QUATERNARY RESEARCH ARTICLE NO.

47, 169180 (1997)

QR971881

A 3500

C yr High-Resolution Record of Water-Level Changes in Lake Titicaca, Bolivia/Peru

MARK B. ABBOTT

Department of Geosciences, Morrill Science Center, University of Massachusetts Box 35820, Amherst, Massachusetts 01003-5820

MICHAEL W. BINFORD

Graduate School of Design, Harvard University, 48 Quincy Street, Cambridge, Massachusetts 02138

MARK BRENNER

Department of Fisheries and Aquatic Sciences, University of Florida, 7922 NW 71st Street, Gainesville, Florida 32653

KERRY R. KELTS

Limnological Research Center, University of Minnesota, 220 Pillsbury Hall, 310 Pillsbury Drive SE, Minneapolis, Minnesota 55455 Received May 7, 1996
Sediment cores collected from the southern basin of Lake Titicaca (Bolivia/Peru) on a transect from 4.6 m above overow level to 15.1 m below overow level are used to identify a new centuryscale chronology of Holocene lake-level variations. The results indicate that lithologic and geochemical analyses on a transect of cores can be used to identify and date century-scale lake-level changes. Detailed sedimentary analyses of subfacies and radiocarbon dating were conducted on four representative cores. A chronology based on 60 accelerator mass spectrometer radiocarbon measurements constrains the timing of water-level uctuations. Two methods were used to estimate the 14C reservoir age. Both indicate that it has remained nearly constant at 250 14C yr during the late Holocene. Core studies based on lithology and geochemistry establish the timing and magnitude of ve periods of low lake level, implying negative moisture balance for the northern Andean altiplano over the last 3500 cal yr. Between 3500 and 3350 cal yr B.P., a transition from massive, inorganic-clay facies to laminated organic-matter-rich silts in each of the four cores signals a waterlevel rise after a prolonged mid-Holocene dry phase. Evidence of other signicant low lake levels occurs 29002800, 24002200, 20001700, and 900500 cal yr B.P. Several of the low lake levels coincided with cultural changes in the region, including the collapse of the Tiwanaku civilization. 1997 University of Washington.

INTRODUCTION

Highly resolved lacustrine records are useful for studying the mechanisms and effects of climate change. Spatial and
temporal resolution must be both sufciently extensive and ne-scaled to describe patterns that appear at the scale of the processes of interest. When the affected processes are certain human activities, this criterion for ne-scale resolution can be achieved by spatially dening the unit of study as a lake and its drainage basin and temporally as the period of habitation by humans. Furthermore, the measurement and description of paleoclimate at a lake basin point will be even more valuable if it is imbedded in a spatially extensive web of other point descriptions. South America has a scarcity of sites with century-scale paleoclimate data sets, yet is extremely important because El Nino/Southern Oscillation events (ENSO) cause major economic hardships, the intertropical convergence zone (ITCZ) migrates over two-thirds of the surface area annually, the vast Amazon basin is the largest remaining forested area in the world (with important climatic and paleoclimatic implications), and several civilizations have developed and collapsed on the continent. The Lake Titicaca drainage basin and associated altiplano in the Peruvian and Bolivian Andes is an endorheic system that was also the site of the Tiwanaku civilization. Nearby alpine glaciers, and the lake itself, contain paleoclimate records. Several previous studies have been done in the Titicaca watershed (Thompson et al., 1985; Wirrmann and Mourguiart, 1995; Abbott et al., 1997). In this paper we describe a nely resolved record of lake-level change driven by climatic variability over the past 3500 yr, and in a compan-

0033-5894/97 $25.00 Copyright 1997 by the University of Washington. All rights of reproduction in any form reserved.

QR 1881

a607$$$$61

03-18-97 16:12:53

AP: QR

ABBOTT ET AL.

ion paper Binford et al. (1997) describe the effects of climate variation on civilization. Low lake stands during the middle to late Holocene have been postulated for Lake Titicaca (Wirrmann and Mourguiart, 1995; Wirrmann et al., 1992; Wirrmann and Oliveira Almeida, 1987), but the timing, rate, and mechanism for declines and returns to higher levels remains poorly described. Here we report evidence that suggests a rapid lakelevel rise of 15 to 20 m about 3500 yr B.P. and several century-scale low stands at 29002800, 24002200, 2000 1700, and 900500 cal yr B.P. These ndings substantially improve our knowledge of the timing, duration, and magnitude of variations in the precipitationevaporation balance of the South American altiplano during the late Holocene. This study also provides the rst accurate AMS radiocarbon chronologies required to resolve century-scale dynamics of precipitationevaporation variations on the altiplano. This paper has four objectives: (1) to determine lakelevel changes by identifying sediment unconformities from detailed core descriptions, smear-slide mineralogy, and the geochemical properties of sediment cores; (2) to dene the magnitude of lake-level changes in the Lake Titicaca system based on a transect of cores from shallow to deeper water (0.7, 4.2, 6.0, and 12.6 m below overow level); (3) to determine a reservoir age model for lake Titicaca to correct 14 C dates prior to calibration and assess whether the age has shifted during the past ca. 3500 14C yr; and (4) to determine a high-resolution chronology for lake-level changes based on 61 AMS radiocarbon dates.

