Terminal Classic drought in the northern Maya lowlands inferred from multiple sediment cores in Lake Chichancanab (Mexico)

Terminal Classic drought in the northern Maya lowlands inferred from multiple sediment cores in Lake Chichancanab (Mexico)

ARTICLE IN PRESS Quaternary Science Reviews 24 (2005) 1413–1427 Terminal Classic drought in the northern Maya lowlands inferred from multiple sedime...

768KB Sizes 0 Downloads 7 Views

ARTICLE IN PRESS

Quaternary Science Reviews 24 (2005) 1413–1427

Terminal Classic drought in the northern Maya lowlands inferred from multiple sediment cores in Lake Chichancanab (Mexico) David A. Hodell, Mark Brenner, Jason H. Curtis Department of Geological Sciences, Land Use and Environmental Change Institute (LUECI), University of Florida, P.O. Box 112120, Gainesville, FL 32611, USA Received 22 June 2004; accepted 22 October 2004

Abstract We present new density records from sediment cores taken along a depth transect in Lake Chichancanab, Mexico. The data reveal in great detail the climatic events that comprised the Terminal Classic Drought (TCD) and coincided with the demise of Classic Maya civilization between ca 750 and 1050 AD. In shallow-water cores, the TCD is marked by a condensed gypsum horizon. In deeper-water sections it is expanded and represented by numerous interbedded gypsum and organic-rich strata. The TCD was not a single, two-century-long megadrought, but rather consisted of a series of dry events separated by intervening periods of relatively moister conditions. We estimate the TCD occurred between ca 770 and 1100 AD and included an early phase (ca 770–870 AD) and late phase (ca 920–1100 AD). The intervening 50-year period (ca 870–920 AD) was relatively moister. Each dry phase is represented by multiple gypsum (density) bands interbedded with organic-rich sediment that indicate alternating dry and wet conditions. Spectral analysis revealed significant periods around 213, 50, and 27 years. Despite uncertainty regarding the ages of these events, their pattern is robust and generally consistent with other proxy records under the same climate regime, such as the marine Cariaco Basin off northern Venezuela (Science 299 (2003) 1731). r 2004 Elsevier Ltd. All rights reserved.

1. Introduction Lake Chichancanab is a long, narrow basin located in the center of the northern Yucatan Peninsula, Mexico, near the border between the States of Yucatan and Quintana Roo (Figs. 1 and 2). Its sediments contain a rich archive of Holocene paleoclimate information for the northern Maya lowlands (Covich, 1970; Covich and Stuiver, 1974; Hodell et al., 1995, 2001). Of all the lake sediment profiles studied thus far on the Yucatan Peninsula, those from Lake Chichancanab have provided the clearest record of droughts during the period of Maya occupation. The characteristics that make this lake system ideal for paleoclimate studies include its effectively closed hydrology (i.e., loss of a substantial fraction of its water volume to evaporation), gypsum Corresponding author. Tel: +1 352 392 6137: fax: +1 352 392 9294.

E-mail address: [email protected]fl.edu (D.A. Hodell). 0277-3791/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.quascirev.2004.10.013

saturation (CaSO4), high carbonate microfossil abundance, and relatively low-density prehistoric occupation of its watershed. In particular, the dual paleoclimate proxies of mineralogy (gypsum deposition) and oxygen isotopes (d18O) provide a powerful tool for inferring past drought history. In 1993, a 4.9-m core was raised in 6.9 m of water and yielded a 9000-year sediment record with a mean sedimentation rate of 0.5 mm/yr. This core provided the first paleolimnological evidence for a protracted drought during the Terminal Classic Period, ca 800–1000 AD, which included a 6-cm thick gypsum bed 65 cm below the sediment surface and a concomitant increase in d18O values of gastropods and ostracods (Hodell et al., 1995). A single seed extracted from the interval of gypsum deposition and peak d18O values was dated by AMS-14C. Radiocarbon analysis yielded an age of 1140735 14C years BP, which translated to a calibrated age range of 780–990 AD

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1414

22˚ N

21˚ N

Punta Laguna

Chichancanab

20˚ N

19˚ N

18˚ N

92˚ W

91˚ W

90˚ W

89˚ W

88˚ W

87˚ W

Fig. 1. Location of Lakes Chichancanab and Punta Laguna on the Yucatan Peninsula relative to contours of annual rainfall in mm.

(95.4% probability). Gypsum precipitation during the Terminal Classic Period suggested a period of drought in the north-central Maya lowlands. In 2000, we retrieved cores from 11 m of water in Chichancanab that permitted us to refine our paleoclimate interpretations (Hodell et al., 2001). Two cores from the same location were combined to yield a composite 1.9-m section that spanned the last 2600 yr. Twelve radiocarbon dates on terrestrial organic matter provided a robust chronology and sedimentation rates averaged 0.8 mm/yr. Several intervals of gypsum deposition were identified representing relatively dry climate conditions: at ca 475 BC, 275–250 BC, 125–210 AD, 750–875 AD, and 1000–1075 AD (Hodell et al., 2001). The event at 125-210 AD was associated with the abandonment of the Mirador Basin in northern Peten, Guatemala (Dahlin, 1983). The latter two intervals coincided with the Terminal Classic Period and suggested that the drought occurred in two distinct phases from about 750–875 AD and 1000–1075 AD. Spectral analysis of the density record showed quasi-periodicities at 208, 100, 50, and 38 years. In March 2004, we returned to Lake Chichancanab and completed the first bathymetric survey of the lake. We also took sediment cores along a transect ranging in water depth from 4.3 to 14.7 m (relative to 2004 water levels), near the deepest point in the lake. Multiple cores were taken at the deepest site to ensure stratigraphic continuity and to evaluate disturbance of gypsum layers during the coring process. Sedimentation rates in the deepest cores average 0.9–1 mm/yr. These new records provide a detailed history of paleoclimate change on the

Yucatan Peninsula during the period of Maya occupation, including the events comprising the Terminal Classic Drought (TCD). This fine-grained paleoclimatic history will help archaeologists evaluate the role that climate change may have played in Maya cultural evolution.

2. Physical and chemical environment Lake Chichancanab is an elongate, N-S oriented basin that is one in a series of lakes separated by lowlying swampy areas (Fig. 2). The lake is 14.5 km long, 0.7 km wide, and had a maximum depth of 15 m in March 2004 (Table 1). We constructed a bathymetric map of the basin using a Garmin GPSMAP 180 Sounder interfaced with a laptop computer. The lake is comprised of a series of sub-basins separated by shallow sills. Lake Esmeralda, its sister lake, lies to the south. The two water bodies have been connected at times of high water. The central and largest basin of Lake Chichancanab lies between 191 520 and 191 540 N (Fig. 1), and has been the focus of our coring efforts since 1993. The lake basin owes its origin to tectonokarst processes related to faulting associated with the Serrita de Ticul to the west, and subsequent dissolution and collapse of limestone bedrock. The eastern shore is bounded by a low ridge that probably represents a fault line that extends southeast from Chichancanab to the town of Bacalar near the coast (Perry et al., 2003). The aquifer on the Yucatan Peninsula consists of a lens of fresh water that floats above denser saline water.

