|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
AN ANALYSIS OF COMPOUND SPECIFIC CARBON ISOTOPES OF LIPID
BIOMARKERS: A PROXY FOR PALEOENVIRONMENTAL CHANGE IN THE
MAYA LOWLANDS OF PETEN, GUATEMALA
SARAH DAVIDSON NEWELL
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
Sarah Davidson Newell
I find the great thing in this world is not so much where we stand, as in what direction we
are moving...we must sail sometimes with the wind and sometimes against it, but we
must sail, and not drift, nor lie at anchor.
Oliver Wendell Holmes
First and foremost I would like to thank my advisor, Dr. David A. Hodell, for his
constant support, encouragement and inspiration. I appreciate most his faith in my
abilities as a research scientist and unspoken prodding to work harder. Over the past two
years I have developed a stronger passion for research for which he is unduly responsible.
Perhaps the most influential person through my master's research was Dr. Jason H.
Curtis, to whom I now owe my first-born child. Over the past two years, Jason has
provided constant and unparalleled support both in the lab and in life. Jason's assistance
with all aspects of my lab work and method development is the sole reason why I was
able to complete my MS in just two years. Jason provided to me a helping hand in the
lab (at the drop of a hat, I should note), an ear for listening to complaints and excitement
alike, and a shoulder during stressful times. I would especially like to thank him for the
hours spent with our heads buried in the GC oven and for putting up with my spastic
A number of other individuals were instrumental in the completion of my research.
I would like to thank Dr. Mark Brenner for constantly reviewing my work and providing
additional advisement. Mark's own excitement for this work was an integral part of my
completing this project. I would like to thank Dr. Thomas P. Guilderson for allowing me
to travel to the CAMS facility at Lawrence Livermore National Laboratory to run my
own radiocarbon dates. The experience was incredible and my project has greatly
benefited from having the additional radiocarbon dates. I would also like to thank Drs.
Prudence Rice, Katherine Emery and Andrew Zimmerman for the exchange of ideas and
thoughtful discussions. I thank Drs. Barbara Leyden, Gerald Haug and Yongsong Huang
for sharing their data with me and Drs. Mark Pagani and Yongsong Huang for assistance
in the development of my methods. I also thank Jenny Slosek for running all of the bulk
isotope data from Lake Sacnab.
There are three individuals to whom sincere thanks are owed for their personal
support. Dave Buck, Ellie Harrison-Buck and Michael Hillesheim have provided me
with friendship that is unequaled. Each one has supported me in their own individual
way and I can not imagine having done any of this without their love and understanding.
Finally, I would like to thank my parents, Michael and Karen Newell, as well as my
siblings, Erik and Jessica Sanderson and Jeremy and Erika Newell. Without their
unconditional love and support, I never would have made it!
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iv
LIST OF TABLES .......................................... viii
LIST OF FIGURES ........................................ .............. ix
A B S T R A C T ...................................................................................................................... x ii
1 IN T R O D U C T IO N ................................................. .............................................. .
2 BACKGROUND .............................. .. .......... .............................8
M o d em S ettin g ..................................................................................................... 8
Cultural History of the Peten ........................ .................. .................... 10
Interactions among the Ancient Maya, Climate and Environment.......................... 13
S tu d y S ite s .............................................................................................................. . 2 2
L ak e S acn ab ......................................................................................................... 2 2
Lake Salpeten .............. ...... ............ .............. ............... 23
Compound-Specific Carbon Isotopic Studies ................ ...................................23
3 M E T H O D S ................................................................................................................ 2 7
C o re C o llectio n ........................................................................................................... 2 7
L ak e S acn ab ......................................................................................................... 2 7
L ak e S alp eten ...................................................................................................... 2 8
C h ro n o lo g y ................................................................................................................ 2 9
L ak e S acn ab ......................................................................................................... 2 9
L ake Salpeten ............................................................................................. 29
Bulk Elemental Geochemical Analyses.................................................................29
Bulk Carbon and Nitrogen Isotopic Analyses ....................................................... 30
Compound-specific Carbon Isotopic Analyses .............. ....................................30
4 R E S U L T S ................................................................................................................. .. 3 3
C h ro n o lo g y ................................................................................................................ 3 3
L ak e S acn ab ......................................................................................................... 3 3
L a k e S a lp e te n ......................................................................................................3 3
E lem ental G eochem ical A nalyses ......................................................... ................ 38
L ake Sacnab ......................................................................................................... 38
B ulk C arbon and N itrogen Isotopes ...................................................... ................ 41
Com pound-Specific Carbon Isotopes.................................................... ................ 42
L ake Sacnab ................................................................................................... 42
L ake Salpeten .......... ................................. ...................... ... .... ... 44
5 D ISC U SSIO N ............................................................................... ...................... 46
Implications for Changes in Land-use.................................................... 46
Sources of O rganic M atter..................................................................... ............... 50
Evidence for R elative Shifts in V egetation ........................................... ................ 58
C om prison w ith Pollen R records .......................................................... ................ 66
Com prison w ith Population Estim ates................................................. ................ 70
Relationship between Climate and Environmental Changes................................73
C o n c lu sio n s............................................................................................................... .. 7 6
A LIPID EXTRACTION PROCEDURE..................................................................80
B SILICA GEL CHROM ATOGRAPHY ......................................................................82
C UREA ADDUCTION AND GC ANALYSIS ............................. ..................... 84
D D A T A T A B L E S ............................................................ ............................................. 8 6
L IST O F R E FE R E N C E S .................................................. ........................................... 97
BIOGRAPH ICAL SKETCH .................. .............................................................. 104
LIST OF TABLES
1: Perkin Elmer 8500 Gas Chromatograph oven program for sample analyses .............31
2: Hewlett Packard 6890 GC oven program for sample analyses. ...............................32
3: AMS 14C dates for samples from sediment core SN-19-VII-97 from Lake Sacnab.
For the LLNL-CAMS samples, backgrounds were scaled relative to sample size
using Pliocene wood blanks prepared at UF (4+1tg)......................... ................ 34
4: AMS 14C dates for samples from sediment core SP-12-VI-02-1A from Lake
S a lp e te n ............................................................................................................... .. 3 5
5: Tie points used to correlate the %CaCO3 record from core SP2-19-VII-99
(Rosenmeier et al., 2002b) and the scanning XRF Ca concentration data from
core SP-12-VI-02-1A (correlation coefficient = 0.505)......................................36
LIST OF FIGURES
1: Location map showing the approximate location of the Peten Lake District in
northern Guatemala. Map adapted from Map #802723 from the Civil
Intelligence Agency (http://www.cia.gov/cia/publications/pubs.html)...................2...
2: Impact of long-term Maya settlement on the terrestrial and aquatic environments in
the Peten, Guatemala. M odified from Rice (1985). ........................... ................ 15
3: Population density estimates versus time (cal yr BP) and time periods for Lake
Sacnab and Lake Salpeten from the Middle Preclassic (cal yr BP) to the Late
Postclassic (cal yr BP). Data from Rice and Rice, 1990. ........................... 16
4: Location map showing: (A) the Yucatan Peninsula and the location of the Peten
Lake District within Guatemala, (B) detail of the Peten Lake District, (C) the
bathymetry of Lake Sacnab and location of core SN-19-VII-97 and (D) the
bathymetry of Lake Salpeten and location of cores SP-12-VI-02 and SP2-99 .......22
5: Histograms from Ficken et al. (2000) showing the molecular distribution of n-
alkanes from the three categories: terrestrial, emergent, and submerged/floating.
Only odd carbon number distributions are shown and bars represent 1 standard
d e v ia tio n ............................................................................................................... ... 2 5
6: Depth versus calibrated age (yr BP) for terrestrial wood, seed and charcoal samples
in Lake Sacnab core SN-19-VII-97. Squares indicate samples analyzed at
LLNL-CAMS and the triangle indicates the sample measured at NOSAMS ..........35
7: Correlation of SP2-19-VII-99 and SP-12-VII-02-1A using %CaCO3 in the SP2-19-
VII-99 core and the scanning XRF Ca concentration data from the SP-12-VII-
02-1A core .............................................................................. ......................... 37
8: Calibrated ages (yr BP) versus depth for correlated tie points in core SP-12-VII-02-
lA as well as radiocarbon dates of terrestrial samples in Lake Salpeten core SP-
12 -V II-0 2 -1A ........................................................................................................ 3 8
9: Magnetic susceptibility, percent calcium carbonate (%CaCO3), percent organic
matter (% OM), percent other (%Other) and percent nitrogen (%N) versus age in
calibrated years before present (cal yr BP) from Lake Sacnab........................ 40
10: Magnetic susceptibility versus age in calibrated years before present (cal yr BP)
for core SP-12-VII-02-1A from Lake Salpeten. The gray highlighted area
represents the "M aya clay" unit. ..............................................................................4 1
11: Bulk carbon and nitrogen isotopes for core SN-19-VII-97 from Lake Sacnab. The
gray highlighted area represents the "Maya clay" unit. .....................................43
12: Compound-specific 613C results of long chain n-alkanes (C29, C31, and C33) from
Lake Sacnab, Guatemala versus age in calibrated years before present (cal yr
B P ) ....................................................................................................... . ....... .. 4 4
13: Compound-specific 613C results of long chain n-alkanes (C29, C31, and C33) from
Lake Salpeten, Guatemala versus age in calibrated years before present (cal yr
B P ) ....................................................................................................... . ....... .. 4 5
14: Comparison of 613C of C33 from Lake Sacnab with magnetic susceptibility. ............51
15: Comparison of 613C of C33 from Lake Salpeten with magnetic susceptibility............52
16: C:N ratios in weight % and bulk organic matter isotopes from Lake Sacnab.............54
17: Comparison of 613C values for bulk organic matter, C31 and C33 from Lake
S a c n a b ........................................................... ...... ................................................ ... 5 5
18: Predominant n-alkane chain length (A) and CPI (B) for Lakes Sacnab and
S a lp e te n ............................................................................................................... .. 5 7
19: Comparison of 613C records for n-alkane chain C33 in Lake Sacnab and Lake
S a lp e te n .............................................................................................................. .. 5 9
20: Diagram comparing the 613C of C33 to the 615N of bulk organic matter in Lake
S a c n a b ........................................................... ...... ................................................ ... 6 1
21: Relative shifts in contribution of C4 vegetation (in %) in Lakes Sacnab and
Salpeten over the last -4500 cal yr BP................................................ ................ 63
22: Diagram showing the comparison between 613C values of C31 from Lake Sacnab,
Lake Salpeten and Lake Quexil (Huang et al., 2001) .........................................65
23: Diagram showing % grass pollen versus 613C of n-alkane C33 in Lakes Salpeten
and Sacnab and the presence of maize pollen in Salpeten (pollen data from
L eyden, 1987) ........................................................................... ........ ............... 69
24: Comparison of the compound-specific carbon isotope records (C33) and population
density estimates versus time in Lakes Salpeten (top) and Sacnab (bottom)...........71
25: Comparison of the compound-specific carbon isotope records (C33) and percent Ti
versus time in Lakes Salpeten and Sacnab. Percent Ti data from Haug et al.
(2 0 0 1 ) ................................................................... ................................................ ... 7 5
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
AN ANALYSIS OF COMPOUND SPECIFIC CARBON ISOTOPES OF LIPID
BIOMARKERS: A PROXY FOR PALEOENVIRONMENTAL CHANGE IN THE
MAYA LOWLANDS OF PETEN, GUATEMALA
Sarah Davidson Newell
Chair: David A. Hodell
Major Department: Geological Sciences
The Peten region of northern Guatemala has been occupied by humans for more
than 3000 years. During that time, the lowland tropical environment experienced a
prolonged period of anthropogenic disturbance. Forest disturbance in the Peten Lake
District of northern Guatemala was associated with Maya agricultural practices as well as
clearing for urban development, construction, and fuelwood both for cooking and for
lime-plaster production. Expansion of the Maya civilization during the Preclassic (-1000
BC to AD 250) and Classic periods (AD 250 to AD 900) was accompanied by increasing
deforestation of Peten watersheds and accelerated rates of soil erosion. Palynological
data from the Peten Lake District illustrate the near elimination of high forest taxa and
prevalence of disturbance taxa (grasses, weeds) during the height of Classic Maya
occupation (-AD 500-800). After flourishing during the Classic period between AD 250
and 800, the Maya civilization in the Peten Lake District experienced a dramatic change
between AD 800 and 900 that some have referred to as the "collapse." Population
densities declined significantly after AD 900, thereby curtailing human pressures on the
landscape. This cycle of population expansion and decline in the Peten Lake District
provides a natural historical "experiment" that has been used to study the response of
tropical vegetation to long-term changes in land-use by humans. A new line of evidence
is used here to complement other archaeological and paleoenvironmental methods: the
use of leaf wax biomarkers in palynological inference studies. The molecular and isotopic
compositions of leaf waxes have been shown to be reliable indicators of vegetative
biomass and are useful for testing palynological inferences. The carbon isotopic
composition of long-chain n-alkanes of leaf waxes were used as a geochemical proxy for
terrestrial vegetation to test palynological inferences of vegetation change in two lake
basins (Lakes Sacnab and Salpeten) in the Peten Lake District of the southern Maya
Lowlands over the past -4500 cal yr BP.
Discrepancies between the 613C of long-chain n-alkanes and vegetation changes
inferred from pollen profiles suggest that the two proxies may be recording different
characteristics of watershed vegetation as well as different airshed/watershed processes.
The data indicate that in the watersheds of Lakes Salpeten and Sacnab, shifts in the
proportion of C3 to C4 vegetation are most likely controlled by a combination of climate
change and human deforestation. The correspondence of 613C records to independent
proxies for climate change from -4500 cal yr BP until -3000 cal yr BP suggest that
regional drying and increased climate variability caused an increase in the contribution of
C4 vegetation during that time. Following this period and beginning with the first Maya
occupation in the watersheds, vegetation change was likely a result of human-driven
deforestation or perhaps a combination of both climate and human impact.
The Peten region of northern Guatemala (Figure 1) has been occupied by humans
for more than 3000 years. During that time, the lowland tropical environment
experienced a prolonged period of anthropogenic disturbance. Forest disturbance in the
Peten Lake District of northern Guatemala was associated with Maya agricultural
practices as well as clearing for urban development, construction, and fuelwood both for
cooking and for lime-plaster production. Expansion of the Maya civilization during the
Preclassic (-1000 BC to AD 250) and Classic periods (AD 250 to AD 900) was
accompanied by increasing deforestation of Peten watersheds and accelerated rates of soil
erosion (Brenner, 1983). Palynological data from the Peten Lake District illustrate the
near elimination of high forest taxa and prevalence of disturbance taxa (grasses, weeds)
during the height of Classic Maya occupation (-AD 500-800) (Islebe et al., 1996;
Leyden, 1987, Vaughan etal., 1985, Deevey, 1978).