STUDY AREA

Lake Titicaca has an area of ca. 8500 km2, a drainage of ca. 57,000 km2, and includes the connected Lago Grande and Lago Winaymarka basins (Fig. 1). Lake Titicaca has undergone measurable lake-level changes during the historic period (1914present) ranging from 3806.2 m in 1943 to 3812.6 m in 1986, with an average annual uctuation of 0.8 m (Roche et al., 1992). Although Lake Titicaca has varied between a hydrologically open and closed system during the Holocene, it lies in the upper part of a much larger endorheic system that includes Lago Poopo and the vast salares in central and southern Bolivia, respectively. Today the lake drains over a 3804-m sill down the RB o Desaguadero from the southwest corner of Lago Winaymarka (Wirrmann, 1992). We use the elevation of the sill as the base datum for reporting lake-level changes as meters BOL (below overow level) because of the strong interannual variability of lake levels. This index horizon facilitates description of core boundary depths and lake-level changes inferred from the cored transect. The sill separating the eastern and western basins of Lago Winaymarka lies at 10 m BOL. When the

water level of Lake Titicaca falls 10 m BOL to 3794 m two separate subbasin lakes are formed. The eastern basin remains connected to Lake Titicaca proper (Lago Grande) by the Tiquina Strait until lake level falls below 16 m BOL (3788 m), and then the Titicaca system separates into three separate lake basins. The four cores from shallow regions of Lago Winaymarka are assumed to represent changes in Lake Titicaca as a whole. This is defensible given the morphology of the connections to the main lake and local stream sources. The Lake Titicaca basin is particularly sensitive to shifts in the precipitationevaporation balance because even with the overowing conditions that prevail today, only 1 to 3% of the lake water is lost by overow. During the recorded period, lake level has remained above the overow level, although most of the water is removed by evaporation. Estimates of the amount of water lost historically by evaporation range from a mean of 91% from 1968 to 1987 to 99% from 1956 to 1978. Estimates for the average residence time of water in Lake Titicaca range from 60 to 175 yr (Carmouze, 1992; Roche et al., 1992; Han, 1995). The water balance of the altiplano is affected by many factors including ENSO events, uctuations in the seasonal location of the ITCZ, and changes in the strength of summer monsoon circulation. Strong ENSO years correlate with drought on the altiplano (Roche et al., 1992). There are strong seasonal contrasts in precipitation, with more than 78% of the average annual precipitation (760 mm/yr basinwide) occurring during the summer wet season (December February), when the ITCZ reaches its southernmost extent. Maximum precipitation in the Lake Titicaca watershed occurs on the high mountains in the northeast corner, reaching totals of 1000 mm/yr, and on the southern shore of Lago Grande, where precipitation totals of 1100 mm/yr are enhanced by lake-effect moisture (Roche et al., 1992).

METHODS

A transect of sediment cores was collected to identify and map the major sediment transitions. Although 15 cores were described, we focused on four representative cores for detailed sediment analysis and high-resolution dating. Cores were taken with a square-rod piston corer (Wright et al., 1984) and a piston corer designed to collect undisturbed sedimentwater interface proles (Fisher et al., 1992). Organic matter was measured by weight loss on ignition (LOI) at 550C (Hakanson and Jansson, 1983) and carbonate con tent was assessed from the weight loss between 550 and 1000C (Dean, 1974). Calcium, magnesium, iron, and potassium in bulk-sediment samples were measured on a JarrellAsh 9000 Inductively Coupled Argon Plasma Spectrophotometer, following ashing at 550C and digestion for 1 hr

a607$$$$62

LATE HOLOCENE FLUCTUATIONS OF LAKE TITICACA
FIG. 1. Map showing the location of Lake Titicaca in South America including Lago Grande and the two subbasins of Lago Winaymarka. The bathymetric map shows core sites A and C in the western basin and B and D in the eastern basin. Water drains from Lago Grande through the Tiquina Strait into Lago Winaymarka and out of Lake Titicaca down the RB o Desaguadero.

in boiling 1 N HCl. Lithology was determined from smearslide mineralogy and detailed inspection of sediments, noting Munsel color, texture, sedimentary structures, and biogenic features. Most stratigraphic levels contained insufcient terrestrial organic material for AMS 14C measurements. Therefore we used calcite shells from the abundant aquatic gastropods (Littoridina andecola and Littoridina sp.). All sample material for 14C measurements was wet-sieved through nested screens (500, 250, and 125 mm), microscopically inspected, sonically cleaned, and archived in precombusted glass containers. Carbonate samples were pretreated with 10% dissolution using HCl. Radiocarbon dates were measured at the Center for Accelerator Mass Spectrometry at the Lawrence Livermore National Laboratory (CAMS). Radiocarbon ages are reported either as 14C yr B.P. (uncalibrated) or cal yr B.P. if corrected and calibrated according to the methods outlined for CALIB 3.0 by Stuiver and Reimer (1993). Abrupt sediment transitions interpreted as erosion surfaces were 14C dated by taking samples 1 cm above and below the disturbed contact to avoid reworked material. Where an abrupt transition was interpreted as an unconformity, a 14C measurement from the upper surface was interpreted as an estimate of the age of transgression. A date from just below the unconformity denes a maximum age for the low lake stand because the amount of eroded sediment is unknown. In some cases, these desiccation surfaces show little or no evidence of erosion.
Although sediment transitions associated with subaerial exposure can be identied in a single core, the rate and magnitude of water-level change was resolved with a transect of cores from shallow to deep water. This core series was used to identify subfacies related to increasing water depths (Binford et al., 1992). Surface sediments yielded information that was used to calibrate sediment subfacies formed in particular depth ranges in Lake Titicaca. Waterlevel reconstructions are based on these criteria. Exposure surfaces were identied by (1) scour marks, (2) mud cracks, (3) abrupt transitions (1 cm) characterized by coarser grained (ne sand) sediments with high bulk density (1 g/cm3) overlying ne-grained organicrich muds (20% organic matter), (4) an abrupt increase in iron and potassium concentration associated with the reducing conditions in water-saturated soils, and (5) highly fragmented shell material in the overlying muds. The presence of one or more of these characteristics combined with an abrupt change in the radiocarbon activity of adjacent strata indicate erosion or nondepositional surfaces. We used detailed core descriptions, a smear-slide mineralogy, and radiocarbon stratigraphy to delimit watersaturated soils and erosion surfaces formed during low water stands and subaerial exposure. Shallow-water subfacies (2 m water depth) were identied by (1) the presence of high concentrations of achenes (seeds) of the littoral sedge Schoenoplectus tatora in a coarse-grained matrix (silt to sand), (2) large amounts