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

Fig. 2. Bathymetry of Lake Chichancanab showing the position of the depth transect of cores taken in 2004 (white squares). Inset map shows that Lake Chichancanab is part of a chain of narrow lakes that are connected during high water levels.

Table 1 Morphometric variables of Lake Chichancanab Maximum length ¼ 14.5 km Maximum width ¼ 0.7 km Perimeter ¼ 33.6 km Area ¼ 5.3 km2 Volume ¼ 22.03  106 m3 Max depth ¼ 15 m Mean depth ¼ volume/area ¼ 4.2 m

The thickness of the fresh water lens increases from 15 m near the coast to approximately 120 m at a distance of 90 km inland (Marin, 1990; Stenich and Marin, 1997). The salt water wedge penetrates 440 km inland (Back and Hanshaw, 1970; Doehring and Butler, 1974; among others). Perry et al. (2002) estimated the elevation of Lake Esmeralda at 4 masl in 1997, and

1415

suggested it occupies a groundwater divide where the water table elevation decreases both east and west of the lake. They also noted that Lake Chichancanab lies close to this hydrologic divide and marks the boundary between the high-permeability rocks of the northwest Yucatan and the less permeable southeastern ‘‘evaporite region’’. This region is defined by high SO4/Cl ratios and lower permeability, the latter expressed as swamps perched above the water table, ephemeral streams, and relatively slow return to normal lake stage after storm events (Perry et al., 2002). The water table in this region is controlled less directly by sea level than it is in other parts of the northern peninsula. Chichancanab in Yucatec Maya means ‘‘little sea’’, which aptly describes its high concentration of dissolved salts (4000 mg/L). The dominant ions are sulfate and calcium (Table 2) and the lake water is saturated with both gypsum (CaSO4) and celestite (SrSO4), as well as calcite, aragonite, and dolomite (Perry et al., 2002). Sulfate comes from the dissolution of bedrock gypsum/ anhydrite in the watershed, either Eocene gypsumbearing evaporites or gypsum-anhydrite from the Chicxulub impact breccia, brought close to the surface by faulting (Perry et al., 2003). The pH of lake water averages 8.5 and sediments are rich in carbonate microfossils. The two factors that contribute to the lake’s high salinity are the high solubility of gypsum/anhydrite in the watershed and intense evaporation of water from the basin. High evaporation rates are indicated by high d18O and dD values relative to rainfall and groundwater. Covich and Stuiver (1974) reported a range of d18O values from 3.4% to 5.4% for lake waters collected in 1973. Perry et al. (2003) reported d18O and dD values of 3.0 and 15.4, respectively, in 1998. We measured mean d18O values in lake water of 3.2% in 1993, 4.0% in 2000, and 3.1% in 2004. The slightly lower d18O values obtained in 2004 coincided with high lake stage reflecting the slow recovery of lake level after Hurricane Isidore struck the peninsula in September 2002, raising the water table throughout northern Yucatan dramatically. Isotopes of regional precipitation and groundwater are similar to one another and average about 4% (d18O) and 22% (dD) (Perry et al., 2003). This indicates that Chichancanab lake water is enriched 7% to 8% in d18O and enriched 37% in dD relative to its hydrologic input waters. Chichancanab water also deviates from the meteoric water line, indicating evaporative enrichment (Fig. 3). Hydrogen isotopes are about five times more enriched than oxygen isotopes. Chichancanab’s sister Lake Esmeralda is less evaporatively enriched in both oxygen and hydrogen isotopes (d18O ¼ 2.4%, dD ¼ 12.7% in 1998 and d18O ¼ 1.8%, dD ¼ 10.7% in 1997; Perry et al., 2003), indicating it is a more open system hydrologically.

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1416

Table 2 Concentration of dissolved ions in Lake Chichancanab

Ca Mg Sr K Na SO4 Cl HCO3 Total

March 1996a mg/L

March 1995a mg/L

June 1993b mg/L

1967c mg/L

1950d mg/L

607 201 15 11 199 2455 234 138 3861

679 240 17 4 210 2625 293 95 4164

693 257 — 14 250 2545 234 18 4011

600–700 360–425 — 20–24 340–400 2750–2950 — — —

813 253 — 3 105 2959 138 71 4342

a

Perry et al. (2002). Hodell et al. (1995). c Covich and Stuiver (1974). d Illescas-Pasquel (1950). b

40

δ D = 8 * δ 18O + 10 Chichancanab 1998

δ D (‰ SMOW)

20 0

Esmeralda 1997

-20

Esmeralda 1998 Dziuche

-40

-60 -80

Rain, Hurricane Mitch 1998

-10

-8

-6

-4

-2

0

2

4

δ O (‰ SMOW) 18

Fig. 3. Oxygen and hydrogen isotopes of rain, groundwater, and lakes from the northern Yucatan Peninsula. Solid line represents the global ‘‘Meteoric Water Line’’ (MWL). Note that Lake Chichancanab water plots below the MWL indicating significant evaporative loss of water from the basin. Dashed lines represent the average values of dD and d18O for groundwater. Data and graph redrawn from Perry et al. (2003).

3. Climate setting Rainfall on the Yucatan Peninsula is highly variable both spatially and temporally. The northwest coast is the driest part of the peninsula and precipitation increases to the south (Fig. 1). There is also a pronounced east–west rainfall gradient across the northern peninsula. The east coast receives 1400 mm/ yr, and precipitation decreases to the west, with a minimum of 1000 mm/yr in the Puuc Hills. This

gradient is the result of prevailing easterly (i.e., trade) winds. Precipitation arises from convection of humid air along the east coast and progressively rains out as storms move from east to west across the peninsula (Giddings and Soto, 2003). Yucatan lies on a main hurricane path and a single storm can contribute significantly to the mean rainfall in a given year (Gray, 1993; Boose et al., 2003). The wet season lasts from May to October with peak precipitation usually occurring during the month of September. The dry season occurs from November through April. Summer precipitation is usually bimodal with July and August typically being a little drier and less cloudy than other months in the rainy season, resulting in the so-called canicula or ‘‘little dry’’ (Magan˜a et al., 1999). The rainy season coincides with Northern Hemisphere summer when the ITCZ and North Atlantic high move northward (Hastenrath, 1984, 1991). During this period, tropical low pressure systems are carried west by the trade winds across the Atlantic into the Caribbean, bringing heavy rainfall, lightning, and strong winds to the Yucatan Peninsula (Wilson, 1980). Precipitation is suppressed during winter months when the ITCZ swings south of the equator and the North Atlantic subtropical high pressure zone (also known as the Azores–Bermuda high) moves south and dominates in the Intra-Americas Sea (Gray 1993). In winter, the northern Yucatan Peninsula is marked by subsidence related to the descending limb of the Hadley cell, which is centered at 201 N (NCEP/NCAR Reanalysis fields 1968–1996; Waliser et al., 1999). The closest weather data collected near Chichancanab come from La Presumida located SW of the lake (Fig. 2). Mean annual precipitation averages 1268 mm and evaporation exceeds precipitation. The net annual water deficit is 300–400 mm/yr. Mean annual temperature is 25.9 1C and mean monthly temperature varies from a low of 23.3 1C to a high of 28.4 1C.