After flourishing during the Classic period between 250 and 800 AD, the Maya
civilization in the Peten Lake District experienced dramatic change between 800 and 900
AD that some have referred to as the "collapse". Population densities declined
significantly after 900 AD, thereby curtailing human pressures on the landscape. This
cycle of population expansion and decline in the Peten Lake District provides a natural
historical "experiment" that can be used to study the response of tropical vegetation to
long-term changes in land-use by humans (Deevey, 1969).
I I L :
,r u r'r; -,
A, h -- r..:
p, I wij --A.
'l ., 1, Guatem- l
n ..rU 1 international boundary
r.., ar, -..Department b,. ndary
Fg r 1:L o i o o th Nationl capital
:u NO t' .- u Department capital
I .n '--1 -1i ; | 1 R a l lroal
northern Guatemala. Map adapted from Map #802723 from the Civil Intelligence
I : ; lu o.1 po ,o ,
ilTl iI r., r "
Fiue1 oainmpsointhaprxmelotinfte P eeten akeDsrict,I in
notenGaemanla.Mpaatdfo apeDis272ctfrom the Civ l ntelligence
Figuen (ht:/Location iapsov/i/pbicgteapprons/pbsatemlocto) fth.ee aeititi
The exact timing and extent of deforestation during the period of Maya
occupation and the reforestation that followed the collapse is debated among
archaeologists and paleoecologists. Evidence currently used in the study of ancient
human impact on the Maya lowland environments comes from paleobotanical,
environmental archaeological and paleolimnological studies. Much of this evidence is
derived from palynological research. However, much of this plant-based evidence is
biased in one way or another and none can be used in isolation. For example, vegetation
"reconstructions" inferred from pollen profiles may be biased because the relative
abundance of pollen grains in a sediment profile cannot be directly related to species
abundance or biomass (Bradley, 1999). Most pollen in lake sediments represents only
the small percentage of tropical vegetation that is pollinated by wind and does not reflect
those plants that depend on pollination by insects or self-fertilization. Because certain
species produce a disproportionately large number of pollen grains, it is difficult to
determine the actual composition of vegetation in the landscape. Furthermore, certain
plants may produce pollen under conditions of stress rather than optimum growth
(Bradley, 1999). Maize pollen has often been used in Mesoamerican studies as a proxy
for agriculture and associated deforestation. One potential concern with traditional
palynology, however, is that maize pollen is not typically included in total pollen counts,
thus causing an over-representation of other taxa. In addition, maize pollen is large and
typically only transported short distances. Alternatively, leaf-waxes derived from maize
are probably transported readily and should reflect maize cultivation in the watershed
accurately. These shortcomings underscore the need to validate palynological
interpretations by independent means.
A new line of evidence is used here to complement other archaeological and
paleoenvironmental methods: the use of leaf wax biomarkers in palynological inference
studies. The molecular and isotopic composition of leaf waxes have been shown to be
reliable indicators of vegetative biomass (Hughen et al., 2004; Huang et al., 1999; Huang
et al., 2001) and are useful for testing palynological inferences (Hughen et al., 2004).
There are several advantages to using leaf wax biomarkers in conjunction with pollen to
assess changes in vegetative biomass. Pollen and leaf waxes are derived from different
vegetative sources and thus record different aspects of watershed vegetation (Huang et
al., 1999). Whereas pollen reflects only reproductive effort for specific groups of plants,
leaf waxes provide a more representative measure of vegetative biomass for various plant
types within a watershed. Leaf waxes reflect the contribution of all land plants whereas
pollen will disproportionately reflect wind-pollinated species. In addition, plants in the
watershed that do not reproduce sexually or are dormant will produce a biomarker signal
but yield no pollen.
The isotopic composition of leaf waxes can be used as a geochemical proxy of
terrestrial vegetation (Hughen et al., 2004; Huang et al., 1999, 2001). Long-chain (C27-
C33) n-alkanes exhibit a strong odd-over-even carbon-numbered dominance and are
produced nearly exclusively by vascular plants as components of epicuticular leaf waxes
(Meyers, 1997). The carbon isotopic composition of terrestrial plant biomarkers in lake
sediments reflects the relative contribution to the sediments of alkanes coming from
plants using the C3- (tropical trees) versus C4- (grasses) metabolic pathway. Plants that
fix carbon by means of the C3 pathway include all the high forest trees (Huang et al.
2001). C4 plants include many tropical grasses and maize, which are associated with
cleared land and agriculture in Mesoamerica. The 613C values of n-alkanes in C3 plants
range from -31%o to -38%o, whereas n-alkanes in leaf waxes of C4 plants typically range
from -19%o to -25%o (Freeman, 2001). Stratigraphic variations in the 613C ratio of long-
chain n-alkanes in lake sediment cores should therefore reflect changes in the proportion
of C3 to C4 vegetation in a lake's watershed (Huang et al. 2001; Meyers, 1997).
Here, the carbon isotopic composition of leaf waxes are used as a geochemical
proxy for terrestrial vegetation type to test palynological inferences of vegetation change
in two lake basins (Lakes Sacnab and Salpeten) in the Peten Lake District of the southern
Maya Lowlands. Three specific questions are addressed in this study:
1) Is there a significant relationship between the 613C of long-chain n-alkanes and
vegetation changes inferred from pollen profiles? If the signals are correlated, then both
long-chain n-alkanes and pollen are most likely recording the same vegetation changes in
a watershed. For example, as disturbance taxa (grasses, weeds) replace high forest taxa,
one would expect the 613C of n-alkanes to increase, reflecting a greater proportion of C4
biomass. Similarly, an increase in the relative proportion of high forest taxa during
reforestation should be accompanied by a decrease in the 613C of n-alkanes. Comparison
of pollen taxa and long-chain, n-alkane 613C records will reveal whether inferred
vegetation changes based on the two proxies are in agreement or contradictory. Because
pollen and carbon isotopes of leaf waxes may reflect different characteristics of
watershed vegetation and different airshed/watershed processes, the two proxies need not
2) Are 613C values of long-chain n-alkanes in lake sediments correlated to Maya
population density estimates within the same watershed? If agricultural land clearance
was tied to population density, then increased disturbance taxa and maize are expected to
increase as populations increased. Because grasses and maize are C4 plants, one would
predict an increase in the 613C of n-alkanes with population growth. As populations
declined and vegetation returned to a C3-dominated forest, a decrease in the compound-
specific carbon isotopic ratios would be recorded. Population density and the 613C of
long-chain n-alkanes may be decoupled if, for example, agricultural practices changed
through time from more extensive to intensive methods, or if alternative crops, other than
maize, became important in the diet. The relation between population density and 613C
of long-chain n-alkanes can be addressed because population data are available from
archaeological transects in the watersheds of both Lakes Sacnab and Salpeten. 3) What
was the relative importance of human- and climate-induced changes in vegetation in the
Central Peten? It is difficult to assess the effect of climate change during the period of
human occupation because it is difficult to tease apart human-induced versus climate-
induced changes in vegetation. However, a comparison of the 613C of n-alkanes with
climate proxies that are not confounded by human impact may reveal whether climate
played a role in vegetation change at all. If vegetation changes in a watershed were
climate-induced, one would predict that long-chain n-alkanes would correlate with
independent, regional paleoclimate records. For example, a period of increased
evaporation/precipitation (E/P) would likely coincide with an increase in the 613C of n-
alkanes, reflecting a greater proportion of C4 biomass. A period of decreased E/P would
likely correspond to a return to a C3-dominated forest. This question can be answered by
comparing proxy records of climate change from Peten and the Caribbean Sea with
compound-specific analyses, and evaluating those data in light of changes in Maya
The Maya area, located in southeastern Mesoamerica, occupies a broad expanse of
land and includes parts of the countries of Mexico, Guatemala, Belize, Honduras and El
Salvador. The area has typically been both culturally and geographically divided by
scholars into the highlands in the south and the lowlands in the north. The highlands,
dominated by a volcanic landscape, refer to the area greater than 1000 ft above sea level
(a.s.1.). and spread from southeastern Chiapas toward lower Central America. The
lowlands consist of the Yucatan Peninsula in Mexico, Guatemala, and Belize.
The topography of the lowlands is dominated by a limestone platform that has
evolved into porous, karst hills with extensive dissolution features. While there are few
permanently flowing rivers and lakes are rare in the northern lowlands, there are
numerous lake basins in the department of Peten. Soils are relatively thin and vulnerable
to rapid erosion with vegetation removal, making agriculture in this region challenging.
Today, the lowland Maya practice shifting, slash-and-burn (swidden) agriculture that
permits the forest to regenerate at intervals. The poor soils, however, can only be planted
for ~2 years; after which, plots are fallowed for 4 to 7 years in the Peten and 15 to 20
years in other parts of the Yucatan (Coe, 1999). These poor soils, in addition to the stress
added by highly seasonal rainfall, make the lowlands a rather harsh place to practice
agriculture (Coe, 1999).
The Peten Lake District is located in the Guatemalan Lowlands of the Yucatan
Peninsula (Figure 1). The region contains numerous terminal basins that are aligned
along a series of east-west aligned en echelon faults (Deevey et al,. 1979). The bedrock
is karst limestone of Cretaceous and Tertiary age with elevations ranging from 100 to 300
m. Most of the lakes have deep troughs at the foot of a steep, north-shore fault scarp
which gives rise to their distinctive bathymetry. The water table in the area is deep below
the ground surface, making groundwater relatively inaccessible. Perched surface waters
often result in seasonally inundated topographic depressions, or bajos, that are clay-
floored and hold water during the rainy season (Deevey et al., 1979). The major water
bodies of the Peten Lake District remain filled throughout the year and extend
approximately 100 km from east to west.
Soils in the Peten region are dominated by well-drained, mineral-rich mollisols that
support a tropical semi-deciduous and evergreen forest (Lundell, 1937). Modern
vegetation is variable throughout the region. The central Peten is mostly semi-deciduous
subtropical moist forest while the southwest Peten is dominated by extensive savannas
with forested hills that support a diverse, fire-resistant herbaceous flora. The savanna
vegetation may have been created during the period of Maya occupation when much of
the forest was burned for cultivation (Leyden, 1987). Alternatively, it may be a
Pleistocene relict or an edaphic assemblage, the consequence of clayey, hydromorphic
soils. The low-lying basins in the northeast Peten are dominated by swamp and
marshland. Vegetation here is dominated by palms, grasses and sedges, but also includes
a semi-deciduous higher forest (Lundell, 1937).
Mean modern annual temperature is approximately 25C and mean annual
precipitation is approximately 1600 mm with a pronounced dry season between January
and May (Deevey et al., 1980). There is very little surface drainage in the area and most
water that falls as rain is either transpired, evaporated, or directly enters the aquifer
(Rosenmeier et al., 2002). The Peten typically receives higher rainfall than the rest of the
Yucatan; there is a pronounced decrease in rainfall from south to north on the Yucatan
Peninsula. There is also considerable interannual precipitation variability. These inter-
annual variations force Maya farmers to practice long-term resource planning. Extensive
droughts occur periodically in this area and previous studies have shown that this
variability is not solely a modern occurrence (Hodell et al., 1995, 2001; Curtis et al.,
Cultural History of the Peten
The ancient Maya civilization of Mesoamerica arose about 2000 B.C. and spanned
a period of 3,000 years before undergoing a period of social and political change in the
Late and Terminal Classic, between 750 and 1050 AD. These changes resulted in the
development of very different Maya political, economic, and ideological systems and
were associated with the cessation of construction of major architecture and elite
monuments, the reduction of inland trade systems designed for the movement of elite
status markers, and the abandonment of some but not all urban centers as populations
dispersed into non-urban village settings, and migrated northward into the northern
Yucatan and Belizean region. These political and social changes are inferred from the
archaeological record of ancient Maya settlement. Maya civilization was unique in
possessing the only written language in the Americas at the time. It also developed the
most sophisticated and accurate calendar of the time, monumental architecture, a
hierarchical social class system, sophisticated agricultural systems including intensive
agriculture, and trade systems extending from northern Mexico to south of Honduras and
El Salvador and perhaps as far as South America. The Maya also had a highly developed
religious system based, during the Classic period, on semi-divine kingship and a noble
class (Sharer, 1994).
The Peten region was continuously occupied by the ancient Maya from the Middle
Preclassic (1000 BC) through the Postclassic (AD 900 to 1525), and up to the Spanish
conquest in 1697 (Rice and Rice, 1990). The Preclassic period (1000 BC to AD 250) in
the Peten region is divided into the Middle Preclassic and Late Preclassic. It is during
this period that the Maya are thought to have developed from hunters and gatherers into a
complex civilization. This period marked the development of cities, temples and
inscribed stone monuments. During the Middle Preclassic, archaeological evidence
suggests the presence of social stratification as well as sophisticated religious and
economic institutions. The first examples of writing appear in the Late Preclassic (Coe,
The Classic Period was characterized by remarkable growth of the civilization both
in terms of population and social complexity. The Classic Period (AD 300- AD 900) is
also subdivided into three periods: the Early Classic, Late Classic and Terminal Classic.
During the Early and Late Classic periods, archaeological evidence suggests the
development of "states" with centralized political and religious authority in addition to
the erection of carved stone monuments (Coe, 1999). An important change in the
political structure occurred at the end of this period, as authority became shared by a
council of many instead of one individual (Coe, 1999). Populations grew slowly, but
exponentially and in many regions displayed peak population densities between AD 700
and 800 AD (Rice and Rice, 1990).
The Terminal Classic and Post-Classic Periods (AD 900 to AD 1525) are marked
by diverse cultural responses throughout Mesoamerica. Much of the southern lowland
region experienced the abandonment of Classic cities, temples and religious centers;
subsequent political fragmentation resulted in a massive cultural decline that has been
referred to as the "collapse" (Coe, 1999). The Maya Terminal Classic, however, is
perhaps one of the most important periods in the Peten region. The uniqueness of central
Peten lies in the ability to study not only the events leading up to the Maya Terminal
Classic period, but the continuity of occupation following the collapse of Classic
civilization (Rice and Rice, 2004). This period in the central Peten was both a center and
a crossroads for Postclassic Maya civilization. Peten-like architecture and iconographic
traits in the northern lowlands are evidence of Late Classic, conflict-driven migration
from the Peten region. Throughout the Late Classic and Terminal Classic, the Peten
region showed demographic loss, whereas the northern lowlands began to be heavily
settled. The lakes region in particular suffered population decline, though the area was
never completely abandoned. Archaeological evidence from every major Peten lake
basin shows continuous occupation from the Late Classic to the Postclassic, with some
indication of population migration between the Rio de Pasi6n region, the Gulf Coast, and
the Peten Lake District during the Postclassic (Rice and Rice, 2004).
The nature of settlements changed from the Late Classic to the Postclassic in the
Peten Lakes District. Whereas Classic settlements expanded throughout the watershed,
most settlement shifted to small, densely populated areas found primarily on islands and
peninsulas in the lakes during the Postclassic. This settlement pattern in itself suggests a
"conflict-driven" society. The location of these sites in poor agricultural zones, the
presence of defensive structures, as well as the presence of intrusive residential
architecture in previously occupied sites, suggests that the Late Classic migration of
small groups between the Rio de Pasi6n and the northern lowlands forced them to settle
into an already-established settlement system (Rice and Rice, 2004).