of aquatic plant macrofossils (Myriophyllum, Chara, and Potamogeton), and (3) sediments containing 90% CaCO3 composed of calcied macrophyte coatings and fragmented mollusk shells. During prolonged low stands, water-saturated soil formation is more intense, as indicated by order-of-magnitude increases in iron and potassium.

RESULTS

Reservoir Age Measurements and Calibration Radiocarbon dates derived from aquatic organisms may be signicantly older than their true age of deposition because of the long residence time of the lake water and the presence of limestone in the drainage basin that is a source of 14C-depleted carbonate. The contemporary reservoir age of Lake Titicaca was estimated by measuring the 14C activity of aquatic gastropods (L. andecola) taken from the A.D. 1900 stratigraphic level (identied by 210Pb dating) to avoid younger samples contaminated by fossil fuels and nuclearweapons testing (Levin et al., 1989). The measured fraction Modern was corrected for radioactive decay since A.D. 1950 and compared with the value expected from Stuiver and Pearson (1993). The result is a 250-yr offset, which is subtracted from the measured 14C ages prior to calibration (Stuiver and Reimer, 1993). When lake level falls below 3804 m, Lake Titicaca has no surface outow and residence time increases. It was thus critical to check whether the 14C reservoir age of Lake Titicaca varied over past centuries. We assessed changes in the 14 C reservoir effect for the past 3500 yr by measuring the 14 C activity of paired samples formed of carbon from aquatic and atmospheric sources, respectively, collected from the same stratigraphic level. Radiocarbon measurements of L. andecola shells and S. tatora achenes from ve nearly equivalent levels at four core sites indicate that the 250-yr offset has been consistent through time (compare CAMS 017006 to 017048, 016995 to 04981, 016998 to 04978, 011976 to 013601, and 013608 to 013609 in Table 1). Sediment Cores Detailed descriptions of sediment cores A, B, and D are included as examples of sedimentary facies from shallow water (5 m BOL), intermediate water (510 m BOL), and deeper water (10 m BOL) sites, respectively. Radiocarbon dating focused on cores A, B, C, and D to develop centuryscale chronologies. The stratigraphy and water depth of core C are similar to core B described below and are therefore not discussed here. Sediment boundaries labeled ES-1 through ES-5 are interpreted as erosion surfaces (ES) and were identied by changes in color, texture, grain size, mineralogy, organic content, biogenic features, and bulk-sedi-

ment geochemistry (Figs. 2 and 3). If these surfaces are interpreted as intervals of continuous sedimentation then the units are labeled S-1 through S-4. The radiocarbon dates from stratigraphic levels above and below the erosion surfaces are used as supporting evidence for periods of erosion or nondeposition. Table 1 lists the radiocarbon dates and Table 2 summarizes the age interpretations of the upper and lower boundaries. The radiocarbon dates on ES-5 are variable partly because the samples were arranged to provide an even spread along the length of the core. The dates bracket the unconformities, but do not dene them exactly. Core A was collected in the western basin of Lago Winaymarka from 12.6 m BOL (16.6 m water depth when the core was collected in August 1993). The core is 6.6 m long and contains one abrupt sediment transition at 14.2 m BOL (ES1) and two layers of nearly pure gastropod shell material at 13.7 (S-3) and 13.1 m BOL (S-5) (Fig. 2). Fourteen radiocarbon dates dene the abrupt ES-1 boundary and two shell layers S-3 and S-5 that coincide with erosion surfaces ES3 and ES-5 in cores B, C, and D. Analyses of smear-slide mineralogy show that the sediments immediately below the ES-1 contact contain a higher concentration of clastic component, coarser grain size (silt to ne sand), and decreased organic matter compared with the sediments directly overlying the boundaries. We interpret the shell layers at the S-3 and S-5 contacts as lag deposits formed during a period of lowered lake level, although water still covered the core site and no erosion occurred. The sediments below 14.2 m BOL are massive, coarsegrained (silt to ne sand), and contain terrestrial sedge seeds suggesting subaerial exposure. Sedimentary structures at the ES-1 boundary are typical erosion scour marks. Aquatic gastropods are absent from the lower boundary. Weakly laminated, ne-grained (clayey-silt) lacustrine muds above ES1 are dated 3510 / 120/040 cal yr B.P. (CAMS-11976), documenting the age of lake-level rise. There is no evidence for the S-2 and S-4 contacts in core A, either because the lake did not drop sufciently or because the accumulation rate of this core is slow relative to cores from the eastern basin. Between 3510 /120/040 and 2270 /50/0150 cal yr B.P. (CAMS-11973) calcium carbonate content increased (40 to 50%), organic matter decreased (40 to 30%), and clastic material remained relatively constant (20%). The S-3 contact is marked by a 1-cm-thick layer of gastropod shells (L. andecola). Coincident increases in grain size (silt), clastic material (50%), and accumulation rate are consistent with a shallow-water environment. The S-3 contact is interpreted as a lag deposit formed during a low lake stand, during which material was transported from recently exposed sites. Likewise, sediments forming the S-5 contact show high concentrations of gastropod shells and an increase