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1417

4. Methods

5. Results

Sediment cores were collected from Lake Chichancanab in 1993, 2000, and 2004. On 7 and 8 March 2004, we took cores along a transect of water depth ranging from 4.3 to 14.7 m of water. Sediments were retrieved with a piston corer using 5.7-cm (OD) clear polycarbonate tubes. Overlapping core sections were retrieved from each site to ensure complete stratigraphic recovery. Multiple cores were taken at the deepest station (CH1) to duplicate the topmost 175-cm of the section and evaluate any disturbance of fine interbedded layers of gypsum and shell-bearing organic matter. Each core section was measured for bulk density by gamma-ray attenuation (GRA) at 0.5-cm increments using a GEOTEK MultiSensor Core Logger. The instrument was calibrated at the start of each day using a 5.7-cm (OD) polycarbonate core liner filled with distilled water and an aluminum standard of varying thickness. After logging, cores were split and imaged using a GEOTEK digital color linescan camera that was calibrated each day using a white ceramic calibration tile. As the core is pushed along the track by a stepper motor, the camera collects line scans across the core surface that are broken into three colors (red, green, and blue) using interference filters mounted in front of each of three CCD line arrays.

Several features of the density signal can be correlated across cores taken from different water depths (Fig. 4). Groups of density peaks (labeled A–D), corresponding to intervals of gypsum deposition, were identified. Each group is comprised of several dense gypsum horizons interbedded with low-density organic-rich deposits. Gypsum layers are compressed together in shallow sections, but are distinct in cores from deeper water. For example, groups A and B appear as a single, 6-cm thick layer in cores from 7 m of water. Cores from the deepest station (CH1 07-III-04) at 14.5 m display the highest sedimentation rate. They contain the most highly resolved record of gypsum deposition. Consequently, deepwater cores were used as the reference section to which all other cores were correlated by matching density features. During the coring process, the act of punching through dense gypsum layers interbedded with organic-rich sediment can distort the stratigraphy by forcing gypsum fragments downward through softer, underlying deposits. We assessed such disturbance by comparing records from multiple cores taken at the same location. It is unlikely that coring in multiple holes will produce the same distortions. We took three full cores (0 to 175 cm) and a deeper section (75–175 cm) at the site in 14.5 m of water. Cores were taken on two days and the distance between core sites was within the accuracy of the GPS (10–20 m).

Density (g/cm3 ) 1

1.2

1.4

1.6

1.8

1

0

1.2

1.4

1.6

1

1.2

1.4

1.6

1

1.8

0

0

CH1 7-III-04 (14.7 m)

1.2

1.4

1.6

1.8

0

CH 23-V-00 (11 m)

CH 21-V-00 (12 m)

CH3 7-III-04 (7 m) CH 25-VI-93 (6.9 m)

50 50

370

50 100

Depth (cm)

50 890

100

870

A 1130

150

860

B B

C 150 200

A

A 100

B

1410 1400 1490

C

990

1350

1140

A&B 1350

D

D D 200

150

D

1800 1770 1830

100 1840

250

2310

300

250 0

5

10

15

Weight %S Fig. 4. Gamma ray attenuation (GRA) bulk density records for a transect of cores taken in Lake Chichancanab in water depths ranging from 7 to 14.7 m. Weight percent sulfur record from the 1993 core taken in 6.9 m of water (dashed line) is shown relative to the density record from the 2004 core taken in about the same water depth. Position of radiocarbon control points in 14C years before present is indicated in each core. Letters (A–D) designate groups of density peaks that are correlative among cores.

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1418

We were especially interested in the interval between 120 and 170 cm in the deep-water cores because they display strong gypsum/organic bands that were deposited during the Late and Terminal Classic Periods when the Maya civilization reached its apex and ultimate

demise. Digital images of this time interval from the four sections are very similar, showing a series of lightcolored gypsum layers interbedded with organic-rich sediments containing shells (Fig. 5). Shells are predominantly gastropods (Pyrgophorus coronatus) and appear in the images as white specks against the dark organic-rich sediment. The digital images were analyzed for their red, green, and blue components and the red/blue ratio (R/B) was used to assess the similarities among the split core images (Fig. 6). Three groups of light-dark alternating beds are identified (designated A, B, and C), and are separated from each other by intervals of uniformly dark, organic-rich sediment. Within each group from a core, individual peaks can be recognized and are designated A1-6, B1-2, and C1-2. Density records among the cores are also very similar, though subtle differences exist (Fig. 7). For example, the two cores taken on 7 March 2004 show a clear interval of low-density between groups A and B whereas cores taken on 8 March 2004 contain small density peaks between groups A and B. Nonetheless, images and color data for all four cores clearly indicate deposition of organic-rich sediments between groups A and B. A radiocarbon date was obtained from deep-water core CH1 08-III-04-MWI-2 on wood found at a depth of 141–142 cm. The age was 1130735 14C years BP, equivalent to 780–1000 AD using OxCal version 3.5 and the atmospheric data from Stuiver et al. (1998). Because

Fig. 5. Images of four split cores taken at the same station in 14.7 m of water. Sediments are composed of interbedded gypsum and organicrich strata containing abundant shell material. The similarity of the cores indicates that sediments were not disturbed during the coring and splitting process. Arrow in Core CH1 08-III-04-MWI-2 designates position of radiocarbon date (1130735 14C years BP; 780–1000 AD).

Red/Blue reflectance 100

CH1 7-III-04-MWI-1

CH1 8-III-04 MWI-2

CH1 7-III-04-1A

CH1 8-III-04 MWI-1

110

120 6 5

Depth (cm)

130

A

4 3 2

140 1

150

2

B 1

160 2

C

1

170

180

Fig. 6. Ratio of red to blue (R/B) color reflectance of core images shown in Fig. 5. Letters designate groups of reflectance peaks and numbers refer to individual peaks that can be correlated among the four cores. Arrow in Core CH1 08-III-04-MWI-2 designates position of radiocarbon date (1130735 14C years BP; 780–1000 AD).

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

CH1 7-III-04-MWI-1 Density (g/cm3) 1

1.2

1.4

1.6

1.8

1419

CH1 8-III-04-MWI-2 Density (g/cm3) 2

2.2

0.8

1

1.2

1.4

1.6

1.8

2

120

6

130

5 4

A

3

Depth (cm)

140

2 1

150

2

2

B

1

160

1

C

170

0.8

1

1.2

1.4

1.6

1.8

Density (g/cm3) CH1 7-III-04-1A

0.8

1

1.2

1.4

1.6

1.8

Density (g/cm3) CH1 8-III-04-MWI-1

Fig. 7. GRA bulk density records of the four cores taken in 14.7 m of water. Letters designate groups of density peaks and numbers refer to individual peaks that can be correlated among the four cores.

the occurrence of terrestrial material is rare in the deepwater cores, additional age control points were obtained by correlating density peaks from cores CH-23-V-00 and CH-21-V-00 (Figs. 4 and 8). The sediment surface in core CH1 08-III-04 was assumed to be 2004 AD. The composite reference section was dated by converting radiocarbon dates to calendar years, and fitting the data with a second-order polynomial equation. The fitted line passes through all points within the 95.4% probability distributions for each calendar date (Fig. 9). The second-order coefficient is small and is only required because the linear sedimentation rate increases toward the top of the core owing to less compaction of surface sediment. Between 120 and 260 cm, sedimentation rates were fairly constant, averaging 0.87 mm yr1 (Fig. 9). Density versus age is described only from the deepwater core at station CH1 because it is the most detailed and complete of the recovered sections (Fig. 10). The base of the core possesses a series of thin, high-density gypsum bands interbedded with low-density organic layers that were deposited between 800 and 250 BC (286 and 241 cm). This was followed by deposition of shell-bearing organic matter with low density between 250 BC and 90 AD (241 and 217 cm). The interval from 90 to 230 AD (217–203 cm) contains two density peaks followed by a long period of deposition of lowdensity organic matter between 230 and 650 AD (203 and 169 cm).