Archaeological evidence overwhelmingly suggests that the Maya population in the
Peten Lakes District did not "collapse", but underwent a significant transformation.
While there was a pronounced demographic decline, the "collapse" in the Peten did not
result in complete depopulation. Instead, intersite conflict led to political restructuring
and thus a redistribution of population, especially into fragile lands. The region was one
of dynamic and contested lands, and remained so until AD -1200 (Rice and Rice, 2004).
Interactions among the Ancient Maya, Climate and Environment
The Maya Lowlands have been studied by researchers seeking answers to
questions about past human-climate-environment interactions. This research has
concentrated on the central Peten, and in particular the Peten Lakes District. In 1972,
Edward S. Deevey began a long-term paleoecological project in the Department of Peten,
Guatemala. The Central Peten Historical Ecology Project (CPHEP) investigated both the
paleolimnology and archaeology of major lake watersheds in the Central Peten (Rice et
al., 1985). The objective of the project was to investigate both the social and natural
history of the region through interdisciplinary research, focusing on ecological change as
a result of human landscape transformations. One assumption of the CPHEP was that
climate change was unimportant during the period of Maya occupation. Recent studies
have recognized, however, that climate was not invariant through this period (Hodell et
al., 1995; Rosenmeier et al., 2002; Curtis et al., 1998). Research design included
paleolimnological studies from other regions that incorporate archaeological perspectives
(Dunning et al., 1997, 1998; Demarest, 1997). In particular, lake sediment cores were
used to develop long-term, high resolution records of environmental change within a
watershed and were used in conjunction with archaeological surveys that estimated the
timing and density of human occupation.
The impacts of long-term Maya occupation in the Central Peten Lake District on
both terrestrial and lacustrine environments were summarized by researchers in the
CPHEP (Binford et al., 1987) (Figure 2). The model is generalized and does not
necessarily describe the history of human and environmental changes for each and every
Archaeological surveys consisted of systematic mapping of settlement remains
and test-pit excavations, and provide estimates of settlement patterns and population
growth in the central Peten (Rice and Rice, 1990). Surveys were completed in each of
three twin-lake basins: Sacnab-Yaxha, Macanche-Salpeten, and Quexil-Petenxil. The
trends of population growth are notably similar in each basin surveyed, although the true
variation was probably not captured by the coarse chronological framework, which was
based on ceramic phases, not radiocarbon. Occupation of the region began at
approximately 1000 BC and was followed by a steady increase until the end of the Late
Classic (AD 750-800), when populations reached their peak in most basins (Figure 3).
Population growth was exponential in the Yaxha-Sacnab basins but not in the other
basins, which experienced a small population decline during the Early Classic.
10,000 4000 3000 2000 1000 0
Cal yr BP
Figure 2: Impact of long-term Maya settlement on the terrestrial and aquatic
environments in the Peten, Guatemala. Modified from Rice et al. (1985).
Sediment cores from numerous lakes in Peten show similar stratigraphic changes
in sediment composition and geochemistry (Figure 2) (Cowgill and Hutchinson, 1966;
Deevey et al., 1979; Brenner, 1983; Vaughan et al., 1985; Rice et al., 1985; Binford et
al., 1987; Leyden, 1987; Curtis et al., 1998; Islebe et al., 1996). Holocene sediment prior
to Maya occupation is comprised of organic-rich (30-60%) gyttja (Brenner et al., 2002).
Overlying the pre-Maya gyttja is a clay-rich horizon known as the "Maya clay." The
pal yrelateda freo ring an d ecet al.1al 16rti
2500 2000 1SO 1000 Soo 0
cal is. BP
Figure 3: Population density estimates versus time (cal yr BP) and time periods for Lake
Sacnab and Lake Salpeten from the Middle Preclassic (cal yr BP) to the Late Postclassic
(cal yr BP). Data from Rice and Rice, 1990.
base of the Maya clay has been dated to approximately 3,000 yr BP (Brenner, 1994;
Rosenmeier et al., 2002) and reflects accelerated erosion associated with early land
clearance by humans (Deevey et al., 1979; Brenner, 1983; Vaughan et al., 1985; Rice et
al., 1985; Binford etal., 1987; Leyden, 1987; Curtis etal., 1998; Islebe etal., 1996).
Studies outside the region per se suggest the increase in colluviation may have been
partly related to regional drying (Hodell et al., 1995; Hodell et al., 1996; Curtis et al.,
2001). Nutrient-rich soils became unstable as forest cover was removed and replaced by
a savanna-like landscape. Increased soil erosion is reflected by high net accumulation
rates of lake sediments. As erosion progressed, organic-rich surficial soils were removed
and the underlying bedrock was weathered and transported to the lake basin. This is
reflected by the high proportion of inorganic sediments. Overlying the Maya clay unit is
another organic-rich layer that is inferred to represent sediments deposited following the
Maya collapse (Brenner et al., 2002).
Deforestation and environmental disturbance accompanied growth of the Maya
population from the Preclassic to the Late Classic (Figure 2). Vaughan et al. (1985)
developed pollen stratigraphies for Lakes Quexil and Sacnab that indicate changes in
vegetation through the period of Maya occupation. The pre-Maya pollen assemblage was
dominated by high forest taxa such as Moraceae, whereas the early to mid-Preclassic is
characterized by an open forest and, probably, culturally-induced savanna. Late
Preclassic to Late Classic sediments are dominated by clay-rich sediments and 70% non-
arboreal pollen. Vaughan et al. (1985) defined three distinct disturbance pollen zones.
The Late Preclassic zone is characterized by savanna trees and shrubs. The landscape
transitions to nearly grassland during the Early Classic, with only one type of arboreal
pollen present (Ramon). The Late Classic to Postclassic contains only grassland taxa.
Some increases in high-forest taxa near the top of this interval show evidence of
reforestation, as does the pollen spectrum during the Postclassic period. If we assume an
anthropogenic cause for the replacement of arboreal with grassland taxa, a sharp rise in
arboreal pollen and a decrease in grassland types imply regional depopulation (Vaughan
et al., 1985). Although each study is thought to reflect primarily local vegetation
changes, multiple subsequent studies from the Peten and surrounding lowlands, with the
exception of the work done by Cowgill et al. (1966) and Dunning et al. (1997), have all
detected a similar decline in high-forest taxa from the Late Preclassic through the
Terminal Classic that is interpreted to reflect human-driven deforestation (Islebe et al.,
1996; Leyden, 1987, Vaughan et al., 1985, Deevey, 1978). The pollen zones were
assigned ages by "cultural zonation", or the correlation with ages of archaeological
periods. Subsequently, they may contain some chronological error. However, these
same pollen zones have now been independently dated in other lake cores (Leyden, 1987
and Rosenmeier et al., 2002) and are considered reasonable.
Leyden (1987) used a 15 m sediment core from Lake Salpeten, Guatemala to
develop a high-resolution pollen record from the basin to reconstruct Holocene
vegetation changes. The sediment record spans from the pre-Maya to the present, and
shows distinct evidence of Maya land clearance. While there is an apparent lack of
disturbance taxa in the early Preclassic, the pollen zones representing the Late Preclassic
through the Postclassic show evidence for abundant terrestrial herbs and strongly suggest
Maya land clearance. Leyden attributes the lack of disturbance taxa in the early
Preclassic to small local populations in the Salpeten basin at that time. Some
discrepancies, however, may lie within the issues of chronological control. Leyden
suggests that a few high-forest taxa actually increased initially, and declined later as
forest removal was intensified. Presence of oak during the Classic period may suggest a
savanna landscape, but this is relatively unclear. Leyden suggests that after the initial
deforestation, the forest structure was relatively stable. However, greater proportions of
maize pollen during the Late Classic through Postclassic indicate intensified agricultural
activities. In the gyttja layer above the Maya clay, the concentration of total pollen grains
nearly doubles, suggesting forest regrowth after depopulation of the lake basin (or slowed
bulk sedimentation rate). Leyden proposes that these forests regenerated rapidly, but
were more open than pre-Maya forests. This may have been related to climate rather than
anthropogenic influences; regional drying may be suggested by the Post-Maya
continuation of open forests. This period is dominated by secondary growth, as
suggested by the increases in successional shrubs and trees.
While most vegetation studies have focused on forest clearance, a select few have
concentrated on revealing the dynamics of afforestation following the Maya collapse in
the central Peten region of Guatemala (Wiseman, 1985; Brenner et al., 1990). Wiseman
(1985) used pollen in lake sediments to track vegetation changes and the nature of these
changes in response to cultural collapse in the Maya Lowlands. Seven cores from Lakes
Petenxil and Quexil were collected and sampled for pollen analyses and used in
conjunction with modern pollen and fauna studies that examined 0.1 ha plots in various
stages of regrowth. This modern analog guided inferences about forest succession based
on the fossil pollen record. All cores that penetrated into Late Classic Period deposits
showed a subsequent replacement of agricultural weeds by secondary forest growth. The
pollen spectrum during the early Classic to Postclassic was similar to samples taken from
soils under swidden agriculture. One notable feature of the sediment cores is that each
core shows essentially the same pollen spectra, suggesting that the conditions in the basin
were in spatial equilibrium. Wiseman cites the decrease in maize pollen and subsequent
recolonization by forest as key evidence for depopulation; however, poor chronological
control makes it difficult to determine the precise timing of both environmental and
Interpreting vegetation change from palynological data is further complicated by
the challenge of obtaining accurate chronological control. Prior to the advent of AMS
dating and the potential to measure 14C in very small (-20.g) samples of organic carbon,
many early radiocarbon ages from Peten lakes were based on dating of bulk sediment or
carbonate shells. Shells may incorporate carbon derived from the dissolution of ancient
limestone and thus appear older than their true age. This problem extends to organic
matter that is fixed photosynthetically within the lake, i.e. autochthonous organic matter.
Primary producers may incorporate carbon derived from dissolved ancient limestone, and
thus display "too-old" ages. This hard-water lake error has compromised the reliability
of early chronologies. To further complicate matters, many subsequent studies assigned
ages to the pollen stratigraphy by correlation with ages of archaeological periods. Dating
error within one pollen record was thus transferred to records from other lake cores.
Chronological imprecision in both sediment cores and archaeological studies creates
challenges for linking environmental disturbance and human population sizes (Yaeger
and Hodell, in press).
Despite the overwhelming evidence for late Holocene vegetation change in the
Peten region, it is difficult to distinguish between the impacts of climate and Maya
occupation on forest composition. While the CPHEP assumed that climate was relatively
constant for the past 10,000 years, recent work has proved otherwise. A sediment core
from Lake Chichancanab in northern Yucatan provides evidence for regional drying that
occurred beginning approximately 1000 BC with a distinct interval of droughts between
AD 800 and AD 1000 that coincided with the Terminal Classic collapse (Hodell et al.,
1995, Hodell et al. in press). Another paleoclimate record from Punta Laguna, on the
northeastern Yucatan Peninsula, indicates alternating wet-dry shifts in the hydrologic
balance through the late Holocene with the driest period lasting from approximately AD
280 to 1080 (Curtis et al., 1998). Four severe drought events were revealed, including
one that occurred during the Maya Hiatus (AD 600) and one that occurred concurrently
with the Late Classic drought recorded in the Chichancanab record.
A climate record from Lake Salpeten indicates a trend similar to the
Chichancanab record (Rosenmeier et al., 2002). Lake levels decreased continuously
from approximately 1000 BC, with the lowest lake level occurring between AD 800 and
900. This inferred drought coincides with the Terminal Classic collapse of the Classic
Maya. The record also showed reduced soil erosion as well as forest recovery after AD
850, likely associated with Maya population decline (Rosenmeier et al., 2002).
Rosenmeier argues, however, that these changes may not be entirely climate-related, but
rather due to the effects of human-induced vegetation change on the lake's hydrology.
The impact of humans on regional vegetation and soil stability, as well as the
influence of climate change on the Classic Maya, illustrate the complex interplay among
climate, humans, and the environment throughout the Late Holocene in the Maya
Lowlands. While it is extremely difficult to separate the signals of human versus
climate-controlled changes in the environment, further investigations of vegetation
change with precise chronological control may provide additional evidence that the Maya
had a profound impact on their lowland tropical environment. The "natural experiment"
provides an excellent opportunity to study the relationship between changes in human
population sizes, land-use activities, climate changes, and vegetation responses.
Additionally, such studies can help us understand how ecosystems respond once human
and climate pressures are curtailed.
Paleolimnological cores from two lakes in the Central Peten Lake District, one in
the east and one in central Peten, were examined for compound-specific carbon isotopes
(Figure 4). Comparison of the records from Lake Sacnab and Lake Salpeten permits a
regional assessment of vegetation change.
B I Lake Yaxha-
Lake Perdida deran,
Lake Peten Itza Lake SapeIn
~u-e'ra .yr Lake Macanche t
Lake Sacpuy Lake Queil
I I ake Pethnxil N
Figure 4: Location map showing: (A) the Yucatan Peninsula and the location of the Peten
Lake District within Guatemala, (B) detail of the Peten Lake District, (C) the
bathymetry of Lake Sacnab and location of core SN-19-VII-97 and (D) the
bathymetry of Lake Salpeten and location of cores SP-12-VI-02 and SP2-99.
Lake Sacnab (1703'N and 8923'W) is located in the eastern part of the Peten
Lake District near the border between Guatemala and Belize (Figure 4c). The lake has a
surface area of 3.9 km2 and is -3.5 km long by 1.5 km wide (Deevey et al., 1980). The
maximum depth is 13 m. The lake is thermally stratified, but occasional mixing is
~x I~w~ -
CCVIxi tf 4n I t
indicated by the absence of permanent stagnation. Lake waters are relatively poor in
nutrients, with the majority of nutrients being supplied by erosion of upland soils
(Deevey et al., 1980). Lake Sacnab has no outflow, thus making the sediments the
primary sink for dissolved and particulate matter that enters the lake.
Lake Salpeten (1658'N and 8940'W) is located -35 km to the WSW of Lake
Sacnab and has a surface area of approximately 2.6 km2 (Rosenmeier et al., 2002) (Figure
4d). The lake has a maximum depth of approximately 32 m (Brezonik and Fox, 1974).
Lake waters are sulfate-rich and have high total dissolved solids (4.76 g L-1) (Deevey et al.
1980). Surface temperatures range from 27C to 30C throughout the year (Rosenmeier et
al., 2002). Lake Salpeten is closed hydrologically and lake-bottom sediments are the
primary sink for dissolved and particulate matter that enters the lake.
