ARTICLE IN PRESS

Quaternary International 161 (2007) 4–21

A 14 kyr record of the tropical Andes: The Lago Chungara´ sequence (181S, northern Chilean Altiplano) A. Morenoa,, S. Giraltb, B. Valero-Garce´sa, A. Sa´ezc, R. Baod, R. Pregoe, J.J. Pueyoc, P. Gonza´lez-Sampe´riza, C. Tabernerb a Pyrenean Institute of Ecology—CSIC, Apdo 202, 50080 Zaragoza, Spain Institute of Earth Sciences ‘Jaume Almera’-CSIC, C/Lluı´s Sole´ i Sabarı´s s/n, 08028 Barcelona, Spain c Faculty of Geology, University of Barcelona, C/Martı´ Franque´s s/n, 08028 Barcelona, Spain d Faculty of Sciences, University of A Corun˜a, Campus da Zapateira s/n, 15071 A Corun˜a, Spain e Department of Marine Biochemistry, Marine Research Institute, CSIC, C/ Eduardo Cabello 6, 36208 Vigo, Spain b

Abstract High-resolution geochemical analyses obtained using an X-ray fluorescence (XRF) Core Scanner, as well as mineralogical data from the Lago Chungara´ sedimentary sequence in the northern Andean Chilean Altiplano (181S), provided a detailed reconstruction of the lacustrine sedimentary evolution during the last 14,000 cal. yr BP. The high-resolution analyses attained in this study allowed to distinguish abrupt periods, identify the complex structures of the early and mid-Holocene arid intervals and to compare their timing with Titicaca lake and Sajama ice records. Three main components in the lake sediments have been identified: (a) biogenic component, mainly from diatoms (b) volcanics (ash layers) from the nearby Parinacota Volcano and (c) endogenic carbonates. The correlation between volcanic input in Lago Chungara´ and the total particles deposited in the Nevado Sajama ice core suggests the Parinacota Volcano as the common source. The geochemical record of Lago Chungara´ indicates an increase in siliceous productivity during the early Holocene, lagging behind the rise in temperatures inferred from the Nevado Sajama ice core. The regional mid-Holocene aridity crisis can be characterized as a number of short events with calcite and aragonite precipitation in the offshore lake zones. r 2006 Elsevier Ltd and INQUA. All rights reserved.

1. Introduction Recent research of past climate oscillations has found that changes between climate modes during the Holocene occurred within decades (Mayewski et al., 2004), a period of time similar to more recent climate changes (Houghton et al., 2001). In this context, scientific efforts over the last few years have been directed towards understanding the timing and mechanisms of abrupt climate changes during the last millennia. Despite the recent increase in the number of high-resolution paleoclimate records from low latitudes (e.g. Hughen et al., 1996, 2004; Kuhlmann et al., 2004), the role of the tropics in abrupt Holocene climate changes remains a matter of debate. Tropical South America exemplifies the complexity of Holocene climate reconstructions, in which high-resolution terrestrial records, essential Corresponding author. Tel.: +34 976 716118; fax:+34 976 716019.

E-mail address: [email protected] (A. Moreno).

to any examination of rapid climate fluctuations, are scarce, with diverse proxy records showing numerous discrepancies (i.e. Grosjean, 2001). Detailed knowledge of the distribution and amplitude of abrupt climate changes in tropical latitudes of the Andean Altiplano is still sparse and the processes responsible for climate variability at different temporal and regional scales are barely understood. The suggested close link between higher lake levels in the Andean Altiplano and cold sea surface temperatures in the Equatorial Atlantic (i.e., Heinrich events, Younger Dryas, 8.2 kyr event or the Little Ice Age) indicated by the Titicaca lake record (Baker et al., 2001a) requires additional records from tropical South America to confirm this paleoclimate teleconnection between the two hemispheres. The evolution of temporal and spatial moisture patterns during the Holocene is one of the main controversies surrounding studies of South American paleoclimate. It has been generally accepted that the northern-central Andes were a generally arid region from
7 to 4 kyr BP

1040-6182/$ - see front matter r 2006 Elsevier Ltd and INQUA. All rights reserved. doi:10.1016/j.quaint.2006.10.020