TABLE 1 AMS Radiocarbon Dates and Calibrated Ages a from Lake Titicaca Sediment Cores
Core depth (cm BOL) Measured radiocarbon age (14C yr B.P.) Measured error (14C yr) Median calibrated age (cal yr B.P.) 7390

CAMS No. 5741 5742

Core A A A A A A A A A A A A A A B B B B B B B B B B B B B C C C C C C C C C C C C C D D D D D D D D D D D D D D D D D D D D
Material Gastropod shell Gastropod shell Gastropod shell Gastroped shell Gastroped shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Fish scale Fish scale Gastropod shell S. tatora achenes Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell S. tatora achenes S. tatora achenes Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell S. tatora achenes Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell S. tatora achenes Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell Gastropod shell S. tatora achenes Gastropod shell Gastropod shell

Calibrated (/) error 80

Calibrated (0) error 50
a Dates were calibrated by rst subtracting the reservoir age (see text) and then calculating the age by using the computer program CALIB 3.0 (Stuiver and Reimer, 1993).

a607$$1881

FIG. 2. Stratigraphy and sediment properties for cores A, B, and D. Note the abrupt shifts in core properties across the disconformable surfaces, as indicated by results from smear-slide mineralogy, loss on ignition at 550 and 900C, dry bulk density measurements, and bulk-sediment geochemistry.
FIG. 3. Transect of cores and age-depth plots showing the location and magnitude of the unconformities. The elevations of cores A, B, C, and D are shown with respect to the overow level of the lake into RB o Desaguadero. The western basin is isolated from the eastern basin when the water falls 10 m BOL (below overow level). Likewise, the eastern basin is isolated from Lago Grande when the water level falls 15 m BOL.

in clastic material from 10% to nearly 40%. Above the S5 contact the organic matter content of the sediments increases to 60% and ne-grained carbonate precipitates comprise 10% of the sediments, consistent with deeperwater sediment facies. Core B was collected in the eastern basin of Lago Winay marka from 6.0 m BOL (11.1 m water depth in August 1992). The core is 5.4 m long and contains four abrupt sediment transitions at 6.7 m (ES-5), 7.9 m (ES-4), 8.6 m (ES-3), and 9.1 m (ES-1) BOL (Fig. 2) within a 3.5-m-thick late Holocene section. Thirteen radiocarbon dates show that the four abrupt transitions coincide with major shifts in the age plots of the sediments. Figure 2 illustrates the massive sediments below the ES1 transition that are dominated by ne-grained calcareous mud (silty clay). This subfacies is characterized by high bulk density (1 g/cm3), low organic matter (10%), and
increased iron content, consistent with subaerial exposure. The sediments overlying the ES-1 contact include gastropod shell debris, silt-sized mineral clastics, S. tatora achenes, and aquatic macrophytes, indicating a shallow-water facies. This facies is equivalent to ES-2 at core site D. The sediments overlying the ES-3 contact are characterized by increased grain size (from clayey silt to silt), clastic content (from 10 to 45%), and bulk density (from 0.3 to 0.5 g/cm3). This is consistent with a shallow-water environment. The sediments overlying the erosion surface contain increased gastropod shell content. The sediments overlying the ES-4 contact are characterized by increased bulk density (0.4 to 0.6 g/cm3), increased Fe content, and increased grain size (clayey silt to silt). The overlying sediments exhibit high concentrations of gastropod shells and aquatic macrophytes, consistent with a shallowwater environment.

a607$$$$63

TABLE 2 Radiocarbon Dates Bracketing Erosion Surfaces ES-1 through ES-5 and Sediment Units S-1 through S-4
Elevation of erosion surface (cm BOL) No erosion 204 No erosion Eroded No erosion 374 No erosion Shallow water Shallow water 480 Elevation of upper age (cm BOL) Upper age (cal yr B.P.) Elevation of lower age (cm BOL) Lower age (cal yr B.P.)
Boundary S-5 ES-5 ES-5 ES-5 S-4 ES-4 ES-4 ES-4 S-3 ES-3 ES-3 ES-3 S-2 S-2 S-2 ES-2 ES-1 ES-1 ES-1 ES-1
Core A B C D A B C D A B C D A B C D A B C D