Compound-Specific Carbon Isotopic Studies
The major source of organic matter in lake sediments is generally derived from
phytoplankton living in the water column or aquatic macrophytes. Land plants may also
provide an important source of organic matter to lake sediments. The relative
contribution of these three sources is influenced by productivity of lacustrine algae,
aquatic macrophytes, and terrestrial plants. Transport processes and preservation may
also influence the ultimate contribution of organic compounds from various sources
(Meyers, 1997). In order to determine relative changes in algal, macrophyte, or
terrestrial plant productivity, it is necessary to discriminate between the sources of
organic matter sequestered in lake-bottom sediments. Certain compounds in lake
sediments, commonly called biomarkers, are uniquely derived from specific sources of
organic matter. Leaf waxes are produced exclusively by vascular plants and serve as a
protective coating on leaves and stems (Eglinton and Hamilton, 1967). The abundance
and molecular and isotopic composition of leaf waxes reflect vegetative biomass and
have been used to discern relative changes in terrestrial vegetation (Figure 5) (Hughen et
al., 2004; Filley et al., 2001; Huang et al., 1999, 2001). The wax particles, introduced
into the atmosphere by wind and dust ablation off live vegetation (Simoneit, 1977), have
a molecular composition that is generally similar to that of their source vegetation (Conte
and Weber, 2002). Leaf waxes can also be remobilized from soils during exposure
(Schefub, 2003). Leaf waxes settle onto lake surfaces from the atmosphere and are
incorporated into lake-bottom sediments. Leaf waxes thus have atmospheric residence
times on the order of days to weeks, making them especially valuable as tracers of abrupt
vegetative change. Leaf waxes record biomass of exposed leaf surface area in the
watershed and thus have the capability of resolving rapid responses to climate or other
The isotopic analysis of long-chain n-alkanes has proven to be a useful new tool
for evaluating qualitative changes in terrestrial vegetation (Hughen et al., 2004; Filley et
al., 2001; Huang et al., 1999, 2001). Long-chain (C29-C33) n-alkanes exhibiting a strong
odd-over-even carbon-numbered dominance are produced nearly exclusively by vascular
plants as components of epicuticular leaf waxes (Meyers, 1997). In addition, n-alkanes
are chemically and biologically resistant and are often found in sediments in quantities
sufficient for analysis. Preservation of lipids, and in particular n-alkanes, is generally
good. Studies have shown greater preservation of n-alkane terrestrial biomarkers versus
algal biomarkers (Meyers et al., 1984; Meyers and Ishiwatari., 1993; Meyers and Eadie,
1993) suggesting that studies utilizing these biomarkers are more robust.
TERRESTRIAL 3B w.
submerged/floating. Only odd carbon number distributions are shown and
bars represent 1 standard deviation.
The carbon isotopic composition of long chain n-alkanes (>C27) reflects the
relative contribution of C3 and C4 plants (Hughen et al., 2004; Filley et al., 2001; Huang
et al., 1999; Huang et al., 2001). Plants that fix carbon by means of the C4 metabolic
pathway include the tropical grasses (including maize) and are associated with land
clearance. C4 plants are more competitive than plants using the C3 pathway under both
high water stress (such as during drought) as well as lower ambient pCO2 levels. Plants
that fix carbon by means of the C3 pathway include all the high forest trees (Huang et al.,
2001). Plants that use the C3 pathway have bulk 613C values in the range of -21%o to -
28%o, whereas C4 plants display bulk 613C values between -11%o and -15%o (Lajtha and
Marshall, 1994). Stratigraphic variations in the 6 13C ratio of long-chain n-alkanes in lake
sediment cores will thus reflect the changes in the proportion of C3 to C4 vegetation in a
lake's watershed (Huang et al., 2001).
Previous studies have shown that it is possible to recognize C4 plant expansions by
examining the carbon isotopic composition of long-chain n-alkanes in sediment cores.
Huang, et al. (2001) examined leaf wax n-alkanes from two sites in Mesoamerica and
found contrasting moisture variations over the last 25,000 years. Data indicate that
regional climate plays a large role in the relative abundance of C3 versus C4 plants.
Enriched 613C values during the Last Glacial Maximum (LGM) suggest an expansion of
C4 plants in the sediments of Lake Quexil during the LGM as a result of low partial
pressure of atmospheric carbon dioxide (pCO2) and increased aridity. Results indicate
that it is possible to recognize vegetation changes by examining the compound-specific
carbon isotopes of preserved organic matter in lake sediments. This study provides
information that complements past studies that analyzed only bulk sediment or pollen and
thus provide a better record of changes in terrestrial vegetation in the Peten Lake District.
On 19 July 1997, two sediment cores were retrieved from Lake Sacnab in 8.0 m
of water. A 152-cm-long core (SN-19-VII-97-MWI) was taken using a piston corer
designed to recover undisturbed sediment-water interface profiles (Fisher et al., 1992).
The sediment core was sampled in the field at 1-cm intervals by vertical extrusion into a
sampling tray fitted to the top of the core barrel. A total of 3.7 m of sediment was
collected in four additional sections (SN-19-VII-97-LEX) using a piston corer with
Cores were extruded and sectioned at 1-cm intervals. Magnetic susceptibility was
measured for each section using a GEOTEK multi-sensor core logger (MSCL) at Florida
State University, Tallahassee. All samples from core SN-19-VII-97 were freeze-dried
and ground in preparation for compound-specific carbon isotopic analyses. Magnetic
susceptibility data were used to determine that there was a 25.5-cm offset between the
mud-water interface core and the piston core. All depths in the deeper core sections were
adjusted using the mud-water interface as the 0-cm datum.
On 12 June 2002, two sediment cores were retrieved from 23 meters water depth
in Lake Salpeten. A short (93 cm) trigger/gravity core (SP-12-VI-02-1A-MWI) and a
longer (550 cm) Kullenberg piston core (SP-12-VI-02-1A) were both collected in
polycarbonate liners. The Kullenberg core was cut in the field into approximately 1-m
lengths for transport to the University of Florida, where cores were stored in a cooler.
Whole core sections were analyzed for magnetic susceptibility, gamma ray
attenuation (GRA) bulk density and p-wave velocity using a GEOTEK MSCL at the
University of Florida. Each section of the Kullenberg core was split lengthwise into
archive and sampling halves and described. The sampling half was U-channeled and
taken to the Ocean Drilling Program Core Repository in Bremen, Germany for elemental
analysis on the X-Ray Fluorescence (XRF) Core Scanner. Additional samples were
taken from the remainder of the sampling half at the University of Florida at 10-20 cm
intervals for compound-specific carbon isotopic analyses. The mud-water interface core
was sampled at 1 cm intervals at the University of Florida by vertical extrusion into a
sampling tray fitted to the top of the core barrel. All samples from core SP-12-VI-02
were freeze-dried and ground in preparation for compound-specific carbon isotopic
analyses. Magnetic susceptibility and density data were used to determine the 7-cm offset
between the mud-water interface core and the Kullenberg core and depths assigned to
deeper sections were adjusted accordingly.
Radiocarbon ages for Lake Sacnab sediments were determined by accelerator
mass spectrometry (AMS) using terrestrial organic matter (seeds, charcoal and wood) at
Lawrence Livermore National Laboratory Center for AMS (LLNL-CAMS) and the
National Ocean Science AMS (NOSAMS) facility at Woods Hole Oceanographic
Institution. Radiocarbon ages were converted to calendar ages using the program OxCal
v 3.9. (Bronk Ramsey, 1995; Bronk Ramsey, 2001) and atmospheric data from Stuiver et
The sample that was analyzed at NOSAMS was pretreated on-site whereas the
samples analyzed at LLNL-CAMS were pretreated at the University of Florida. For the
LLNL-CAMS samples, backgrounds were scaled relative to sample size using UF
processed Pliocene wood blanks to determine the modern-C contribution (41 pg). All
radiocarbon ages are adjusted to a 613C value of-25 per mil.
Radiocarbon ages for Lake Salpeten sediments were determined by AMS 14C
dating of terrestrial organic matter (seeds, charcoal and wood) at LLNL-CAMS.
Radiocarbon ages were converted to calendar ages using the program OxCal v 3.9 (Bronk
Ramsey, 1995; Bronk Ramsey, 2001) and atmospheric data from Stuiver et al. (1998).
Bulk Elemental Geochemical Analyses
Total carbon (TC) and total nitrogen (TN) were measured on all samples from
core SN-19-VII-97 using a Carlo Erba NA 1500 CNS elemental analyzer with
autosampler. Analytical precision for TC and TN is approximately 0.5%. Total
inorganic carbon (TIC) in the sediments was measured by coulometric titration
(Engleman et al., 1985) with a UIC (Coulometrics) Model 5011 CO2 coulometer coupled
with a UIC CM5240-TIC inorganic carbon preparation device. Analytical precision is
approximately 0.5% based on analysis of 16 calcium carbonate internal standards.
Organic carbon (OC) was calculated by subtracting TIC from TC. Weight percent
calcium carbonate (%CaCO3) was calculated by multiplying IC by 8.33. Weight percent
organic matter (%OM) was estimated by multiplying OC by 2.5.
Bulk Carbon and Nitrogen Isotopic Analyses
Bulk organic sediment samples were analyzed on-line for carbon and nitrogen
isotopes using a VG PRISM Series II isotope ratio mass spectrometer with a triple
trap preparation device linked to a Carlo Erba NA 1500 CNS Elemental. Bulk carbon
isotopic results are reported in standard delta (6) notation relative to the Vienna PeeDee
Belemnite (VPDB) standard. Precision for 613C samples was approximately 0.15%o
based on nine analyses of NBS-22. Bulk nitrogen isotopic results are reported in
standard delta (6) notation relative to atmospheric N2. Precision for 615N samples was
approximately 0.20%o based on nine analyses of peptone.
Compound-Specific Carbon Isotopic Analyses
The extraction and isolation methods used for lipids in this study were patterned
after Silliman et al. (2000) and M. Pagani (personal communication, 2003). Prior to
extraction, approximately 15 .g of C34 n-alkane was added to each sample as an internal
standard for n-alkanes. Lipids were extracted from approximately 3-5 g of dry sediment
using 2:1 methylene chloride/methanol in a Dionex Accelerated Solvent Extractor (ASE)
(for program see Appendix A). Extraction efficiencies averaged approximately 89% for
all samples. The samples were then evaporated and solvent exchanged to hexane using a
hot water bath while adding a stream of dinitrogen gas (N2). Samples were dissolved in 1
mL of hexane and added to a 1 cm x 29 cm glass column filled with 2.5 g of 5%
deactivated silica gel. 15 mL of hexane was used to elute the n-alkanes. Full isolation
scheme procedures are outlined in Appendix B. Samples were urea adducted to obtain
clean n-alkanes for gas chromatography as outlined in Appendix C. After urea
adduction, samples were concentrated in 200 [.L of hexane in preparation for analysis on
a Perkin Elmer 8500 Gas Chromatograph (GC) to determine purity and appropriate
concentrations for GC-IRMS analyses. Samples of 4 U.L were injected into the GC with a
30 m DB-1 column (0.25 mm ID). The gas chromatograph was used in split injection
mode with a ratio of 20:1 and equipped with a FID detector. The GC oven temperature
was programmed to maximize alkane separation (Table 1). The necessary dilution for
GC-IRMS analysis was calculated and samples were transferred to a glass auto-sampler
vial and sealed with a Teflon crimp cap. Samples were typically dissolved in 20-100 UL
hexane. Carbon isotopic analyses were performed using a Hewlett Packard 6890 GC
connected to a Finnigan MAT Delta+ XL Mass Spectrometer via a GC-C III interface.
Table 1: Perkin Elmer 8500 Gas Chromatograph oven program for sample analyses.
Rate Temperature Time
C/min 0C (min)
6 300 20
The n-alkanes were separated on a fused silica capillary column (30 m x .32 mm
i.d.; .25uim film thickness) using helium as the carrier gas. The GC oven temperature
was programmed to maximize alkane separation (Table 2). Alkanes were combusted into
CO2 in a ceramic oxidation reactor containing three braided NiO/CuO/Pt wires. Three
pulses of a standard, calibrated CO2 reference gas were injected via the GC-C II interface
to the IRMS for the measurement of 613C values of individual alkanes. A laboratory
working standard (UFIS) consisting of three n-alkane chains (C19, C25 and C30) was
measured along with unknowns at the beginning and end of each run, as well as after
every fourth sample analysis within a run. UFIS was calibrated to a set of standard n-
alkanes (Mix A) from Indiana University with known 613C values. Long-term analytical
precision, based on repeated analysis of n-alkanes in UFIS, was 0.4%o. Data were
acquired and processed using ISODAT NT 2.0 software. All reported carbon isotopic
compositions for samples represent averaged values for duplicate analyses. Duplicate
analyses had a standard deviation of approximately 0.5%o. Long-term analytical
precision based on analysis of the internal standard C34 was 0.4%o.
Table 2: Hewlett Packard 6890 GC oven program for sample analyses.
Rate Temperature Time
C/min 0C (min)
6 299 0
10 300 4
Twelve AMS 14C dates were obtained from core SN-19-VII-97 and yielded a
maximum age of 4580 14C yr BP on a sample at 439.5 cm (Table 3). The core
chronology was established by converting sediment depths to age using three linear
regression equations (Figure 6):
196.5-0 cm: age (cal yr BP) = 11.863 x depth, r2 = 0.9628;
369.5-196.5 cm: age (cal yr BP) = 4.6821 x depth + 1495, r2 = 1;
439.5-369.5 cm: age (cal yr BP) = 17.683 x depth 3293.2, r2 = 0.9317.
The age of the base of the core (445 cm) is estimated to be -4650 cal yr BP based on
extrapolation of the linear regression from 369.5 to 439.5. Sedimentation rates for the
three intervals are:
196.5-0 cm = 0.081 cm/yr;
369.5-196.5 cm = 0.214 cm/yr;
439.5-369.5 cm = 0.052 cm/yr.
The wood date at 363.5 cm had a large error because of its small size and was not used to
establish the chronology.
Two AMS 14C dates were obtained from core SP-12-VI-02-1A and yielded a
maximum age of 3985 14C yr BP (Table 4). Additional chronological control was
obtained by correlating the %CaCO3 record from core SP2-19-VII-99 (Rosenmeier et al.,
2002b) and the scanning XRF Ca concentration data from core SP-12-VI-02-1A
(correlation coefficient = 0.505) using AnalySeries v. 1.0 (Paillard et al., 1996). Ten tie
points were used to correlate the records (Figure 7, Table 5).
Table 3: AMS 14C dates for samples from sediment core SN-19-VII-97 from Lake
Sacnab. For the LLNL-CAMS samples, backgrounds were scaled relative to
sample size using Pliocene wood blanks prepared at UF (4+1 pg). All
radiocarbon ages are adjusted to a 613C value of -25 %o. Radiocarbon ages
were converted to calendar ages using the program OxCal v 3.9 (Bronk
Ramsey, 1995; Bronk Ramsey, 2001) and the atmospheric data set of Stuiver
et al. (1998). All ages reported in this thesis are in calendar years before
present (relative to AD 1950).