The density and color records between 600 and 1100 AD are expanded to illustrate detailed changes during the Late and Terminal Classic Periods (Fig. 10). A thin, high-density gypsum bed occurs at 670 AD (between 169 and 165 cm) followed by continued organic matter deposition to 770 AD (158 cm). Two thick gypsum horizons occur between 770 and 870 AD (158 and 149 cm), followed by a short period of low density from 870 to 920 AD (149–143 cm). Between 920 and 1100 AD (143–124 cm), four large density peaks are followed by two smaller ones. From 1100 to present (124–0 cm), sediments are dominated by shell-bearing organic matter with low density. Time series analysis was performed in both the depth and time domains for all four density records from the deep-water site (CH1) between 120 and 170 cm (Fig. 11A). All spectra show three peaks in the depth domain but the frequencies vary among the records, perhaps as a consequence of coring effects but more likely due to small differences in sedimentation rates. A mean spectrum was calculated by averaging the power from the four records for each frequency interval (i.e., 0.001). The mean spectrum in depth domain shows significant power at periods of approximately 21, 5, and 2.6 cm (Fig. 11A). Spectra were also calculated in the time domain after applying the 2nd order age-depth equation. The averaged spectrum for all four density records shows significant peaks at 213, 50, and 27 years (Fig. 11B).

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

Density (g/cm3) 1

1.2

1.4

1.6

0

1.8

0

Depth (cm)

50

50

sed. rate = 0.87 +/- 0.08 mm yr

120

Depth (cm)

1420

160 200 240

100

280 800

1600

2400

Age (cal yr BP)

150

100

Depth (cm)

860

200

890

150

870 990 1130 1350 1350 1410 1400 1490

200

250

1800 1700 1830 1840

2310

250 0

500

1000

1500

2000

2500

3000

Age (cal yr BP) Fig. 9. Age-depth relationship for Core CH1 7-III-04. Line represents a 2nd order polynomial fit of the data (Age ¼ 14.926+5.2572 * depth+0.016909 * depth2). Error bars are the 95.4% probability range for each radiocarbon date converted to calendar years using the program OxCal version 3.5 (Bronk Ramsey, 1998). Inset shows the age control points for the interval from 120 to 260 cm, fit with a linear regression line. Sedimentation rates for this interval average 0.8770.08 mm/yr.

300 Fig. 8. GRA bulk density record from Core CH1 7-III-04 with the position of the radiocarbon dates that were assigned to various depths by correlation of density signals shown in Fig. 4.

6. Discussion 6.1. Chichancanab sediments as recorders of climate change Several attributes make the sediments of Lake Chichancanab excellent archives for paleoclimate reconstruction. First, oxygen and hydrogen isotopes of lake water indicate that it is a relatively closed basin that loses a large fraction of its water budget to evaporation (Fig. 3). Second, precipitation of gypsum from Lake Chichancanab’s water provides a sensitive proxy for past changes in hydrology that can be measured by nondestructive means (i.e., density) at high spatial resolution. Third, the Chichancanab basin was not densely occupied during the Classic Period as were most watersheds of the southern lowlands. High salinity and sulfate concentrations probably discouraged use of the lake as a potable water source. The total dissolved solids and sulfate concentration (Table 2) exceed recommended drinking water standards. Sulfate has a laxative effect and the US Public Health Service recommends

that sulfate in drinking water should not exceed 250 mg/ L. Chichancanab’s sulfate concentration is 10 times this value. Pollen analysis of Chichancanab sediments shows some evidence for disturbance vegetation and maize cultivation in the watershed during the Classic Period (Leyden, 2002); however, Lake Chichancanab’s hydrology was probably not impacted significantly by human modification of the relatively low-stature vegetation in the watershed (Rosenmeier et al., 2002a, b, in press). In closed basins, the volume and d18O of the lake water is controlled by the balance between water lost to evaporation and water gained by precipitation, runoff, and groundwater input. Runoff and groundwater inflow are generally related to precipitation, so the hydrologic budget of a closed-basin lake is essentially dependent on the difference between precipitation and evaporation. Lake Chichancanab may not be completely closed because water lost via downward leakage can be significant, even in karst basins with clay seals (Deevey, 1988). Nonetheless, the high d18O and dD values of Chichancanab lake water and the relatively slow recovery of stage and lakewater isotope values following precipitation events such as hurricanes indicate the majority of its hydrologic budget is lost to evaporation. Lake Chichancanab water is just above saturation for gypsum today (average saturation index ¼ 0.05; Perry et al., 2002). Gypsum precipitates near the shoreline

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

Density (g/cm3) 1

1.2

1.4

1.6

1421

Density (g/cm3) 1 1100

1.8

2000

1.2

1.4

1.6

1.8

4

1500

3

1000

A 2

1

Date (AD)

500

BC / AD

Date

1000 900

2

B 800

1

0

700 500 C

1000

600 1.2

1.6

2.0

2.4

Red/Blue reflectance Fig. 10. GRA bulk density record from Core CH1 7-III-04 versus time based on the age model in Fig. 9. Gray area is an expanded plot of GRA bulk density and red/blue color reflectance for the interval from 600 to 1100 AD.

where evaporation is intense, but remains in solution in the open lake because minerals often do not precipitate unless the water is supersaturated. When the lake volume is reduced by decreased rainfall and/or increased evaporation, the saturation of gypsum increases until it precipitates from solution and is deposited on the lake floor. Occurrence of gypsum layers in sediment cores therefore provides a proxy for past increases in E/P. Sediments containing gypsum are denser than shellbearing organic matter, thus sediment bulk density is highly correlated with gypsum deposition and weight percent sulfur. Sediment density is a better proxy for gypsum than is weight percent sulfur (S wt%) because density can be measured non-destructively by gamma ray attenuation at high spatial resolution (e.g., every 0.5 cm) while the sediment is still in polycarbonate core tubes. This avoids sediment disturbance that can result during core splitting or extrusion processes, as well as inherent difficulties of sampling dense gypsum beds at fine resolution for S analysis. Sediment color also can serve as a proxy for gypsum because it is lighter than organic matter (Fig. 5). Imaging, however, is done following the core splitting process, which can result in disturbance of fine gypsum-organic interbeds.