203 765

No date See ES-2250 2240

673 615

See ES-2470

721 381

The sediments overlying the ES-5 contact are characterized by coarser grain sizes (from clayey silt to silt), clastic content (from 20 to 35%), and bulk density (from 0.3 to 0.5 g/cm3). Coincident increases in Fe and K are consistent with the shallow-water oxidizing facies overlying the ES-5 boundary in core D. The thickness of this interval is less in core B, consistent with an increasing water level from 6.7 to 2.1 m BOL. The muds overlying the ES-5 boundary have increased organic content and decreased concentrations of gastropod shells, consistent with deeper-water facies. Core D was collected in the eastern basin of Lago Winay marka from 0.7 m BOL (5.8 m water depth in August 1992). The core is 4.4 m long and contains four abrupt sediment transitions (1 cm thick) at 2.0 (ES-5), 3.8 (ES-3), 4.4 (ES2), and 4.8 m (ES-1) BOL in the late Holocene section. Twenty radiocarbon measurements reveal that the four abrupt contacts coincide with shifts in the radiocarbon age of L. andecola shells, suggesting periods of erosion or nondeposition. The ES-1 boundary at 4.9 m BOL is marked by a shift from massive-gray clays to weakly laminated, organic silts. The sediments below the ES-1 transition contain 5% organic matter, 70% ne-grained calcite (silty clay), elevated Fe (2 mg/g), and high wet bulk density (1 g/cm3) (Fig. 2). They also contain less organic matter, fragmented highly weathered gastropod shells, and S. tatora achenes, consistent with a facies formed by reworking of exposed older shallow-

LAKE-LEVEL HISTORY

A water-level history for Lake Titicaca is based on calibrated core chronologies for the past 3500 cal yr B.P. Five periods of signicant lake-level depression are documented by ve erosion surfaces (Fig. 3: ES-1 through ES-5) dened by the criteria discussed above and supported by signicant age differences between adjacent strata (Table 2). The lake was probably no more stable during the periods of high and low water than it has been during the 20th century. Therefore, the lake-level curve is represented as a broad band to indicate the level of variability constrained either by radiocarbon dates or by observations over the past century (6 m) (Fig. 4). Prior to 3500 cal yr B.P. the lake level was lower than 15 m BOL based on ages of basal sediments in six cores, four of which are discussed in this paper. Lake level rose rapidly after 3500 cal yr B.P., nearing the overow stage by 3350 cal yr B.P. High accumulation rates in core D around 3350 cal yr B.P. suggest large-scale erosion and reworking of shorelines. Hiatuses in core D between 3300 and 2900 cal yr B.P. and variations in shallow-water facies in cores B and D suggest that water level was variable at this time, uctuating between the overow stage and 8 m BOL. The age of the upper boundary of the ES-2 surface in core D indicates that lake level rose after 2900 cal yr B.P. to within 2 m of, and possibly above, the overow stage. Erosion surfaces preserved in all cores penetrating 10 m BOL and a lag deposit at 13.8 m BOL in core A indicate
that lake level dropped between 10 and 12 m BOL by 2400 cal yr B.P. Shallow-water lacustrine facies in core D indicate lake level increased abruptly to at least 2 m BOL by 2200 Cal yr B.P. Sediments at sites B and D experienced marked erosion during this low stand and have 14C ages that are similar to those for sediments overlying the ES-3 contact (Table 2). Lake level fell between 10 and 12 m BOL after 1900 cal yr B.P., as indicated by the ES-4 surface in core B. Deposits covering this period are absent in core D because they were eroded during a subsequent low stand when the ES-5 surface formed. Further supporting evidence for this low stand is found in shallow-water subfacies in core C from the western basin. Shallow-water subfacies in core B indicate lake level rose after 1650 cal yr B.P. to near the overow level. The latest prolonged low stand culminated after 700 cal yr B.P., as indicated by the age of the sediments immediately overlying the ES-5 surface in cores B, C, and D. Massive erosion occurred at core sites D and B, making it difcult to estimate the timing of the onset of lower water levels. Shallow-water facies formed in cores B and D at 600 cal yr B.P. These subfacies have characteristics consistent with deeper water sediments deposited after 500 cal yr B.P. The collapse of the Tiwanaku civilization, which relied on high lake levels for raised-eld agriculture, occurred about 800 cal yr B.P. (Binford et al., 1997), which is coincident with shallow-water sediment facies formed during a period of low lake level.