Composi Calibrated Age 95.4%
te Depth Accession Sample Age 1 a Age (cal Probability
Sample ID (cm) Number Material (14C yr BP) yr BP) (AD/BC)
SN-19-VII-97-MWI_45.5 45 CAMS 58754 280 60 390 115
SN-19-VII-97-LEX1 109.5 84 CAMS 58755 1210 50 1165 AD 800 110
SN19VII97MWI 97-98 97.5 CAMS 106265 seed 990 50 880 115
SN19VII97MWI 121-123 122 CAMS 106266 material 1650 60 1560 AD 400 155
SN-19-VII-97-LEX1 152.5 127 CAMS 58756 1620 50 1515 AD 400 120
SN19VII97MWI 145-147 146 CAMS 106267 material 1740 100 1650 AD 300 235
SN19VII97LEX2 221-223 196.5 CAMS 106268 charcoal 2370 40 2415 500 BC 90
SN-19-VII-97-LEX4_414.5 363.5 OS 18657 wood 2500 160 2555 600 BC 400
leaf 1300 BC
SN19VII97LEX4 394-396 369.5 CAMS 106269 material 3030 60 3225 160
leaf 1600 BC
SN19VII97LEX4 404-407 379.5 CAMS 106270 material 3300 70 3540 155
leaf 1800 BC
SN19VII97LEX4 434-436 409.5 CAMS 106271 material 3450 40 3740 105
leaf 2600 BC
SN19VII97LEX4 464-466 439.5 CAMS 106272 material 4050 80 4580 275
4250 y = 17.683x 3293.2
4000- R2 = 0.9317
r 3250 -y = 4.6821x + 1495
S 3000 R2
? 2750 -
) 2500 -
1750 y =11.863x
o 1500- R2 0.9628
0 50 100 150 200 250 300 350 400 450 500
Figure 6: Depth versus calibrated age (yr BP) for terrestrial wood, seed and charcoal
samples in Lake Sacnab core SN-19-VII-97. Squares indicate samples
analyzed at LLNL-CAMS and the triangle indicates the sample measured at
Table 4: AMS 14C dates for samples from sediment core SP-12-VI-02-1A from Lake
Composite Age 1 a Calibrated Age 95.4%
Sample Depth Accession (14C yr Age (cal Probability
Sample ID Material (cm) Number BP) yr BP) (AD/BC)
SP12VI02 CAMS 1790 AD
STA_110-111 charcoal 110.5 106273 210 60 165 170
ST1A 110-111 106273 170
SP12VI02 CAMS 2500 BC
STA_442-445 charcoal 443.5 106274 3985 50 4455 200
ST1A 442-445 106274 200
Table 5: Tie points used to correlate the %CaCO3 record from core SP2-19-VII-99
(Rosenmeier et al., 2002b) and the scanning XRF Ca concentration data from
core SP-12-VI-02-1A (correlation coefficient = 0.505) using AnalySeries v.
1.0 (Paillard et al., 1996). Included are the respective ages for each set of
correlated depths. Ages for depths in core SP2-19-VII-99 were determined
using the chronology outlined in Rosenmeier et al. (2002) while ages for
depths in SP-12-VI-02-1A were determined using the correlation (Figure 4).
Depth in SP- SP2-19-VII-
12-VI-02-1A 99 Age (cal yr BP)
152 151 1660
201.5 201 2030
215 213 2090
236.5 236 2210
246 248 2270
265.5 266 2390
293.5 289 2590
310.5 301 2730
330 309.5 2860
376 331 3250
The core chronology was established by converting sediment depths to age with
two equations derived by linear regression for two intervals (Figure 8). The top of the
core was assumed to be modem.
152-0 cm: age (cal yr BP) = 10.976 x depth, r2 = 1;
376-152 cm: age (cal yr BP) = 6.8806 x depth + 599.66, r2 = 0.9947.
The age of the base of the core (563.5 cm) was estimated to be -4500 Cal yr BP based on
extrapolation of the linear regression from 152 to 376 cm. The radiocarbon date at 110.5
cm was not used to construct the chronology because of its small sample size and
suspicion that the charcoal fragments may have been displaced down core. Based on a
comparison between magnetic susceptibility records from SP-12-VI-02-1A and SP2-19-
VII-99, the radiocarbon date at 443.5 cm was not used to construct the chronology
because it caused deeper sediments to appear too old based on a comparison with core
SP-80-1. Sedimentation rates for the two intervals are as follows:
196.5-0 cm = 0.091 cm/yr;
563.5-196.5 cm = 0.142 cm/yr.
10 20 30 40 50 60 70 80 4 6 8 10 12 14 16
% CaCO3 5-pt Smooth Ca (103)
Figure 7: Correlation of SP2-19-VII-99 and SP-12-VII-02-1A using %CaCO3 in the SP2-
19-VII-99 core and the scanning XRF Ca concentration data from the SP-12-
0 100 200 300 400 500
Figure 8: Calibrated ages (yr BP) versus depth for correlated tie points in core SP-12-
VII-02-1A as well as radiocarbon dates of terrestrial samples in Lake Salpeten
core SP-12-VII-02-1A. Triangles indicate dates analyzed at LLNL-CAMS
while the diamonds represent tie points.
Elemental Geochemical Analyses
From 4500 cal yr BP to -3300 cal yr BP, sediments are dominated by organic
matter as reflected by high %OM and %oN concentrations and low magnetic susceptibility
and %CaCO3 (Figure 9). The %Other is low (-20%) at 4500 cal yr BP, and increases
continuously to the base of the Maya clay at 3300 cal yr BP. Beginning at the base of the
Maya clay unit at -3300 cal yr BP, magnetic susceptibility and %CaCO3 begin to
increase and %OM and /oN decrease abruptly. Sediment composition between -3300
and 1200 cal yr BP is dominated by inorganic sediment (clay and detrital carbonate) as
reflected by relatively high values of magnetic susceptibility, %CaCO3 and %Other.
While the %OM and %N are relatively constant during this time, the magnetic
susceptibility, %CaCO3 and %Other show some variation. %Other remains relatively
unchanged from 3300 to 2600 cal yr BP, reaching maximum values of 90% during this
time before decreasing to 50% between 2700 and 2400 cal yr BP. %Other remains
relatively constant from 2400 cal yr BP to 1600 cal yr BP, when values begin to increase
and reach 60% at 1200 cal yr BP. Following this maximum, %Other decreases
continuously from 1200 cal yr BP to the present. At -3300 cal yr BP, magnetic
susceptibility increases, rising to a maximum of -20 SI at 2700 cal yr BP before
decreasing to an average of 10 SI for the remainder of the Maya clay unit. %CaCO3
increases from 3300 cal yr BP to 2400 cal yr BP, reaching a maximum of -45%.
%CaCO3 then decreases to -4% at 1200 cal yr BP and remains low until the present. At
the top of the Maya clay unit, %OM and %N begin to increase and do so continuously
from 1200 cal yr BP to present.
Magnetic susceptibility increases from the base of the core to -4100 cal yr BP
(Figure 10). Values remain high from 4100 to -2000 cal yr BP reflecting the high clay
content of the sediment. Beginning at -2000 cal yr BP, magnetic susceptibility begins to
decline with two distinct steps centered at 1700 and 1300 cal yr BP. From 800 cal yr BP
to the present, magnetic susceptibility values remain low and relatively unchanged.
Magnetic Susceptibilily 1SlI
-10 -5 0 5 10 15 20 25
0 20 40 60 80 100
0 0.5 1 1.5 2 2.5 3
0 20 40 60 80 100 0 20 40 60 80 100
%CaCOi % Other
Figure 9: Magnetic susceptibility, percent calcium carbonate (%CaCO3), percent organic
matter (% OM), percent other (%Other) and percent nitrogen (%N) versus age in
calibrated years before present (cal yr BP) from Lake Sacnab. The gray highlighted area
represents the "Maya clay" unit.
5(000 I I I I I
-1 -0.5 0 0.5 1 1.5 2 2.5 3
Magnetic Susceptibility (cgs)
Figure 10: Magnetic susceptibility versus age in calibrated years before present (cal yr
BP) for core SP-12-VII-02-1A from Lake Salpeten. The gray highlighted area
represents the "Maya clay" unit.
Bulk Carbon and Nitrogen Isotopes
The carbon and nitrogen isotopic records of bulk organic matter show similar
patterns during the last 4650 years (Figure 11). Bulk 613C values range from -21%o to -
28%o throughout the core, whereas bulk 615N values vary between 0%o and 5%o. Carbon
isotopic values generally increase from -27%o at the base of the core to -23%o at -3300
cal yr BP. The 613C values of bulk organic matter average -23%o from 3300 to 2500 cal
yr BP, and then decrease between 2500 and 2000 cal yr BP. Bulk organic 613C remains
unchanged and averages -25%o from 2000 to 1300 cal yr BP. A sharp 2.5%o decrease in
613C occurs at 1250 cal yr BP, followed by a slight trend toward increasing values toward
The 615N of bulk organic matter averages -2%o from 4500 cal yr BP to 3500 cal
yr BP. From 3500 cal yr BP to 3300 cal yr BP, 615N values increase by 2%o and remain
high from 3300 cal yr BP to 1200 cal yr BP, averaging 3%o. The nitrogen isotopic
composition then decreases abruptly by 3.5%o at 1200 cal yr BP, and averages 1%o from
1200 cal yr BP to present.
Compound-Specific Carbon Isotopes
Samples were analyzed for the 613C of long-chain n-alkanes (C29, C31, and
C33) in Lake Sacnab sediments at approximately centennial resolution since 4650 cal yr
BP (Figure 12). The 613C signals for C31 and C33 show similar trends and will be
described in unison, whereas the 613C of C29 differs somewhat and will be described
The 613C of C31 and C33 averaged -33%o at the base of the core and gradually
increased to -29%o at -3300 cal yr BP. Values generally remain high from -3300 cal yr
BP to -2500 cal yr BP and averaged -28%o. From 2500 to 2100 cal yr BP, 613C
decreases from -28%o to -32%o and values remain unchanged from 2100 to 1300 cal yr
BP. At 1300 cal yr BP, the carbon isotopes of C31 and C33 record a rapid decrease with
values reaching -39%o and -37%o, respectively. Following this excursion, the 613C of n-
alkane chains C31 and C33 increases at 1100 cal yr BP and averages -32%o to the top of the
-1 0 1 2 3 4 5 -28 -26 -24 -22 -20
6 -N (%96, air) 8' C (%., PDB)
Figure 11: Bulk carbon and nitrogen isotopes for core SN-19-VII-97 from Lake Sacnab.
The gray highlighted area represents the "Maya clay" unit.
The 613C of C29 at the base of the core is -32%o and generally remains unchanged
from -4500 cal yr BP to -2500 cal yr BP, with an average ratio of approximately -29%o.
From 2500 cal yr BP to 2100 cal yr BP, 613C decreases from -29%o to -33%o and values
remain unchanged from 2100 cal yr BP and 1300 cal yr BP. At 1300 cal yr BP, the
carbon isotopes of C29 record a rapid decrease, reaching a value of -35%o. Following this
excursion, the 613C of C29 is highly variable, with values ranging from -24%o to -33%o
during the period from 1100 cal yr BP to the present.
8~"C C, (%., PDB)
-39 -36 -33 -30 -27 -24
-39 -36 -33 -30 -27 -24 -21 -39 -36 -33 -30 -27 -24
S13C C29 (%., PDB) &g1C C13 (%', PDB)
Figure 12: Compound-specific 613C results of long chain n-alkanes (C29, C31, and C33)
from Lake Sacnab, Guatemala versus age in calibrated years before present
(cal yr BP).
Samples were analyzed for the 613C of long-chain n-alkanes (C29, C31, and C33) in
Lake Salpeten sediments at an approximate resolution of 200 years for the past 6000 cal
yr BP (Figure 13). The 613C signals for C29, C31, and C33 show similar trends and will be
described in unison; variations in 613C of C29, however, are muted relative to C31 and C33.
The 613C of long-chain n-alkanes averaged -32%o at the base of the core and
gradually increased to -28%o at -5200 cal yr BP. Values generally remained unchanged
from -5200 cal yr BP to -3300 cal yr BP and averaged -30%o. From 3300 to 2500 cal yr
BP, 613C increased from -30%o to -27%o. The isotopic ratio decreased from 2500 cal yr
BP to the present, reaching minimum values that range from -33%o to -36%o.
a8 C31 (%o, PDB)
-35 -34 -33 -32 -31 -30 -29 -28
-34 -32 -30 -28
8 C C2, (%o, PDB)
-36 -34 -32 -30 -28
61C C33 (%o, PDB)
Figure 13: Compound-specific 613C results of long chain n-alkanes (C29, C31, and C33)
from Lake Salpeten, Guatemala versus age in calibrated years before present
(cal yr BP).
Implications for Changes in Land-use
Changes in land-use can be inferred from the lithological composition of
lacustrine sediments, which reflect changes in material transfer from the watershed to the
lake. With increased land clearance, the detrital load to a lake basin will increase.
Sediment cores taken from Lakes Sacnab and Salpeten show similar changes in sediment
composition. This common pattern has been documented throughout the central Peten
lake district and is interpreted to reflect sediment compositional and rate changes related
to the history of Maya settlement and population in the region (Cowgill and Hutchinson,
1966; Deevey et al., 1979; Brenner, 1983; Vaughan et al., 1985; Rice et al., 1985;
Binford etal., 1987; Leyden, 1987; Curtis etal., 1998; Islebe etal., 1996).
The sediment composition and magnetic susceptibility records in Lake Sacnab
(Figure 6) document changes in material transfer from the catchment to the lake.
Magnetic susceptibility is a proxy for the concentration of magnetic minerals contained
within the sediment, which are derived from the erosion of soil and bedrock in the
watershed and subsequently transported to the lake. CaCO3 in sediments can be derived
from either authigenic precipitation within the lake or by the weathering and transport of
detrital carbonate from soils or limestone bedrock. There is little to no observed
lacustrine carbonate in the sediments of Lake Sacnab, which may suggest that most of the
CaCO3 is detrital. Inorganic carbon and oxygen isotopes measured on bulk material from
the lake have values that range from -8%o to 0%o and -4%o to 0%o, respectively. These
values are generally consistent with measured mean values of Peten soils surrounding
Lake Quexil (613C -10%o to -7%o, 6180 -3%o to 0%o,) (Leyden et al., 1993) and
indicate that the source for sediment CaCO3 is dominantly allochthonous CaCO3 as
opposed to diffuse authigenic CaCO3. Thus, relative shifts in %CaCO3 may serve as an
additional indicator of erosion in the catchment. Organic matter in lacustrine sediments is
derived from multiple sources including terrestrial vegetation, lacustrine algae and
bacteria, and aquatic macrophytes. The %Other may also serve as a proxy indicator of
landscape erosion. Because there is relatively little contribution to the sediments other
than CaCO3 and organic matter, the %Other is a likely indicator of non-carbonate clastic
material (clay) eroded from the watershed. It is important to note the effects of the
closed-sum problem inherent in using weight % data. Dilution may play a major role in
down-core variations of any of the three components. As the % of any component
increases, the other two components will respond by decreasing in % to sum to 100%.