6.2. Inferred climate history The climate near Lake Chichancanab has been marked by alternating wet–dry periods for the past 2800 years. Gypsum was deposited during the following broad periods: 800–250 BC, 90–230 AD, and 650–1100 AD. This does not imply that climate was always dry during these periods, but rather that droughts were more frequent or protracted during these intervals. The latter period is of great interest because it includes the Late Classic Period (600–800 AD) when the Maya reached their cultural apex and the Terminal Classic Period (800–1000 AD) when they suffered a dramatic demographic decline. A brief drought event occurred at 670 AD represented by a thin (0.5 cm) gypsum band (Fig. 10). This was followed by a relatively wet period lasting for 100 years from ca 670 to 770 AD. The next century included the first phase of the TCD which is marked by at least two prominent gypsum layers occurring between 770 and 870 AD. A brief 50 year period of relatively moister conditions occurred between 870 and 920 AD, which we refer to as the Terminal Classic moist period. The second phase of the TCD lasted from 920 to 1100 AD and is marked by at

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1422 5

6

21 cm

5 4

Power

4 3

3

4.8 cm

2

2

1

2.6 cm 0

1

0

0.2

0.4

0.6

0.8

1

0 0

0.2

(A)

0.4

0.6

0.8

1

Frequency 60

60 50

213 yr

50

2001) and 2004. Comparison of the density record with a bandpass filter of 50 years indicates that the first phase of drought (ca 770–870 AD) was comprised of two 50-year cycles (Fig. 12). The density signal then skipped a cycle (ca 870–920 AD), hence the name Terminal Classic moist period. The second drought phase was marked by about four 50-year cycles between 920 and 1100 AD. The density signal suggests that drought was a recurrent phenomenon (Fig. 10; Hodell et al. 2001). This might be expected given the steep north–south precipitation gradient on the Yucatan Peninsula (Fig. 1). Sporadic, severe droughts have been simulated for the peninsula using a coupled global climate model (Hunt and Elliott, in press). In historical precipitation records of Yucatan, rainfall deficits are prominent whereas excesses are much less evident. Analysis of historical precipitation in Yucatan revealed distinct periodicity in the range of 2–24 years, but the cause of the cyclicity is unknown (Giddings and Soto, 2003).

40

Power

40

6.3. Chronology of the TCD

30 20

30

50 yr 10

20

26.7 yr

0 0

0.02

0.04

0.06

0.08

0.1

10

It is important to understand the uncertainties associated with the dating of gypsum deposition and inferred drought events because comparisons will be drawn between paleoclimate and archaeological data. Ages for climate events in Chichancanab are based on assignment of 15 radiocarbon dates from two cores to

0 0.02

0.04

0.06

0.08

0.1

Density (g/cm3)

Frequency

Fig. 11. Time series analysis of GRA bulk density in Core CH1 7-III04 between 120 and 170 cm in depth (A) and time (B) domains. Power spectra were calculated using the program Analyseries (Paillard et al., 1996) by interpolating the signals at constant depth and time increments and linear detrending. All estimates were made with a Bartlett window, 13 lag; and a prewhitening constant of 0.5. Inserts show the power spectra for all four cores which were then averaged at each frequency step to produce a mean spectrum. In depth domain, significant peaks occur at 21, 4.8 and 2.6 cm. In time domain, peaks are centered at 213, 50, and 26.7 yr.

least four prominent gypsum horizons and two thin gypsum beds. Spectral analysis of the density signal in the interval between 600 and 1200 AD suggests that drought was recurrent during this interval with periods of 213, 50, and 27 years (Fig. 11). These periods may represent harmonics and are similar to those recognized previously in other Chichancanab sediment cores. For example, the density signal in CH-23-V-00 revealed power concentrated at periods of 208, 100, 50, and 39 years (Hodell et al., 2001). The 213-year period is close to the 208-year period and the 50-year cycle was recognized in cores taken in both 2000 (Hodell et al.,

1

1.2

1.4

1.6

1200

1100

Date (AD)

0

(B)

1000

900

800

700

600 Fig. 12. GRA bulk density record from Core CH1 7-III-04 (black line) shown relative to a bandpass filter (dashed gray line) of the signal centered at 50 years.

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

levels in the deep-water reference section that contains one date. This is justified given the scarcity of terrestrial organic material found in Chichancanab cores and the great similarity of density signals among cores. For each calibrated radiocarbon date, the midpoint of the 95% probability distribution was used and the age-depth data were fit with a 2nd-order polynomial equation. According to this age/depth model, we estimate that the TCD occurred between 770 and 1100 AD. Ages are not significantly different if we use a linear model for the interval between 120 and 260 cm instead of the 2ndorder equation (Fig. 9). Dating uncertainty associated with the TCD can be assessed by considering the radiocarbon results of terrestrial materials from below, within, and above the intervals of gypsum deposition in the various Chichancanab cores. The onset of the TCD is constrained by analysis of two carbonized twigs from 105 and 106 cm just below the base of the lower gypsum unit in Core CH-23-V-00-MWI. The two samples gave identical ages of 1350750 14C years BP (Fig. 13). These dates convert to 690 AD 790 years at 95.4% probability, which falls within the late Classic Period (600–800 AD). The data indicate that the TCD started after 690 AD 790 years. Two terrestrial carbon samples were dated from within the interval of gypsum deposition in two cores. A seed was dated at 65 cm in Core CH-25-VI-93-MWI (water depth ¼ 6.9 m) and gave an age of 1140 735 14C years (Hodell et al., 1995). In Core CH1 08-III-04-MWI-2 (water depth ¼ 14.5 m), charcoal was dated from 141–142 cm and yielded an age of 1130 735 14C years. The two dates from within the zone of gypsum deposition are identical within error and convert to 890 AD 7110 years and 885 AD 7105 years at 95.4% probability. The age range of these dates (790–1000 AD) falls exactly within the Terminal Classic Period (800–1000 AD). A date of 990760 14C years BP was obtained at a depth of 87 cm in Core CH-23-V-00-MWI, near the top of the second phase of gypsum deposition. This date converts to 1080 AD 7130 and occurs close to

Fig. 13. Calibrated age distributions for radiocarbon dates that constrain the timing of gypsum precipitation in Chichancanab cores. Position of dated samples relative to gypsum layers is shown in right column. Ages are in mean calendar years AD795.4% probability range. Figure produced using the program OxCal version 3.5 (Bronk Ramsey, 1998).