unpublished data). Large dust clouds that we observed could be the source of the dust fallout at Quelccaya. A second ice core collected from Huascaran in the north-central Andes of Peru provides a longer record that shows a marked increase in dust concentration at 2200 yr B.P. (Thompson et al., 1995). This is coincident with a 10- to 12-m low stand identied in this study and with an erosion surface in Lago Taypi Chaka Kkota, a small glacial-fed lake at 4300 m in the southeast corner of the Titicaca Watershed (Abbott et al., 1997). This was unexpected because Huascaran is located in northwestern Peru outside of the altiplano climate zone and therefore suggests a large-scale shift in the precipitation evaporation balance of the tropical Andes. Martin et al. (1993) used eld evidence obtained from four locations in South America to propose long intervals dominated by the Southern Oscillation Low Phase. These century-scale phenomena are manifested either as prolonged periods of ENSO-like conditions or as high frequencies of
individual ENSO-like events. Specically, the eld studies suggest that conditions on the altiplano were (1) drier prior to 3900 yr B.P., (2) wetter between 39003600 and yr B.P., and (3) wetter ca. 2200, 1300 yr B.P., and in the recent past. Whereas the general directions and uctuations proposed by Martin et al. (1993) are consistent with our results, we identify more events and estimate the magnitude of the water balance shifts by bracketing extent of lakelevel changes with multiple radiocarbon dates on a series of cores. The possibility remains that a connection exists between the century-scale variations in the moisture balance of the altiplano region observed in this study and the longterm variations in the frequency of ENSO events. Although ENSO events have been reviewed and categorized over historic times (cf. Anderson, 1992; Quinn et al., 1987), little is known about the changes in the longer term frequency of these events (Sandweiss et al., 1996). Mourguiart (1990) reconstructed water-level changes in Lake Titicaca based on a transfer function using the modern ostracod fauna. The results suggest the following lake level history: (1) 20 m lower prior to 7700 14C yr B.P., (2) dry or uctuating ca. 20 m lower 77003900 yr B.P., (3) 35 m below present level 39001500 yr B.P., and (4) modern levels after 1500 yr B.P. (see also Wirrmann et al., 1990). We show that the mid-Holocene dry phase ended abruptly between 3500 and 3350 cal yr B.P. We also note a 8- to 12m decrease in lake level ending after 650 cal yr B.P., as well as three other uctuations in water level. Wirrmann et al. (1990) noted a dry period at 25002300 yr B.P. in a core from 8 m of water depth from the western basin of Lago Winaymarka. We date a similar dry phase ending at 2200 cal yr B.P. Wirrmann and Mourguiart (1995) modied their original lake-level scenario, suggesting that the Lake Titicaca system rose gradually in two steps: (1) the lake level increased to 3797 m (7 m BOL) at 3800 yr B.P. and remained relatively constant until 2200 yr B.P. and (2) the water level stabilized at 3800 m (4 m BOL) at 2000 yr B.P. before rising to its present level at 1600 yr B.P. This is inconsistent with the results of our study, probably because of our closer dating and sampling resolution and because we include a carbonreservoir correction and calibration.

CONCLUSIONS

B.P. All of the cores collected in Lago Winaymarka for this study penetrate previously exposed sediments, indicating that Lake Titicaca was 15 m BOL prior to 3500 cal yr B.P. A second low stand of 58 m BOL occurred between 2900 and 2800 cal yr B.P. Shallow-water subfacies dating between 3500 and 2900 cal yr B.P. suggest that the lake remained below the overow stage during this period. The third low lake stand of 1012 m BOL ended after 2200 cal yr B.P. The duration of this dry phase is estimated to have been 200 cal yr long. The fourth low stand of 1012 m BOL ended after 1650 cal yr B.P. The duration of this phase remains unknown, but was likely ca. 200 yr based on sediment accumulation rates. The nal low lake level of 712 m BOL began prior to 900 cal yr B.P. and ended after 700 cal yr B.P. Shallow-water subfacies suggest that water level probably remained low until 500 cal yr B.P. The nal low lake stand is coincident with the decline of raised-eld agriculture and the collapse of Tiwanaku culture. Even more provocatively, at least one other low stand was approximately coincident with the transition from the earlier Chiripa culture to the Tiwanaku. This dry period was severe enough to appear in Core A, the deep water core. We do not speculate on the relationships between climate uctuation and the cultural transition because very little is known about the Chiripa culture. The cause of the observed changes in the precipitation evaporation balance remains uncertain. The abruptness (100200 yr) of the shifts between high and low lake stands, however, suggests that changes in the mode of atmospheric circulation is a likely cause. However, our data shed no new light on the mechanism.

ACKNOWLEDGMENTS

We thank Maria Elena Angulo, Clemente Catari, Jason Curtis, Barbara Leyden, and Erin Lowry for assistance in the eld. Ing. Julio Sanjines, director of the Proyecto Especial de Lago Titicaca, provided daily lakelevel data and arranged for permission to work on both the Bolivian and Peruvian sides of the Lake. Sandy P. Harrison and an anonymous referee worked very hard to help us revise an earlier manuscript and deserve special thanks. Parts of this research were supported by U.S. National Science Foundation Grant DEB-9207878 to M.W.B., U.S. National Oceanic and Atmospheric Administration Grant NA56GP0360 to M.W.B., and U.S. National Science Foundation Research training grants to the University of Minnesota.
This study demonstrates that the level of Lake Titicaca has uctuated with an amplitude 22 m during the past 3500 cal yr and that four of the low stands were both profound and occurred abruptly over a period of 100200 yr. The middle to late Holocene has been hydrologically eventful on the South American altiplano, with ve periods of low water stands in Lake Titicaca. The earliest occurred during a prolonged mid-Holocene dry phase ending after 3500 cal yr