Therefore, variation in the delivery of any of the sediment components, especially during
the deposition of the Maya clay, would significantly alter the apparent input of the other
sediment components. To avoid the closed-sum" problem, accumulation rates (g/cm2/yr)
can be calculated. Accumulation rates of the three individual sediment components
provide a better indication of erosion and sediment deposition. Unfortunately, the density
data necessary to calculate accumulation rates are not available for Core SN-19-VII-97
from Lake Sacnab. Anselmetti et al. (in prep), however, calculated accumulation rates
for specific time intervals in Lake Salpeten for the period between 8500 cal yr BP and the
present. Data indicate that erosion rates were lowest during the early to mid-Holocene
and increased beginning in the Early Preclassic. Erosion rates were highest during the
Late Preclassic period (2200 to 1700 cal yr BP) and subsequently decreased over time
even during the height of Maya occupation in the watershed (1700 and 1100 cal yr BP).
The sedimentological history and associated watershed erosional characteristics
determined by Anselmetti et al. (in prep) are consistent with those determined in this
In Lake Sacnab, Holocene sediment prior to Maya occupation is classified as
gyttja. Overlying the pre-Maya gyttja is a clay-rich horizon known as the "Maya clay",
associated with increased sedimentation rates. The onset of Maya clay deposition
appears to have begun at 3300 cal yr BP and lasted until -1200 cal yr BP. This unit is
common in all the central Peten lakes and has been interpreted previously to reflect
accelerated erosion associated with Maya land clearance (Deevey et al., 1979; Brenner,
1983; Vaughan et al., 1985; Rice et al., 1985; Binford et al., 1987; Leyden, 1987; Curtis
et al., 1998; Islebe et al., 1996) and/or possibly regional drying (Hodell et al., 1995;
Curtis et al., 1998). In the Lake Sacnab watershed, soils likely became unstable as forest
cover was removed at the onset of Maya occupation. The common soil type in the Peten
are mollisols, consisting thin (typically <1 m) mineral soils that develop over CaCO3-rich
material called sascab (Brenner et al., 2002).
Down-profile erosion of Peten soils following deforestation is evidenced in the
sedimentological history. Changes in the ratio of %Other to %CaCO3 suggest that the
organic- and clay-rich surface horizon was eroded between 3300 and 2500 cal yr BP.
Once the watershed had been denuded of organic-rich top soils, the weathering and
erosion of more carbonate-rich, deep soils (sascab) ensued from 2500 to 1300 cal yr BP,
as evidenced by the increase in the influx of CaCO3 and decrease in %Other (Figure 6).
This period (between 3300 and 1200 cal yr BP) was characterized by enhanced delivery
of detrital material to the lake and resulted in sediments with a low organic content,
which is likely a dilution effect.
Overlying the Maya clay unit is another organic- and clay-rich layer that is
inferred to represent decreased erosion following the cessation of agriculture and
increased contribution of lacustrine organic matter to the sediment. Stratigraphic changes
in sediment composition in Lake Sacnab are similar to those in other lakes in the Peten
determined in previous studies (Deevey et al., 1979; Brenner, 1983; Vaughan et al.,
1985; Rice et al., 1985; Binford et al., 1987; Leyden, 1987; Curtis et al., 1998; Islebe et
al., 1996) and suggests that the history of human and environmental changes is similar
throughout the central Peten region.
The magnetic susceptibility record from Lake Salpeten shows a similar trend to
Lake Sacnab in the transfer of watershed material to the lake but with a distinct
difference in timing (Figure 10). The onset of Maya clay deposition, indicated by the
enhanced delivery of watershed-derived detrital material resulting from landscape
denudation, appears to have begun at 4100 cal yr BP and lasted until -1100 cal yr BP.
Previous work (Rosenmeier et al., 2002) in Lake Salpeten, however, has indicated that
the Maya Clay was deposited between -3300 and 1100 cal yr BP. The apparent
discrepancy between the two cores is most likely related to the chronology of core SP-12-
VI-02-1A (this study), which was correlated to core SP2-19-VII-99 (Rosenmeier et al.,
2002) using ten tie points (see Chapter 4). In using this method, there is the possibility
for miscorrelations that would cause error in the chronology. Core SP-12-VI-02-1A was
dated indirectly and ages for the Maya clay from Rosenmeier et al. (2002) are probably
more accurate. Archaeological settlements in the Lake Salpeten watershed are
documented beginning at -3000 cal yr BP (Figure 3) (Rice and Rice, 1990) and roughly
coincided with the sedimentological evidence for human occupation of Rosenmeier et al.
For the most part, compound-specific carbon isotopes track magnetic
susceptibility closely throughout the records for both Lake Sacnab and Lake Salpeten
(Figures 14, 15). The magnetic susceptibility record is generally considered a proxy for
changes in amount of detrital material transported from the landscape into the lake that,
in turn, is affected by the amount of vegetation in a watershed. In both lakes, periods of
inferred increased erosion occur simultaneously with periods of higher relative
contributions of C4 vegetation. As vegetation shifts from high-forest taxa to a more
savanna-like landscape during forest removal, rapid erosion of watershed soils resulting
from soil destabilization ensues. Because both magnetic susceptibility and the
compound-specific records are influenced by vegetation cover and type and are indicative
of landscape conditions, the correlation of these proxies was not unexpected.
Sources of Organic Matter
Organic matter in lacustrine sediments is derived from multiple sources including
terrestrial vegetation, lacustrine algae, and aquatic macrophytes. The C/N ratio can
provide information about the proportions of algal versus terrestrial plant contribution to
organic matter (Prahl et al., 1980; Meyers, 1994; Kaushal and Bindford, 1999). Organic
matter from algae and bacteria has C/N weight ratios ranging between 5 and 12, whereas
organic matter from vascular land plants usually has weight ratios of 24 and greater
(Meyers, 1994). While C/N ratios between 36 and 48 are generally characteristic of
cellulose-rich vascular plants, weight ratios between -14 and 20 suggest a mixture of
both algal and vascular plant material (Ertel and Hedges, 1985).
I5"C C-, J%. PDI
-34 -36 -IS
SI I I I I I
W 25 20 15 10 5 0 -5 -10
Magnetic Susceptibility (Sl)
Figure 14: Comparison of 613C of C33 from Lake Sacnab with magnetic susceptibility.
6~C vy~., P 1B1
3 2.5 2 1.5 I 0.5 0 -0.5 -I
Magnetic Susceptibility (cgs)
Figure 15: Comparison of 613C of C33 from Lake Salpeten with magnetic susceptibility.
In Lake Sacnab, the C:N ratio suggests that the source of organic matter to the
lake may have changed over the last 4500 years (Figure 16). Values range from 10 to 20
in the earliest part of the record, from 4500 to 2700 cal yr BP, and suggest a mixture of
both algal and terrestrial plant material. From 2700 cal yr BP to 1300 cal yr BP,
increased ratios of C:N suggest the predominant contribution of terrestrial vegetation to
sediment organic matter. Values are low from 1300 cal yr BP to the present, ranging
only between 9 and 13.5, suggesting a return to dominantly aquatic organic matter.
The distinguishing C:N ratios among the different types of organic matter
generally survive both sinking and sedimentation in lacustrine environments. However,
certain diagenetic processes such as dissolution, oxic/anoxic cycles and bacterial
processes may modify the original ratios (Muller and Mathesius, 1999). For this reason,
C:N values are often used together with the carbon isotope ratio of bulk organic matter to
determine both the sources) and composition of organic matter.
The period between 4500 and 2700 cal yr BP has bulk isotope values that suggest
an increasing contribution of a mix of C3 and C4 land plants while the C:N ratio suggests
additional contribution from algae. From 2700 to 1300 cal yr BP, both the C:N and the
bulk isotopes suggest a dominant contribution from C4 terrestrial plants and is followed
by a period between 1300 cal yr BP and the present in which there is a higher
contribution from algae, aquatic macrophytes and C3 terrestrial vegetation.
0 5 10 15 20 25 30 35 40 45 50
Figure 16: C:N ratios in weight % and bulk organic matter isotopes from Lake Sacnab.
-27 -26 -25 -24 -23 -22 -21 -20
a'C (%o, PDB)
< 3000 -
o Bulk Organic Matter
5000 C 1 1
-40 -35 -30 -25 -20
813C (%o, PDB)
Figure 17: Comparison of 613C values for bulk organic matter, C31 and C33 from Lake
The 613C of bulk organic matter in Lake Sacnab sediments shows similar trends
as the compound-specific record which might indicate that the bulk organic matter within
the lake sediment is dominated by terrestrial organic matter with only a small
contribution from aquatic macrophytes and algae (Figure 17). However, the C:N ratios
suggest that the sources) of organic matter changed over time and is likely a better proxy
for the sources) of organic matter. The bulk and compound-specific isotopes show the
most distinct variation in trend in the latest part of the record, from -1200 cal yr BP to
the present and may be explained by the increased contribution from aquatic macrophytes
and lacustrine algae as demonstrated by the C:N ratio. Carbon isotopic values of bulk
organic matter are enriched by -5%o to 7%o, however, relative to individual leaf wax n-
alkanes. This is expected because lipids, including n-alkanes, are commonly depleted in
13C relative by ~6%o -to 8%o relative to other biosynthetic products (Hayes, 2001). Thus,
much of the terrestrial organic matter in the lake deposits owes its origin to non-lipid
An important concern in analyzing the isotopes of individual n-alkanes is
preservation potential. The carbon preference index (CPI) is often used as a proxy for the
preservation potential of the organic matter when there is a clear predominance of
epicuticular leaf waxes of terrestrial plants (Hedges and Prahl, 1993). CPI values are
generally highest in living plants and surface sediments. Lowered values indicate
increasing maturity and degradation and tend to decrease to a final value of 1 (Hedges
and Prahl, 1993). However, this index cannot be used as a preservation potential proxy in
all cases. For example, a CPI of 1 may indicate immature organic matter with a low
contribution from higher plants rather than mature organic matter. In order to determine
whether the CPI can be used as a proxy for preservation potential, the examination of the
predominant n-alkane chain length can provide some insight into the provenance of
organic matter. The predominant n-alkane chain length is shown in Figure 18a, which is
interpreted as an indicator of the origin of the organic debris input (algae, aquatic
macrophytes or land plants) (Cranwell et al., 1987; Ficken et al., 2000). The
predominance of n-alkanes of high-molecular-weight indicates that terrestrial plants are
the predominant organic matter source and suggests that the CPI can be used as a proxy
for preservation potential.
With respect to the CPI, most of the values are greater than 1.5 with few
exceptions (Figure 18b). The CPI shows a predominance of immature, plant-derived
material and, therefore, very little removal of components during transport and post-
deposition (Ortiz et al., 2004).
20.00 2200 24.00 26_00 2B.00 30 00 3200
5 10 15 20 25
Figure 18: Predominant n-alkane chain length (A) and CPI (B) for Lakes Sacnab and
Evidence for Relative Shifts in Vegetation
Sediments from Lakes Sacnab and Salpeten show a range of values (-38%o to -
27%o,) during the last 4500 years that indicate changing relative proportions of C3 to C4
vegetation in the watershed (Figure 11). A comparison of the 613C records of C33 from
Lake Sacnab and Lake Salpeten reveals similar absolute values and long-term trends over
the past 4500 cal yr BP (Figure 19). At the base of the core, more depleted values
suggest a higher proportion of C3 versus C4 vegetation. A shift towards more enriched
values from 4500 cal yr BP to 3300 cal yr BP reflects an increase in the relative
abundance of C4 (grasses, etc.) to C3 (trees) vegetation. Decreasing 613C values from
2500 cal yr BP to 2000 cal yr BP indicates some forest regrowth and a higher proportion
of C3 taxa than in the previous interval. The proportion of C3 to C4 vegetation is
relatively unchanged from 2000 cal yr BP to 1300 cal yr BP, which is surprising given
the population changes that were occurring in this period. At 1300 cal yr BP in the
Sacnab watershed only, a sharp decrease in the 613C indicates a rapid transition (-100
years) to a highly C3-dominated landscape. During the last -1200 cal yr BP, the carbon
isotopic record from both lakes suggests a gradual shift towards greater contribution of
C3 vegetation toward present.
There are several distinct differences in the 613C records of the two cores. A
peak in the 613C in Lake Salpeten at approximately 3900 cal yr BP and subsequent
decrease until 3300 cal yr BP is not observed in Lake Sacnab. Rather, values are
continually increasing during this time in Lake Sacnab. In addition, the 613C peaks at
approximately 3100 cal yr BP in Lake Sacnab, whereas maximum 613C values in Lake
Salpeten do not occur until 2800 cal yr BP. The rapid transition in 613C at 1300 cal yr BP
in Lake Sacnab is not present in the Salpeten record. The differences in the 613C records
may be due partly to sampling resolution, which is approximately two times higher in
Lake Sacnab than in Lake Salpeten. In addition, some variation may be expected in the
history of vegetation changes in the two watersheds. While the general trends of land-use
and human occupation are thought to be similar between watersheds, these systems are
dynamic and the history of land use was probably variable throughout the region.
-36 -34 -32 -30 -28 -26
S3C C33(?o, PDB)
Figure 19: Comparison of 613C records for n-alkane chain C33 in Lake Sacnab and Lake
The rapid decline in 613C in the Lake Sacnab core at 1300 cal yr BP may reflect
increased dominance of a single species with a relatively depleted carbon isotopic value
after agricultural abandonment. Various chronosequence studies of abandoned
agricultural fields have determined that soil nitrogen is often the limiting factor during
the entire succession of vegetative regrowth (Tilman, 1984; Tilman 1987; Knops and
Tilman, 2000). In many cases, legumes are the first plants to colonize abandoned fields
because they are able to fix atmospheric nitrogen. If the pioneering species were
legumes, then it would have a low 615N value because of nitrogen fixation (615N = 0),
which involves little isotopic fractionation between plant tissues and air. In fact, the
rapid transition in the 613C record in Lake Sacnab occurs at the same time (-1300 cal yr
BP) as a rapid decrease in the 615N of bulk organic matter (Figure 20). These data
support an increased contribution of sediment organic nitrogen from N2 fixers because
sedimentary 615N values at this time are close to zero. While the average 613C of
legumes is not necessarily more negative than other plants, they are exclusively C3 plants
and thus have low 613C values relative to C4 vegetation.
An alternative explanation for the rapid transition at 1300 cal yr BP is the result of
dominance of a non-leguminous pioneer species. For example, it has been documented
that in certain systems, bracken fern often dominates the vegetation in a field
immediately following abandonment (Pakemen et al., 1994). Eventually, other high
forest taxa out-compete this weed and would thus change the isotopic composition of
sediments gradually over time. While the 613C value of bracken fern is not reported in
the literature and is not necessarily the source for the depleted values in this study,
previous work shows that one species may dominate following agricultural abandonment.