1423

the end of the TCD. At 81 cm in the same core, above the upper gypsum horizon, a date of 870 750 was obtained, indicating that the TCD was over by 1150 AD 7120 years. Errors associated with individual calibrated radiocarbon dates that constrain the TCD are on the order of 790 to 130 years (Fig. 13). This limits our ability to achieve any better than general temporal correlations between paleoclimate and cultural events. Although the ages of the drought events and intervening wet periods are uncertain, the pattern and frequency of drought events are robust. Given the errors inherent in radiocarbon dating and calibration to calendar years, the chronology cannot be improved significantly using 14C alone. Efforts are under way to date the gypsum horizons directly using uranium decay series (U/Th). If this method is not confounded by detrital Th and gypsum behaves as a closed system, then it should be possible to refine the dates of individual drought events. 6.4. Correlation with other paleoclimate records Curtis et al. (1996) reported a high-resolution oxygen isotope record from Lake Punta Laguna, located 150 km northeast of Chichancanab (Fig. 1). Comparison of the Chichancanab reference density record with the d18O record of the ostracod Cytheridella ilosvayi from Punta Laguna reveals excellent agreement. The Punta Laguna d18O record shows two increases in d18O during the terminal Classic Period, from ca 750 to 850 AD and ca 910 to 990 AD, supporting a two-phase subdivision of the TCD with an intervening period (ca 850–910 AD) of relatively moister conditions (Fig. 14). The two dry periods correspond with the deposition of gypsum units A and B in the deep-water reference section from Lake Chichancanab. The end of the TCD in Punta Laguna is marked by decreasing d18O values from ca 1060 to 1100, and coincides approximately with the cessation of gypsum precipitation in Chichancanab. The consistency of the Punta Laguna and Chichancanab records lends support to inferences about both the pattern and timing of the TCD described here for the north-central Maya lowlands. Haug et al. (2003) reported a seasonally resolved marine sediment record of titanium from the Cariaco Basin that indicated the Terminal Classic Period was marked by an extended regional dry period, punctuated by more intense multi-year droughts centered at 760, 810, 860, and 910 AD (Fig. 15). Not counting the duration of the droughts themselves, the spacing between drought events was about 40–47 years (75). We observed a similar period of 50 years in density records from Lake Chichancanab (Figs. 11 and 12) (Hodell et al., 2001). The general pattern of dry climate conditions in the Terminal Classic Period is consistent between Lake Chichancanab and the Cariaco Basin,

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

1424

δ18O Cytheridella ilosvayi (‰, PDB) -1.5

-1

-0.5

0

0.5

1

1.5

+ A

terminal

1000

900

B

800

Classic

Date (AD)

early

1100

Postclassic

1200

late

700 C

+

early

600

500 1

1.2

1.4

1.6

1.8

Density (g/cm3)

Fig. 14. Comparison of the GRA bulk density record from Chichancanab Core CH1 7-III-04 (solid line) and the 5-pt running mean of the d18O record (solid line with open circles) of the ostracod Cytheridella ilosvayi from Punta Laguna Core PL 23-VI-93 (Curtis et al., 1996). Crosses represent the positions of 14C dates in the Punta Laguna core.

although the records differ in detail. The estimated start of the TCD (770–780 AD) is similar between the Chichancanab and Cariaco records, but the inferred drought lasted until 1100 AD in Chichancanab whereas it ended by 1000 AD according to the Cariaco Ti record. The Chichancanab record suggests the TCD occurred in two phases, from ca 770 to 870 AD, and from ca 920 to 1100 AD, whereas the Cariaco Basin record indicates extremely dry events at ca 760, 810, 860, and 910 AD (Fig. 15). Although the Cariaco chronology is based on varve counting, the absolute dates for the drought events ‘‘float’’ in time because the chronology is referenced to an assumed date of 930 AD for a rise in Ti that marks the local onset of Medieval Warm conditions. Nonetheless, the pattern and relative spacing of events in the Cariaco Ti signal are robust. The Cariaco Basin is located at 101N whereas Lake Chichancanab is at 201N. Both are part of the same Caribbean climate regime where precipitation is related to the annual cycle. The seasonal rainfall pattern is related to shifts in the position of the Intertropical Convergence Zone (ITCZ) and North Atlantic (Azores–Bermuda) subtropical high-pressure system. Interannual variability in Caribbean rainfall is controlled by a mechanism similar to that of the annual cycle, and involves a competition between the North Atlantic high pressure system and the eastern Pacific ITCZ (Giannini

Fig. 15. Comparison of the GRA bulk density record from Chichancanab Core CH1 7-III-04 (solid line with crosses) and the 30-pt running mean of relative titanium concentration (solid line) from Ocean Drilling Program Hole 1002D, in the Cariaco Basin (Haug et al., 2003). Asterisks designate minima in the Ti record at 760, 810, 860, and 910 AD inferred to represent severe drought events in the terminal Classic Period.

et al., 2000, 2001a, b). The general similarity of the temporal pattern and frequency of inferred drought between Lake Chichancanab and the Cariaco Basin suggests that climate changes during the TCD were regionally pervasive and related to large-scale departures in oceanic and atmospheric fields. 6.5. Implications for Maya archaeology Physical evidence of the TCD from Lake Chichancanab sediments has generated considerable debate among Maya archaeologists. Popular books such as ‘‘The Great Maya Droughts’’ by Gill (2000) and ‘‘The Fall of the Ancient Maya’’ by Webster (2002) have argued for and against drought as an agent of cultural change, respectively. Our purpose here is simply to review the physical evidence of prehistoric droughts found in Lake Chichancanab sediments, and point out potential implications with respect to Maya cultural change. The so-called collapse, decline, transition, or transformation of Classic Maya civilization can be viewed as a progressive chain of events that occurred in a 300-year period from ca 750 to 1050 AD (Rice et al., 2004).

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

Transformation began earliest in the southern lowlands and culminated with the fall of Chichen Itza in the northern lowlands during the 11th century (Yaeger and Hodell, in press). The Terminal Classic transition was a complex process, however, with considerable local, regional, and temporal variability associated with the decline of Maya polities (Demarest et al., 2004). The Chichancanab record leaves little doubt that there were significant climate changes occurring in at least part of the Maya lowlands during the Terminal Classic Period. Our dating of the TCD to between 770 and 1100 AD falls squarely in the period of most rapid cultural transformation. Similar to the cultural changes that occurred in the terminal Classic Period, the climate history is also considerably more complex than previously thought. Our earlier work (Hodell et al., 1995) has been misinterpreted as suggesting that the entire period from 800 to 1000 AD was dry, and the term ‘‘megadrought’’ has been used to refer to this twocentury long period. This false impression was provided by the thick gypsum horizon in the 1993 core, collected in 6.9 m of water. The gypsum lens was dated by AMS radiocarbon analysis of a seed that yielded a calibrated age range of 780–990 AD. The gypsum horizon in shallow water is clearly condensed (Fig. 4) and the age range refers to the 95.4% probability distribution of the calibrated radiocarbon date, not the duration of the drought. Expanded, higher-resolution sections from Chichancanab cores taken in deeper water clearly indicate a complex pattern of two drought phases separated by an intervening moister period. Within each of the drought phases, drier periods alternated with wetter periods with a recurrence interval of 50 years. The finer-grained paleoclimatic history presented here will aid archaeologists in evaluating the potential role that climate change may have played either locally or regionally in the cultural transformation that occurred in the Terminal Classic Period. Both the paleoclimate and archaeology communities, however, face formidable challenges in accurately and precisely dating past climatic and cultural changes. These problems are amplified when trying to correlate the two data sets (Yaeger and Hodell, in press). Because correlations are subject to the uncertainties inherent in radiocarbon dating, it will be difficult to achieve more than a general temporal correlation between paleoclimate and archaeological data without other methods of dating. The records from Lakes Chichancanab and Punta Laguna provide the most convincing case for climate change in the northern Maya lowlands during the terminal Classic Period. How representative are these records of other areas in Mesoamerica? Discrepancies in oxygen isotope records among lakes in different parts of the Maya lowlands have been cited as evidence that the TCD was not widespread throughout the region. These oxygen isotope discrepancies, however, do not necessa-