REFERENCES

Abbott, M. B., Seltzer, G. O., Kelts, K. R., and Southon, J. (1997). Holocene paleohydrology of the tropical Andes from lake records. Quaternary Research 47, 7080. Anderson, R. Y. (1992). Long-term changes in the frequency of occurrence of El Nino events. In El Nino: Historical and Paleoclimatic Aspects of the Southern Oscillation (H. F. Diaz and V. Markgraf, Eds.). Cambridge Univ. Press, London.
ABBOTT ET AL. Limnological Knowledge (C. Dejoux and A. Iltis, Eds.), pp. 6383. Kluwer Academic, Boston. Sandweiss, D. H., Richardson, J. B., Reitz, E. J., Rollins, H. B., and Maasch, K. A. (1996). Geoarcheological evidence from Peru for a 5000 years B.P. onset of El Nino. Science 273, 15311533. Stuiver, S., and Pearson, G. W. (1993). High-precision bidecadal calibration of the radiocarbon time scale, AD 1950-500 BC and 2500-6000 BC. Radiocarbon 35, 124. Stuiver, M., and Reimer, P. J. (1993). Extended 14C data base and revised CALIB 3.0 14C age calibration program. Radiocarbon 35, 215230. Thompson, L. G., Mosley-Thompson, E., Bolzan, J. F., and Koci, B. R. (1985). A 1500-yr record of tropical precipitation in ice cores from the Quelccaya Ice Cap, Peru. Science 229, 971973. Thompson, L. G., Davis, M. E., Mosley-Thompson, E., and Liu, K-b. (1988). Pre-Incan agriculture activity recorded in dust layers in two tropical ice cores. Nature 336, 763765. Thompson, L. G. (1992). Ice core evidence from Peru and China. In Climate since A.D. 1500 (R. S. Bradley and P. D. Jones, Eds.), pp. 517 548. Routledge, London. Thompson, L. G., Mosley-Thompson, E., Davis, M. E., Lin, P.-N., Henderson, K. A., Cole-Dai, J., Bolzan, J. F., and Liu, K.-b. (1995). Late Glacial stage and Holocene tropical ice core records from Huascaran, Peru. Sci ence 269, 4650. Wirrmann, J. P. (1992). Morphology and bathymetry. In Lake Titicaca: A Synthesis of Limnological Knowledge (C. Dejoux and A. Iltis, Eds.), pp. 1623. Kluwer Academic, Boston. Wirrmann, D., and Fernando De Oliveira Almeida, L. (1987). Low Holocene level (7700 to 3650 years ago) of Lake Titicaca (Bolivia). Palaeogeography, Palaeoclimatology, Palaeoecology 59, 315323. Wirrmann, D., Mourguiart, P., and Fernando de Oliveira Almeida, L. (1990). Holocene sedimentology and ostracods distribution in Lake Titicaca-paleohydrological interpretations. In Quaternary of South America and Antarctic Peninsula, Vol. 6 (J. Rabassa, Ed.), pp. 89129. A. A. Balkema, Rotterdam. Wirrmann, J. P., Ybert, J. P., and Mourguiart, P. (1992). A 20,000 years paleohydrological record from Lake Titicaca. In Lake Titicaca: A Synthesis of Limnological Knowledge (C. Dejoux and A. Iltis, Eds.), pp. 4048. Kluwer Academic, Boston. Wirrmann, D., and Mourguiart, P. (1995). Late Quaternary spatio-temporal limnological variations in the Altiplano of Bolivia and Peru. Quaternary Research 43, 344354. Wright, H. E., Mann, D. H., and Glaser, P. H. (1984). Piston corers for peat and lake sediments. Ecology 65, 657659.

Binford, M. W., Brenner, M., and Engstrom, D. E. (1992). Temporal sedimentation patterns in the nearshore littoral of Lago Winaymarka. In Lake Titicaca: A Synthesis of Limnological Knowledge (C. Dejoux and A. Iltis, Eds.), pp. 2937. Kluwer Academic, Boston. Binford, M. W., Kolata, A. L., Brenner, M., Janusek, J., Seddon, M. T., Abbott, M. B., and Curtis, J. H. (1997). Climate variation and the rise and fall of an Andean civilization. Quaternary Research 47(2). Carmouze, J.-P., Arze, C., and Quintanilla, J. (1992). Hydrochemical regulation of the lake and water chemistry of its inow rivers. In Lake Titicaca: A Synthesis of Limnological Knowledge (C. Dejoux and A. Iltis, Eds.), pp. 98112. Kluwer Academic, Boston. Dean, W. E., Jr. (1974). Determination of carbonate and organic matter in calcareous sediments and sedimentary rocks by loss on ignition: comparison with other methods. Journal of Sedimentary Petrology 44, 242248. Fisher, M. M., Brenner, M., and Reddy, K. R. (1992). A simple, inexpensive piston corer for collecting undisturbed sediment/water interface proles. Journal of Paleolimnology 7, 157161. Hakanson, L., and Jansson, M. (1983). Principles of Lake Sedimentol ogy, p. 316. Springer-Verlag, New York. Han, Y. (1995). Stella-Based Simulation of Oxygen and Carbon Isotopic Behavior of Lake Systems. pp. 87106. Unpublished M.S. dissertation, University of Minnesota. Kolata, A. L. (1993). The Tiwanaku: Portrait of an Andean Civilization. Blackwell, Cambridge, MA. Levin, I., Schuchard, J., Kromer, B., and Munnich, K. O. (1989). The Continental European Suess effect. Radiocarbon 31, 431440. Martin, L., Fournier, M., Mourguiart, P., Sifeddine, A., Turcq, B., Absy, L. M., and Flexor, J.-M. (1993). Southern oscillation signal in South America paleoclimatic data of the last 7000 years. Quaternary Research 39, 338346. Mourguiart, P. (1990). Une approche nouvelle du probleme pose par les reconstructions des paleoniveaux lacustres: Utilisation dune fontion de transfert basee sur les faunes dostracodes. Geodynamique 5, 151166. Ortloff, C. R., and Kolata, A. L. (1993). Climate and collapse: Agro-ecological perspectives on the decline of the Tiwanaku State. Journal of Archaeological Science 20, 195221. Quinn, W. H., Neal, V. T., and Antunez de Mayolo, S. E. (1987). El Nino occurrences over the past four and a half centuries. Journal of Geophysical Research 92, 1444914461. Roche, M. A., Bourges, J. C., and Mattos, R. (1992). Climatology and hydrology of the Lake Titicaca basin. In Lake Titicaca: A Synthesis of