It may be possible to determine whether a pioneer species accounts for the dramatic
isotope shift by analyzing pollen samples at close intervals above, within, and below the
613C C33 (%o, PDB)
-39 -37 -35 -33 -31
C 1--I I-I
-29 -27 -25
Figure 20: Diagram comparing the 613C of C33 to the 685N of bulk organic matter in Lake
0 1 2 3 4
815N Bulk (%o, air)
Reported 813C values of n-alkanes in C3 plants range from -31%o to -38%o,
whereas n-alkanes in leaf waxes of C4 plants typically range from -19%o to -25%o
(Freeman, 2001). The 613C values for n-alkanes in this study, however, range from -27%o
to -38%o which might suggest that C3 vegetation dominated throughout much of the study
period. Previous studies of aerosols and sediments have translated 613C values into
percentages of C3 versus C4 plants using a two-component mixing equation (Huang et al.,
2000; Schefub et al., 2003). It has been found, however, that the isotopic values of C3
plants usually become more depleted with increasing carbon number while C4 plants have
isotopic values that are consistent over all chain lengths (Collister et al., 1994). The
variation is suggested to result from the production of different leaf wax lipids in
different proportions during a leafs growth cycle and averages approximately 2.4%o.
This variation translates into a potential 16% error in the calculation of C3 to C4
abundance. This characteristic makes it difficult to quantify the contribution of C3 versus
C4 plants. In order to account for this effect, the values for the lowest-number terrestrial
n-alkane chain, C29, are most often used to translate 613C values into % contributions.
Because aquatic macrophytes can contribute significant amounts of C29, it is preferable to
calculate %C4 using an n-alkane of greater chain length. For this study, values were
calculated using a two-component mixing equation assuming end-member 613C values of
n-alkane C33 of -36.4%o for C3 plants and -19%o for C4 plants, respectively. These values
were adapted from those of Collister et al. (1994), who reported -34%o for the C3-
endmember and -19%o for the C29 C4-endmember.
For Lake Salpeten, the contribution of C4 vegetation to the C33 n-alkane pool
ranges between 14% and 57.5% (Figure 21). The % contribution of C4 taxa increases
between 4500 and -2700 cal yr BP when the greatest proportion of C4 occurred. The
contribution of C4 biomass declines from nearly 58% at 2700 cal yr BP to just 14% at
C4 Vegetation (%)
0 10 20 30 40 50 60 70
Figure 21: Relative shifts in contribution of C4 vegetation (in %) in Lakes Sacnab and
Salpeten over the last -4500 cal yr BP. Values were calculated using a two-
component mixing equation with C3 and C4 inputs represented by 613C values
of n-alkane C33 of-36.4%o and -19%o, respectively (Collister et al., 1994).
Lake Sacnab shows similar shifts to Lake Salpeten with respect to % contribution
of C4 biomass. From 4500 cal yr BP until -3300 cal yr BP, the % contribution increases,
reaching peak values of ~55%o between 3300 and 2700 cal yr BP. From 2700 cal yr BP
to 1300 cal yr BP, the contribution of C4 vegetation generally decreases. At 1300 cal yr
BP, the contribution of C4 vegetation is 0%, but rapidly increases to a greater contribution
(30%) by 1200 cal yr BP. The contribution of C4 vegetation gradually decreases from
1200 cal yr BP to the present.
Changes in water-use efficiency (WUE), i.e. the ratio of carbon gained to water
lost during gas exchange, may also affect the 613C of vegetation. A negative correlation
exists between precipitation and 613C of vegetation among tropical sites (Leffler and
Enquist, 2002). Consequently, the 613C of C3 vegetation would increase during periods
when the precipitation decreased significantly. Consequently the calculated % C4
contribution may be greater than the true value. It is thus necessary to consider the
potential influence of climate on the 613C record.
The 613C values of n-alkanes in Lakes Sacnab and Salpeten are similar to values
obtained by Huang et al. (2001) in Lake Quexil (Figure 22). The 613C values during the
Maya clay interval in Quexil averaged -31%o and are similar to values measured in the
Maya clay from Lake Sacnab (average = -30%o). Values from Lake Quexil in the pre-
Maya gyttja average -34%o and increased by -5%o between 3300 and 1100 cal yr BP,
with a peak in 613C during the time of the Maya Terminal Classic period. They attribute
this increase to both anthropogenic forest clearance and regional drying. The 613C from
Lake Sacnab, however, reaches peak values much earlier at -3000 cal yr BP than in Lake
Quexil. The Quexil carbon isotope record is much lower resolution than either Lakes
Sacnab or Salpeten, and may not capture the structure observed in the higher-resolution
1"C C, (%o, PDB)
-40.0 -35.0 -30.0
- 0 ---
13 C 31 (%o, PDB)
Figure 22: Diagram showing the comparison between 613C values of C31 from Lake
Sacnab, Lake Salpeten and Lake Quexil (Huang et al., 2001).
In the records from Lake Salpeten and Sacnab, the 613C values of n-alkanes from
3300 cal yr BP to 2700 cal yr BP are comparable to values during the LGM in Lake
Quexil, further supporting the interpretation of enhanced C4 input. The data imply that
vegetation during the period between 3300 cand 2700 cal yr BP may have been quite
similar to the vegetation during the Last Glacial Maximum, when climate was
significantly cooler. It is remarkable that vegetation may have changed as drastically
during the period of human occupation as it did during a glacial-interglacial cycle.
Comparison with Pollen Records
Compound-specific carbon isotope records can be compared with pollen records
from the same or nearby lakes to test if vegetation changes inferred from the 613C of
long-chain n-alkanes are the same as those inferred from pollen profiles. A comparison
of the % disturbance taxa from Lake Salpeten (Leyden, 1987) and the long-chain n-
alkane 613C record from Lake Sacnab reveals significant differences in the two proxies
(Figure 23). Disturbance taxa include grasses, sedges, and herbs from the following
families: Amaranthaceae (C3 and C4), Ambrosia (C3), Compositae (C3 and C4),
Cyperaceae (C3 and C4) and Gramineae (C3 and C4). While these pollen data may
provide an accurate representation of changes in the relative abundance of the selected
taxa, they are not ideal for distinguishing changes in the relative contributions of C3
versus C4 plants because both plant types are represented by the disturbance taxa.
Perhaps a better way to compare the compound-specific and disturbance taxa is to
compare 613C to grasses only, which would be most representative of C4 contribution.
A comparison of compound-specific carbon isotopes to % grasses reveals that
while both records increase from the base of the record to 3300 cal yr BP, only the 613C
of C33 peaks between 3300 and 2500 cal yr BP (Figure 24). The % grasses in the pollen
record, however, is relatively low during this period (9%) and does not peak until -2300
cal yr BP. Both the % grass and isotope records generally decline from 2300 cal yr BP to
present. The pollen (% grasses) record is lower resolution than the isotope record; which
may explain some of the discrepancy between the two vegetation proxies.
Perhaps the largest potential reason for the discrepancy may lie in the fact that
maize pollen is not included in the total pollen count because of its large size, thus
causing an over-representation of other taxa. Maize pollen has often been used in
Mesoamerican studies as a proxy for agriculture and associated deforestation and it is
well-documented that maize pollen is abundant during the period of Maya occupation
(Leyden, 1987; Islebe et al., 1996). The 613C of n-alkanes, however, should be very
sensitive to large stands of maize in the watershed.
There are additional potential shortcomings in using the 613C of n-alkanes as a
proxy for vegetation change. For example, maize was probably an important (C4) plant
in the vegetation of the Peten Lake District, especially during times of high population
density. Shoreline cultivation of maize would strongly influence the compound-specific
613C while not altering the pollen profile. The higher contribution of C4 vegetation during
the Preclassic (-3300 cal yr BP to 2500 cal yr BP) may represent early shoreline maize
cultivation in the watershed. As populations increased into the Classic period, the more
desirable shorelines may have become residential areas as opposed to agricultural areas,
which would have then been moved further from the lake. Abandonment of near-shore
fields would alter the compound-specific 613C but would not be apparent in the pollen
profile. Alternatively, maize cultivation on shorelines may have ceased in the Late
Preclassic due to soil depletion. As the soils immediately surrounding the lake were
exhausted, the ancient Maya may have moved agricultural fields further into the
While there are potential explanations for the discrepancy between pollen and
compound-specific 613C values, it is necessary to fully understand that both records are
recording different aspects of watershed vegetation. Vegetation inferences from pollen
percentages represent the relative abundance of pollen grains in a sediment profile and do
not necessarily reflect species abundance or biomass on the landscape (Bradley, 1999)
whereas compound-specific carbon isotopes do not reveal any information about forest
Pollen and leaf waxes are derived from different vegetative sources and thus
record different aspects of watershed vegetation (Huang et al., 1999). Whereas pollen is
a measure of only reproduction, leaf waxes provide a more representative measure of
vegetative biomass within a watershed. The exclusion of maize pollen provides yet
another complication for interpreting pollen profiles. Compound-specific carbon
isotopes, on the other hand, do not reveal any detailed information about forest
composition. They simply allow estimation of the relative contribution of C3 versus C4
n-alkanes to the sedimented organic matter. When examined in conjunction with pollen
accumulation rates, however, leaf waxes and pollen may provide a better estimate of
vegetative biomass. The carbon isotopic composition of leaf waxes is a good
geochemical proxy for testing palynological inferences for vegetation changes within a
S130 033 (%o, PDB)
Figure 23: Diagram showing % grass pollen versus 613C of n-alkane C33 in Lakes
Salpeten and Sacnab and the presence of maize pollen in Salpeten (pollen data
from Leyden, 1987).
Comparison with Population Estimates
The correlation between Maya population densities and 613C values of long-chain
n-alkanes in lake sediments suggests that while vegetation change over the past 4500 cal
yr BP may have been tied to changes in population density, additional factors may have
affected vegetation in the watershed in the early part of the record. Pollen records
(Leyden, 1987; Islebe et al., 1996) reveal that disturbance taxa and maize increased as
populations increased. Pollen changes were most likely tied to agricultural land
clearance. Because grasses and maize are C4 plants, one would predict an increase in
the 613C of n-alkanes as populations grew. The largest shifts in vegetation, however, in
both Lakes Sacnab and Salpeten, occurred well before the peak in late Classic Maya
populations (Figure 24). The highest 613C values, indicating the largest contribution
from C4 vegetation, occurs in Lake Sacnab between 3300 cal yr BP and 2700 cal yr BP
and in Lake Salpeten at 2700 cal yr BP, during the Preclassic Period. In both lakes there
appears to be a decrease in C4 biomass between 2700 cal yr BP and 1300 cal yr BP, the
time period during which population density was greatest.
This decoupling of population density and the 613C of long-chain n-alkanes
between 4500 cal yr BP and 1300 cal yr BP may be a result of several factors. For
example, if agricultural practices changed through time from more extensive to intensive
methods, a return to a C3-dominated landscape may coincide with both population growth
and agricultural developments. The question regarding whether the Maya maintained
economic trees and house gardens within cities remains unanswered, but researchers have
made speculations in some instances. For example, Leyden (1987) notes that the
presence of ramon (Bromsimum) pollen suggests that the Maya were "arboriculturists",
possibly growing tree gardens in typically residential and common areas. Selective
clearing and/or tree planting, however, is not indicated by the rest of the pollen record.
Instead, greater proportions of corn pollen during the Late Classic through Postclassic
indicate intensified agricultural activities. The interpretations made from the pollen
record (that ramon does not necessarily indicate arboriculture) are supported by the work
of Lambert and Arnason (1982), who showed that ramon is attracted to constructions of
limestone and is positively correlated with Maya sites; this would indicate not
arboriculture, but instead settlement expansion.
Lake Salpeten 0
-34 40 I
0 1000 2000 3000 4000 5000
Age (cal yr BP)
Figure 24: Comparison of the compound-specific carbon isotope records (C33) and
population density estimates versus time in Lakes Salpeten (top) and Sacnab
It is also possible that, as discussed in the previous section, the higher
contribution of C4 vegetation during the Preclassic (-3300 cal yr BP to 2500 cal yr BP)
may represent early shoreline maize cultivation in the watershed. As populations
increased into the Classic period, the more desirable shorelines may have shifted to
residential areas as opposed to agricultural areas, which would have then been relocated
further from the lake. Abandonment of near-shore fields would be represented by a
decoupling of the compound-specific 613C and population because the agricultural fields
may have expanded away from the lake shore as populations expanded in the Classic
Another possibility for the lack of coherence between population estimates and the
compound-specific vegetation record may lie within the population estimates themselves.
The population estimates in the Peten are based on a limited number of survey transects
and house-mound excavations. There is a possibility that existing sites may not have
been revealed during survey transects. In addition, older Preclassic population may have
been underestimated if house mounds of the period were poorly preserved in the
archaeological record. In addition, the chronology for population estimates was based on
ceramic phases and inherently has a significant amount of error associated with it. While
estimates of population densities in the watersheds may not be entirely accurate and may
potentially be off by orders of magnitude the relative changes in population density are
correct, and populations were certainly greater in the Classic than Preclassic Periods.
Despite the divergent trends in population density and proxy vegetation in the early
part of the record, the records display similar trends between 1200 cal yr BP and the
present. A decrease in the compound-specific carbon isotopic ratios begins at 1200 cal
yr BP and correlates with a significant decline in population associated with the decline
of the Maya civilization. As human pressures on the landscape were curtailed and fields
were abandoned, watershed vegetation shifted from a C4 to a C3-dominated landscape.
This indicates that the decline in population density was likely the main cause of late
Holocene vegetation change.
Relationship between Climate and Environmental Changes
Finally, the relative importance of humans versus climate for vegetation change in
the Central Peten can be determined by comparing the 613C of n-alkanes with climate
proxies that are not confounded by human impact. These comparisons may reveal
whether climate played any role in environmental change. If vegetation changes in a
watershed were climate-induced, long-chain n-alkanes should correlate with independent
evidence for regional climate change. Proxy records of climate change are available for
both the Yucatan and the Caribbean Sea and can be compared with compound-specific
analyses and evaluated in regards to changes in Maya population densities.
A comparison with the %Ti record from the Cariaco basin (Haug et al., 2001)
(Figure 25) reveals a relationship between %Ti and the compound-specific records from
each lake during the early part of the record. In the Cariaco Basin, %Ti in the sediments
is used as a proxy for terrigenous sediment input, which is influenced by regional dry/wet
cycles. Higher %Ti reflects higher terrigenous input and greater precipitation.
The %Ti record from 10,000 cal yr BP to 4000 cal yr BP is high and relatively
unchanging, indicating relatively mesic conditions. At approximately 4000 cal yr BP,
however, %Ti values decrease and show greater variability indicating generally reduced
precipitation, perhaps related to the southward migration of the Intertropical Convergence
Zone (ITCZ). Increased variability in %Ti is especially pronounced from 4000 cal yr BP
to 2000 cal yr BP. The apparent relationship of 613C of n-alkanes to climate suggests that
vegetation may have been partly controlled by changes in regional climate rather than
local cultural changes over some time period.