1425

rily reflect different climate histories. Rather, they may reflect differences in a lake’s sensitivity to climate change (e.g., the degree to which the basin is closed or open) or other factors such as human disturbance of watershed hydrology. For example, lakes in the southern Maya lowlands were heavily impacted by high population densities in the Classic Period. Human deforestation of watersheds can alter lake hydrology by changing surface runoff and groundwater inflow in ways that mimic climate change (Rosenmeier et al., 2002a, b, in press). Some lakes in the northern lowlands suffer from poor carbonate microfossil preservation, and their chronologies may be compromised by reliance on bulk radiocarbon dates that are unreliable (Whitmore et al., 1996; Leyden et al., 1996, 1998). Small cenotes or aguadas can switch quickly between being hydrologically open or closed by plugging or unplugging of conduits, which can alter lake water d18O independent of climate change (Hodell et al., in press). All lakes are not appropriate for studying paleoclimate (see discussion in Brenner et al., 2003). We believe that Lake Chichancanab’s sediments provide a robust record of regional climate change for the Maya lowlands because of the sensitivity of the aquatic system to climate change, and because the record was not seriously confounded by human disturbance in the watershed. The similarity of the Chichancanab record with marine data from the Cariaco Basin off northern Venezuela (Haug et al., 2003) suggests that the TCD was a widespread climate phenomenon and not restricted to north-central Yucatan. Other proxy records from the Yucatan Peninsula such as lakes (with little human impact), tree rings (Stahle, pers. comm.) and speleothems (Webster, 2000) are needed to evaluate intraregional differences in climate histories. Assessing the role of climate change in the cultural transformations that occurred in the Terminal Classic Period requires accurate determination of the timing, magnitude, and spatial extent of paleoclimatic and paleoenvironmental change. Nevertheless, the patterns of paleoclimate and social change should be compared keeping in mind the chronological and methodological limitations of the paleoclimate and archaeological data.

7. Conclusion Lake Chichancanab provides one of the most sensitive and reliable paleoclimate records of the Maya Lowlands because of its substantial water loss to evaporation, precipitation of gypsum (CaSO4), and relatively low-density occupation of its watershed. New sediment cores from deep water have revealed details of climate change associated with the terminal Classic Period when the Maya civilization underwent decline. The TCD was not a single megadrought lasting

ARTICLE IN PRESS 1426

D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427

for centuries, but rather consisted of two phases of dry climate separated by an intervening period of relatively moister conditions. Dates on paleoclimate events remain uncertain because of the inherent errors associated with radiocarbon dating, but the pattern and frequency of climate variations are robust. We estimate that the TCD lasted from ca 770 to 1100 AD and was comprised of two drought phases (ca 770–870 AD and 920–1100 AD) and an intervening 50-year period (ca 870–920 AD) of more mesic conditions. Within each of the dry phases, climate was marked by alternating dry and moist conditions with an interval of drought recurrence of about 50 years. The bipartite drought history inferred from Chichancanab is supported by oxygen isotope records from nearby Punta Laguna. The general pattern is also consistent with findings from the Cariaco Basin off northern Venezuela (Haug et al., 2003), suggesting that the TCD was a widespread phenomenon and not limited to north-central Yucatan.

Acknowledgements We thank Alma Albornaz Pat, Enrique Ildefonso Chan Can, and Roger Medina Gonzalez of the Universidad Autonoma de Yucatan for logistical field support. Radiocarbon analyses were made at the Center for AMS under the auspices of the US Department of Energy by University of California Lawrence Livermore National Laboratory Contract W-7405-Eng-48. We thank G. Haug for his thoughtful review and making available the titanium data from ODP Hole 1002D. References Back, W., Hanshaw, B.B., 1970. Comparison of chemical hydrogeology of the carbonate peninsulas of Florida and Yucatan. Journal of Hydrology 10, 330–368. Boose, E.R., Foster, D.R., Barker Plotkin, A., Hall, B., 2003. Geographical and historical variation in hurricanes across the Yucatan Peninsula. In: Gomez-Pompa, A., Allen, M.F., Fedick, S.L., Jimenez-Osornio, J.J. (Eds.), The Lowland Maya: Three Millennia at the Human–Wildland Interface. Haworth Press, Binghamton, NY, pp. 495–516. Brenner, M., Hodell, D.A., Curtis, J.H., Rosenmeier, M.F., Anselmetti, F.S., Ariztegui, D., 2003. Paleolimnological approaches for inferring past climate change in the Maya region: recent advances and methodological limitations. In: Gomez-Pompa, A., Allen, M.F., Fedick, S.L., Jimenez-Osornio, J.J. (Eds.), The Lowland Maya: Three Millennia at the Human–Wildland Interface. Haworth Press, Binghamton, NY, pp. 45–75. Bronk Ramsey, C., 1998. Probability and Dating. Radiocarbon 40 (1), 461–474. Covich, A.P., 1970. Stability of molluscan communities; a paleolimnologic study of environmnental disturbance in the Yucatan Peninsula. Ph.D. Dissertation. Yale University, 163 pp. Covich, A., Stuiver, M., 1974. Changes in oxygen 18 as a measure of long-term fluctuations in tropical lake levels and molluscan populations. Limnology and Oceanography 19 (4), 682–691.