a607$$$$64

A few of the residents of Thompson started to talk about fire protection and formed the Thompson Hose Company in 1935. The Thompson Hose Company Incorporated in 1938. The Department started with a 1937 Dodge Truck with a front mount pump built by a local resident Merrill Hathaway. This Truck was replaced by a 1952 Ford with a 500 gallon Darley Pump which at the time cost $10,000 dollars. A few years later a 1959 Jeep was constructed for fighting brush fires. Merrill Hathaway also constructed this vehicle. In the 1960's a second pumper was added to the Company due to a growing population and service to the neighboring townships. A Maxim 1,000n GPM pump was purchased from a Fire Company on Long Island New York. At this time the Company had out grown the present building so an addition was built on the right side of the present building. This building was located on the corner of Main and Jackson Streets in the Borough of Thompson. In the early 70's a 1939 Dodge 1,000 gallon tanker was purchased from the East Main Fire Company in Broome County New York. This was purchased for a $ 1.00 and was put into service in order to a larger water supply to the fleet. In 1974 a new Ford Pumper with a 750 GPM Darley pump and 750 gallon tank replaced the Maxim truck. The cost of this truck was $ 29,000 dollars. In 1974 the Thompson Hose Company purchased a 1975 Chevy Brush truck. They also purchased a 1972 International Tanker from a local milk hauler. The Tanker had a 2200 gallon capacity and was stainless steel in construction. These vehicles replaced the Jeep and the Dodge Tanker. In 1979 the Hose Company purchased a 1980 Dodge Power Wagon with a 750 GPM Hale pump. This Mini Pumper is still in service today.
In 1979 Dr Noyes saw a need for our community to have a ambulance. Dr. Noyes approached the Fire Dept to see if they would be interested in staffing an ambulance for the community. Dr. Noyes bought a seventy something Cadillac Ambulance and donated it to the Department. The Ambulance arrived before all of the staff were trained but the Thompson Hose Co. was well on their way to adding ambulance service to the list of community services. With this addition we had to add on to the building again, this time it was on the left side. This construction was done by a few of the membership. This made a 5 bay fire house, with our meeting room upstairs. We continued to use this ambulance for a few years until we could afford a new ambulance. In 1984 we purchased a Yankee Coach on a Ford Econoline van chassis. At that time we had 15 ambulance crew members and 11 of them were Emergency Medical Technicians (EMT). In 1999 the Yankee Coach was replaced with a 1999 Ford E450 Wheeled Coach modular type ambulance. We are still running with 15 EMT's and 4 First Responders. In 1989 the Company grew again with the addition of a 3,000 gallon Tanker. This was on a Ford chassis and was purchased from Miller Place Fire Company located on Long Island New York. At this time the Company has 2 Pumpers, 2 Tankers, and Brush Truck and 1 Ambulance. In 1993 it was decided to build a new station on Water Street. We moved into this building in summer of 1993. Our new building has a banquet room as well as our engine room. This allowed us to have everything on one floor which made or meeting room handicap assessable. The cost for this project was $ 220,000. In 1995 the Company received a donation from Father Cappelloni of a 1964 Haun Fire Engine. He purchased this engine for $ 1200.00 dollars. This replaced the 1974 Pumper. The Hahn was taken out of service and in December of 2004 and was
purchased by 2 members for $ 1100.00. They formed Thompson Hose Co. Antique Association. In early 2000 we accepted a donation from Mastic Beach Fire Dept of a Ford 900 pumper with a 1000 GPM Waterous Pump This is our present Engine 5. We also purchased a 1993 Western Star Tractor from a Kenworth dealer in Dunmore. We removed the fifth wheel and had a 3500 gallon tank built on this chassis. This Truck is powered by a 425 Caterpillar motor and has a13 speed transmission. We also
had put together a Rescue truck with donations from Columbia Hose Co. and Susquehanna Fire Dept. This rescue truck was short lived at Thompson Hose Co. and was sold to Equinunk Fire dept for a $ 1.00. In 2004 we purchased a new engine for $250.000 dollars. This replaced the Hahn. Our new engine is a 2004 Crimson which is on a Spartan chassis. It is powered by a 400 horse power Cummins diesel. This has a 1250 GPM Hale pump. We had applied for grants for new equipment and was awarded $85,000 for equipment. With this money we purchased new Jaws of Life, 5 in Hose, Radio equipment, and 10 new sets of turnout gear. 2005 we purchased a 1997 GMC pick up truck and converted it to a Brush Truck. The tank was donated from Mastic Beach Fire Department. After a few modifications it was ready for service in Late March. The Brush truck also has a plow frame on it for winter time. 2007 The Thompson Hose Company added a second ambulance. This is a 2007 AEV on a 4 Wheel Drive chassis. The second ambulance is especially helpful when we respond to Motor Vehicle Crashes. The four wheel drive is for the inclement weather that is noted for Susquehanna County.

For current pictures of the Thompson Hose Company please click here. The Thompson Hose Company currently has approximately 50 members. To Date the Thompson Hose Co. has had 11 different Fire Chiefs with Chad Wallace as the current chief. They train regularly and attend classes put on by The Fire Academy to serve our community better. The Ambulance service at the Thompson Hose Co. has had 6 different Captains with Kim Conklin as the current Captain. Our EMT's need 24 hours of continuing education to remain certified. Some of the Fire Chiefs and Ambulance Captains have been in charge of our company before, taken a break and then been reassigned as chief or captain.