Increased climate variability during the period between 4000 cal yr BP and 2000
cal yr BP may have altered precipitation patterns and the vegetation in Peten. An
extended dry period with sporadic periods of greater precipitation is an effective way to
stress vegetation and to erode the landscape. Prior to 2000 cal yr BP, changes in regional
climate may have influenced vegetation in addition to human disturbance.
Further evidence for climate-influenced vegetation change can be found in
sediment records from numerous lakes in northern Yucatan. A sediment core from Lake
Chichancanab provides evidence for regional drying that occurred beginning at
approximately 3000 cal yr BP with a distinct interval of droughts between 1300 cal yr BP
and 1100 cal yr BP, which coincided with the Terminal Classic collapse (Hodell et al.,
1995; Hodell et al., in press). Compound-specific carbon isotopes (Figure 15 ) during the
period between 3300 and 2700 cal yr BP in both Lakes Sacnab and Salpeten represent the
highest proportion of C4 versus C3 vegetation, as would be expected during a period of
increased climate variability. One would expect to see continued increased contribution
of C4 vegetation throughout the period of increased climate variability but this is not the
case. This discrepancy may suggest that while climate influenced vegetation change
beginning at 3000 cal yr BP, human land-use change exacerbated the deforestation
process from 3000 cal yr BP until Maya populations decreased at 1100 cal yr BP.
Ti (%, 3-pt smooth)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
-22 -24 -26 -28 -30 -32 -34 -36 -38
S130 033 (%o, PDB)
Figure 25: Comparison of the compound-specific carbon isotope records (C33) and
percent Ti versus time in Lakes Salpeten and Sacnab. Percent Ti data from
Haug et al. (2001).
The 6180 of ostracod carbonate in Lake Miragoane, Haiti (Hodell et al., 1991)
records a similar climate history to that for the Cariaco basin (Haug et al., 2001). The
isotope record, indicating changes in evaporation/precipitation and lake level, indicates a
switch from dry forest vegetation to more mesic conditions at approximately 8200 cal yr
BP. Beginning approximately 3800 cal yr BP, however, the isotopic value of shell
carbonate begins to increase coincident with a loss of mesic forest trees and a regional
drying event is suggested. These data correspond to local vegetation changes in the
Sacnab and Salpeten basins; as climate became drier, vegetation shifted from a C3 to a
C4-dominated landscape. This further supports the interpretation that (-4000 cal yr BP to
3000 cal yr BP), climate may have played some role in vegetation change.
The questions addressed in this study explore the dynamics of environmental
change with respect to ancient Maya population shifts and late Holocene climate changes.
While the analysis of compound-specific 613C values to study vegetation shifts is still in
its infancy, this is the first late Holocene record for the Peten and the first to compare the
613C of n-alkanes with a local pollen record.
The data indicate that in the watersheds of Lakes Salpeten and Sacnab, shifts in the
proportion of C3 to C4 are most likely controlled by a combination of climate change and
human deforestation. The correspondence of 613C records to independent proxies for
climate change from -4500 cal yr BP until -3000 cal yr BP suggest that regional drying
and increased climate variability caused an increase in the contribution of C4 vegetation
during that time. Following this period and beginning with the first Maya occupation in
the watersheds, vegetation change was likely a result of human-driven deforestation or
perhaps a combination of both climate and human impact.
The lack of concordance between the compound-specific carbon isotopic record
and the pollen profiles between 3000 cal yr BP and 1100 cal yr BP may be due to the
underlying ambiguities in comparing percent pollen data with compound-specific carbon
isotopic measurements. The only true assessment of the C4 pollen signal is to examine
shifts in the absolute values of grass and maize pollen influx. The higher contribution of
C4 vegetation during the Preclassic inferred from compound-specific 613C (-3300 cal yr
BP to 2500 cal yr BP) may represent early shoreline maize cultivation in the watershed.
As populations increased into the Classic period and shorelines shifted to residential as
opposed to agricultural areas, cultivated fields would have then been moved further into
the surrounding watershed. This abandonment of fields and subsequent cessation of
maize cultivation on the shoreline may strongly alter the compound-specific 613C but
would not be apparent in the pollen profile.
Discrepancies between vegetation inferred from pollen profiles versus 613C of n-
alkanes may also lie within the interpretation of the compound-specific 613C record itself
While 613C of n-alkanes is an excellent proxy for estimating the relative changes in
biomass contribution of C3 and C4 vegetation, it cannot be used to interpret forest
composition or the dynamics of tropical reforestation. In addition, there is no clear
understanding of how issues such as differential leaf production and the amount of leaf
wax a plant produces may affect the 613C of n-alkanes. Another potential problem in
interpreting 613C values is that variation in canopy density that alters both light regimes
and the 613C of source CO2 air for plants, which in turn affects the 613C of plant material.
While an important factor, it is difficult to quantitatively assess the canopy effect on 613C
values of n-alkanes. Lastly, the relationship between 613C and water-use efficiency
(WUE) in C3 plants may cause 613C values to appear more depleted and thus suggest a
higher apparent contribution of C3 vegetation. With these limitations in mind, it is
necessary to interpret compound-specific carbon isotope records with caution.
While the decoupling of population density and the 613C of long-chain n-alkanes
may suggest that agricultural practices changed through time from more extensive to
intensive methods, it is impossible to determine this from the 613C from long-chain n-
alkanes alone. The 613C from long-chain n-alkanes need not necessarily track population
change if a shift in the proximity of agricultural lands (and proximity of maize pollen,
which is large and does not travel far) to the lake was influencing the 613C. The
movement of fields and subsequent cessation of maize cultivation on the shoreline would
strongly alter the compound-specific 613C but would not necessarily be reflected by
changes in population densities.
The interpretation of compound-specific carbon isotopes of long-chain n-alkanes is
dependent upon the area surrounding the lake over which the compounds integrate.
There is unfortunately very little information regarding transport and deposition of n-
alkanes in lacustrine studies. If compound-specific carbon isotopes of long-chain n-
alkanes are more strongly influenced by vegetation on water-edge lands, than they may
not provide an accurate picture of overall watershed vegetation and land-use. Studies that
attempt to calibrate the compound-specific carbon isotopic value of modem sediments
with the surrounding forest structure may greatly aid in interpreting sediment records of
Compound-specific carbon isotopes and pollen profiles record represents different
aspects of watershed vegetation and are best used in tandem to infer past changes in
watershed vegetation. While this study complements past studies that analyzed only bulk
sediment or pollen and further supports the record of changes in terrestrial vegetation in
the Peten region of the Maya Lowlands, further research is needed to understand the
dynamic changes in forest structure associated with deforestation and reforestation. In
addition, calculating grass and maize pollen accumulation rates from the same core as the
613C may provide a more robust assessment of the C4 pollen signal. And lastly, it is
important to utilize new archaeological information from surrounding watersheds to
better understand how changes in land-use correlate to environmental change.
LIPID EXTRACTION PROCEDURE
33 mL stainless steel cell
2:1 methylene chloride (DCM)/methanol
-All quartz sand should be ashed at 450C for at least two hours in a muffle
furnace. ASE cells should be cleaned after each use by washing with DI water
and solvent rinsing with methanol 3 times. Store ASE cells in oven, or with top
and bottom screwed in place.
1) Screw base onto 33 ml Accelerated Solvent Extractor cell. Place two filters at the
base and add -5-10 mL of sand.
2) Weigh out and add approximately 10-20 tg of an internal standard (C34) to each
sample so as to allow calculation of yields and quantification of lipid compounds
later in the process.
3) Weight out approximately 3-5 g of sample that has been frozen, freeze dried and
4) Record the weight and then add the sample to the cell.
5) Fill remaining space in the cell with sand and screw on the top.
6) Place cells in the Accelerated Solvent Extractor and use the following method to
extract the phospholipids:
Solvent: 2:1 DCM/Methanol
Pressure: 1500 PSI
Temperature: 100 C
Heat: 5 mins
Static: 5 mins
Purge: 200 sec
7) Solvent exchange to hexane using a hot water bath under a stream of nitrogen gas.
This is performed by evaporating samples to dryness and then adding -10 mL of
hexane. To ensure full solvent exchange, repeat this process three times.
8) Transfer residue to a labeled 20 mL glass scintillation vial using ~15ml of hexane.
For maximum compound retention, do this in multiple wash/transfer steps.
9) Evaporate sample in a hot water bath under a stream of nitrogen gas.
10) Samples (in scintillation vials) are now ready for silica gel chromatography.
SILICA GEL CHROMOTAGRAPHY
GC Resolv or Optima grade Hexane
GC Resolv or Optima grade methylene chloride (DCM)
Teflon squirt bottle
Small stainless steel spatula
5% deactivated silica gel
Pasteur pipette with long pipette tips
20 mL beaker
2 graduates cylinders
-All glassware, glass wool, and silica gel should be solvent rinsed and ashed at
450C for at least two hours in a muffle furnace.
1) To make 5% deactivated silica gel, use the following relation:
a. (%deactivation/ (100-%deactivation)) = (mL water/weight of silica (g))
b. (%deactivation/(100-%deactivation)) (weight of silica (g)) = mL of water
Add 1 mL of DI water to 20g of 100% activated silica gel
c. Shake bottle for 10 minutes.
d. Store in dessicator (good for only three days).
2) Assemble stopcock to the base of the column to regulate flow.
3) Set up column by placing glass wool at the base of a glass column and pushing it
down with a glass rod until the column is clogged.
4) Set up column vertically in the hood with a clamp.
5) Add approximately one column full of hexane and allow the hexane to drain to
clean the glass plug.
6) Weight out 2.5 grams of 5% deactivated silica gel in a 20-ml beaker. Cover with
hexane quickly to keep if from absorbing any moisture from the air.
7) Place a glass funnel in the top of the column. Mix silica gel/hexane thoroughly
with the spatula creating a suspension. Add the mixture to the column using the
funnel and allow the silica to settle while tapping gently to avoid gas bubbles. Be
sure to scrape all silica gel from the sides of the beaker into the funnel. Squirt the
gel with hexane to force it through the funnel.
8) Remove the funnel and fill with hexane.
9) Drain excess hexane while tapping the column to guarantee full settling. Close
the stopcock when the hexane level reaches just above the silica gel. Free any
remaining silica gel adhering to the sides by tapping the column gently.
10) Rinse the column with -10ml of hexane to wash remaining silica gel, again
stopping when the solvent level reaches just above the silica-gel. Rinse the
stopcock tip with hexane.
11) Place a collection vial labeled for n-alkanes under the column.
12) Fill a graduated cylinder with 16-17 mL of hexane. Use a pipette to bring the
level to 15 mL, thus rinsing the pipette.
13) Add 1 mL of hexane from the graduated cylinder to the scintillation vial
containing the sample. Rotate the sample vial between fingers while holding the
vial at a 45 angle to dissolve the residue.
14) Transfer the sample in hexane to the column, gently dispensing sample -lcm
above silica gel.
15) Drain and collect the sample and solvent.
16)Repeat steps 13-15 twice more before adding the remaining hexane (-12 mL) in
the graduated cylinder to the column.
17) Drain the sample and solvent to just above the level of the silica gel.
18) Remove the collection vial and a place second vial labeled for the "non-n-alkane
organic fraction" under the column.
19) Using a graduated cylinder, measure 15 mL of DCM and add it to the column.
20) Drain the DCM into the collection vial, thus eluting all other lipid compounds.
21) Archive for potential later use.
22) Concentrate n-alkane fractions in collection vials by evaporating to dryness under
a stream of N2.
23) Re-dissolve n-alkanes in 200 [tL of hexane.
24) Run samples on a Perkin Elmer 8500 Gas Chromatograph (PE 8500 GC) injecting
4 [tL of sample in order to determine purity and approximate concentrations using
the following oven program:
T ramp Hold
Rate Hold Temp. Time
C/min 0C (min)
6 300 20
UREA ADDUCTION AND GC ANALYSIS
GC Resolv or Optima grade hexane
GC Resolv or Optima grade acetone
GC Resolv or Optima grade pentane
GC Resolv or Optima grade urea-saturated methanol
DCM-extracted DI water
GC Resolv or Optima grade methanol
Pasteur pipette with long pipette tips
20 mL beaker
2 graduates cylinders
gas-tight Teflon coated syringes
-All glassware should be solvent rinsed and ashed at 450C for at least two hours
in a muffle furnace. All of the following solvent transfers are performed using
glass pipettes or Teflon-coated gas-tight syringes.
1) After running each alkane sample on the PE 8500 GC, transfer samples to a 4 mL
glass screw top vial and evaporate to dryness under a stream of nitrogen.
2) Add 1 mL each of acetone, pentane, and urea-saturated methanol to each sample.
3) Place in a -4C freezer for -30 minutes.
4) After 30 minutes, remove samples from freezer. Evaporate excess solvent under a
stream of nitrogen. The urea crystals that remain contain the straight-chain n-
alkanes only, while the non-adducts, branched and cyclic compounds, remain
outside of the crystals.
5) Wash urea crystals with hexane three times, with the wash hexane being pipetted
into a separate 4 mL vial labeled as "non-adducts" to be archived for potential
6) Dissolve the remaining urea crystals in 1 mL of methanol and 1 mL of DCM-
extracted DI water. This fraction contains the straight-chain n-alkanes.
7) Add approximately 2 mL of hexane to this fraction. Shake vigorously using a
Vortex vibrator to extract hydrocarbons.
8) Centrifuge vials at 500 rpm for 5 minutes or until hexane and urea/methanol/water
9) Remove the hexane from the top with a gas-tight syringe and place into a separate
4 mL vial labeled as "adducts". Repeat the hydrocarbon extraction (Steps 7-8)
twice more to guarantee full recovery of n-alkanes.
10) Evaporate excess solvent from the adduct fraction under a stream of nitrogen.
11) Repeat steps 2-10 twice more on the adduct fraction.
12) After triple adduction, evaporate the adduct fraction to dryness under a stream of
nitrogen and re-dissolve in 200 UL of hexane.
13) Run samples on a Perkin Elmer 8500 GC using the following oven program and
injecting 4 UL of sample:
Rate Temperature Time
C/min 0C (min)
6 300 20
14) Evaporate the sample to dryness again under a stream of nitrogen and transfer to a
1 mL glass crimp-top autosampler vial using multiple washes to ensure full
transfer of n-alkanes.
15) Evaporate the sample to dryness under a stream of nitrogen. Dissolve in a known
amount of hexane according to obtain the approximate concentration necessary
for GC-IRMS analysis (ex.: samples that had a voltage of -20-30 mV on the PE
8500 GC should be dissolved in 100 UL of hexane whereas samples with a
voltage of -10 should be dissolved in 25 UL of hexane).
16) The adduct fraction containing straight-chain n-alkanes is now ready for GC-
Isotope Ratio Mass Spectrometry (IRMS) analysis.