Curtis, J.H., Hodell, D.A., Brenner, M., 1996. Climate variability on the Yucatan Peninsula (Mexico) during the past 3500 years, and implications for Maya cultural evolution. Quaternary Research 46, 37–47. Dahlin, B.H., 1983. Climate and prehistory on the Yucatan Peninsula. Climate Change 5, 245–263. Deevey, E.S., 1988. Estimation of downward leakage from Florida lakes. Limnology and Oceanography 33, 1308–1320. Demarest, A.A., Rice, P.M., Rice, D.S., 2004. The terminal classic in the Maya lowlands: assessing collapses, terminations, and transformations. In: Demarest, A.A., Rice, P.M., Rice, D.S. (Eds.), The Terminal Classic in the Maya Lowlands. University Press of Colorado, Boulder, CO, pp. 545–572. Doehring, D.O., Butler, J.H., 1974. Hydrological constraints on groundwater management in the Yucatan Peninsula, Mexico. Journal of Hydrology 10, 330–368. Giannini, A., Kushnir, Y., Cane, M.A., 2000. Interannual variability of Caribbean rainfall, ENSO, and the Atlantic Ocean. Journal of Climate 13, 297–311. Giannini, A., Kushnir, Y., Cane, M.A., 2001a. Seasonality in the Impact of ENSO and the North Atlantic High on Caribbean Rainfall. Physics and Chemistry of the Earth, Part B: Hydrology, Oceans and Atmosphere 26, 143–147. Giannini, A., Kushnir, Y., Cane, M.A., 2001b. Interdecadal changes in the ENSO teleconnection to the Caribbean Region and the North Atlantic Oscillation. Journal of Climate 14, 2867–2879. Giddings, L., Soto, M., 2003. Rhythms of precipitation in the Yucatan Peninsula. In: Gomez-Pompa, A., Allen, M.F., Fedick, S.L., Jimenez-Osornio, J.J. (Eds.), The Lowland Maya: Three Millennia at the Human–Wildland Interface. Haworth Press, Binghamton, NY, pp. 77–89. Gill, R.B., 2000. The Great Maya Droughts: Water, Life, and Death. University of New Mexico Press, Albuquerque 464 pp. Gray, C.R., 1993. Regional meteorology and hurricanes. In: Maul, G.A. (Ed.), Climatic Change in the Intra-Americas Sea. Edward Arnold, London, pp. 87–99. Hastenrath, S., 1984. Interannual variability and the annual cycle: mechanisms of circulation and climate in the tropical Atlantic sector. Monthly Weather Review 112, 1097–1107. Hastenrath, S., 1991. Climate Dynamics of the Tropics. Kluwer Academic Publishers, Boston 488 pp. Haug, G.H., Gunther, D., Peterson, L.C., Sigman, D.M., Hughen, K.A., Aeschlimann, B., 2003. Climate and the collapse of Maya civilization. Science 299, 1731–1735. Hodell, D.A., Curtis, J.H., Brenner, M., 1995. Possible role of climate in the collapse of Classic Maya civilization. Nature 375, 391–394. Hodell, D.A., Brenner, M., Curtis, J.H., Guilderson, T., 2001. Solar forcing of drought frequency in the Maya lowlands. Science 292, 1367–1369. Hodell, D.A., Brenner, M., Curtis, J.H., Medina, R.G., Ildefonso, E., Albornaz, A., Guilderson, T.P., in press. Climate change on the Yucatan Peninsula during the Little Ice Age. Quaternary Research, in press. Hunt, B.G., Elliott, T.I., in press. A simulation of climatic conditions associated with the collapse of the Mayan civilization. Climatic Change. Illescas-Pasquel, F., 1950. Fisi-quimica de las agues de Yucatan. Boletı´ n de la Sociedad Mexicana de Geografı´ a y Estadı´ stica 69, 205–224. Leyden, B.W., 2002. Pollen evidence for climatic variability and cultural disturbance in the Maya lowlands. Ancient Mesoamerica 13, 85–101. Leyden, B.W., Brenner, M., Whitmore, T., Curtis, J.H., Piperno, D.R., Dahlin, B.H., 1996. A record of long- and short-term climatic

ARTICLE IN PRESS D.A. Hodell et al. / Quaternary Science Reviews 24 (2005) 1413–1427 variation from Northwest Yucatan: Cenote San Jose Chulchaca. In: Fedick, S.L. (Ed.), The Managed Mosaic: Ancient Maya Agriculture and Resource Use. University of Utah Press, Salt Lake City, pp. 30–52. Leyden, B.W., Brenner, M., Dahlin, B., 1998. Cultural and climatic history of Coba, a lowland Maya city in Quintana Roo, Mexico. Quaternary Research 49, 111–122. Magan˜a, V., Amador, J.A., Medina, S., 1999. The midsummer drought over Mexico and Central America. Journal of Climate 12, 1577–1588. Marin, L., 1990. Field investigations and numerical simulation of ground-water flow in the karstic aquifer of northwestern Yucatan, Mexico. Ph.D. Dissertation. Northern Illinois University, 183 pp. Paillard, D., Labeyrie, L., Yiou, P., 1996. Macintosh program performs time-series analysis. Eos 77, 379. Perry, E., Velazquez-Oliman, G., Marin, L., 2002. The hydrogeochemistry of the karst aquifer system of the northern Yucatan Peninsula, Mexico. International Geology Review 44, 191–221. Perry, E., Velazquez-Oliman, G., Socki, R.A., 2003. Hydrogeology of the Yucata´n Peninsula. In: Gomez-Pompa, A., Allen, M.F., Fedick, S.L., Jimenez-Osornio, J.J. (Eds.), The Lowland Maya: Three Millennia at the Human–Wildland Interface. Haworth Press, Binghamton, NY, pp. 115–138. Rice, P.M., Demarest, A.A., Rice, D.S., 2004. The terminal classic and the ‘‘Classic Maya Collapse’’ in perspective. In: Demarest, A.A., Rice, P.M., Rice, D.S. (Eds.), The Terminal Classic in the Maya Lowlands. University Press of Colorado, Boulder, CO, pp. 1–11. Rosenmeier, M.F., Hodell, D.A., Brenner, M., Curtis, J.H., 2002a. A 4000-year lacustrine record of environmental change in the southern Maya lowlands, Peten, Guatemala. Quaternary Research 57, 183–190. Rosenmeier, M.F., Hodell, D.A., Brenner, M., Curtis, J.H., Martin, J.B., Anselmetti, F.S., Ariztegui, D., Guilderson, T.P., 2002b. Influence of vegetation change on watershed hydrology: Implica-

1427

tions for paleoclimatic interpretation of lacustrine d18O records. Journal of Paleolimnology 27, 117–131. Rosenmeier, M.F., Brenner, M., Hodell, D.A., Martin, J., Binford, M., and Islebe, G., in press. Quantitative assessments of Holocene environmental change in Pete´n, Guatemala: Catchment hydrology and lake water d18O predictive models. Quaternary Research, in press. Stenich, B., Marin, L.E., 1997. Determination of flow characteristics in the aquifer in northwest Yucatan, Mexico. Journal of Hydrology 191, 315–331. Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen, K.A., Kromer, B., McCormac, F.G., van der Plicht, J., Spurk, M., 1998. INTCAL98 radiocarbon age calibration 24,000-0 cal BP. Radiocarbon 40, 1041–1083. Waliser, D.E., Shi, Z., Lanzante, J.R., Oort, A.H., 1999. The Hadley circulation: assessing NCDP/NCAR reanalysis and sparse in-situ estimates. Climate Dynamics 15 (10), 719–735. Webster, D., 2002. The Fall of the Ancient Maya: Solving the Mystery of the Maya Collapse. Thames and Hudson, London 368 pp. Webster, J.M., 2000. Speleothem evidence of late Holocene climate variation in the Maya lowlands of Belize, Central America, and archaeological implications. Ph.D. Dissertation, University of Georgia, 232 pp. Whitmore, T.J., Brenner, M., Curtis, J.H., Dahlin, B.H., Leyden, B.W., 1996. Holocene climatic and human influences on lakes of the Yucatan Peninsula, Mexico: An interdisciplinary, palaeolimnological approach. Holocene 6 (3), 273–287. Wilson, E.M., 1980. Physical geography of the Yucatan Peninsula. In: Moseley, E., Terry, E.D. (Eds.), Yucatan: A World Apart. University of Alabama Press, pp. 5–40. Yaeger, J., Hodell, D.A. [in press]. Climate–culture–environment interactions and the collapse of Classic Maya civilization. In: Sandweiss, D.H., Quilter, J. (Eds.), El Nin˜o, Catastrophism, and Culture Change in Ancient America. Dunbarton Oaks, Washington, DC.