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

5

as observed in lacustrine (Abbott et al., 1997; Baker et al., 2001a; Grosjean et al., 2003; Paduano et al., 2003; Tapia et al., 2003) and ice-core records (Thompson et al., 1998; Thompson et al., 2000). This hypothesis is also supported by archeological evidence (Nu´n˜ez et al., 2002). In addition to moisture reconstructions, a recent study of long-chain alkenones from Titicaca lake sediments also points to enhanced regional temperatures during the mid-Holocene (Theissen et al., 2005). However, other recent studies support a more complex spatial and temporal pattern, and even periods of increased humidity during the midHolocene (Holmgren et al., 2001; Latorre et al., 2003; Servant and Servant-Vildary, 2003). Paleoclimate sedimentary records possessing a robust and accurate chronological framework are therefore imperative to understanding both the regional significance and the timing of abrupt humidity changes detected during the mid-Holocene. The overall goal of this study was to document the regional pattern of climatic change for the last 14,000 cal. yr BP using a sedimentary record from Lago Chungara´ (Andean Altiplano, 181S). This paper reports a high-resolution geochemical record from the lake obtained by an X-ray fluorescence (XRF) core scanner together with other paleoenvironmental indicators (i.e. physical properties, mineralogy, opal content and total organic carbon (TOC)). The resulting high-resolution analyses, in tandem with a multi-proxy approach, allowed us not only to infer the paleoclimate signal from the Lago Chungara´ record, but also to contribute to the identification, correlation and understanding of abrupt climate change during the Holocene in tropical regions of South America. 2. Location, climate and limnology of Lago Chungara´ Lago Chungara´ (181150 S, 691100 W, 4520 m asl) is located in the highest and westernmost fluvio-lacustrine basin in the Andean Altiplano (Northern Chile, Fig. 1a). This lake sits in the central part of a small hydrologically closed subbasin at the northeastern edge of the Cenozoic Lauca Basin. The intense volcanic activity and, to a lesser extent, the movement of synsedimentary faults are significant factors for sedimentation in the Chungara´ subbasin. The Lago Chungara´ subbasin was formed after the collapse of the Parinacota Volcano (Fig. 1a), which produced a huge debris avalanche blocking the Paleo-Lauca River at about 15–17 kyr BP (Wo¨rner et al., 1988; Wo¨rner et al., 2000; Wo¨rner et al., 2002). However, the age of this collapse is controversial, and it has been estimated at 8 kyr by other authors (Clavero et al., 2002, 2004). The local vegetation is dominated by tussock-like grasses, shrubs, Polylepsis, a dwarf tree of the Rosaceae family, as well as extensive soligenous peatlands (‘‘bofedales’’) (Schwalb et al., 1999; Earle et al., 2003). Lago Chungara´ is climatically located in the arid Central Andes. This region is dominated by tropical summer moisture stemming from the Amazon Basin, and is controlled by the southward migration of the subtropical

Fig. 1. (a) Location of Lago Chungara´ and other paleoclimatic records on the Northern Chilean Altiplano. (b) Position of sediment cores in Lago Chungara´. Isobaths and main inflows are indicated.

jet stream as well as the intensification of the Bolivian highpressure system (Garreaud, 2001; Garreaud et al., 2003). Average annual rainfall in the region is about 350 mm. A significant fraction of the inter-annual variability of summer precipitation is currently related to the El Nin˜o Southern Oscillation (ENSO) (Vuille, 1999). Thus, wet summers on the Andean Altiplano are associated with an ENSO-related cooling of the tropical Pacific (La Nin˜a phase). Lago Chungara´ has a maximum water depth of 40 m, a surface area of 22.5 km2 and a volume of about 426  106 m3 (Risacher et al., 2003; Herrera et al., 2006). The main inflow is the Chungara´ River (300–460 l s1) and several springs on the western margin. Although there is no surface outlet, groundwater outflow was estimated as 0.2 m3 s1 (Montgomery et al., 2003) and the water lost by potential evaporation measuring about 1230 mm yr1 (Risacher et al., 2003). The lake is polymictic, meso to eutrophic and contains 1.3 g l1 total dissolved solids

ARTICLE IN PRESS 6

A. Moreno et al. / Quaternary International 161 (2007) 4–21

(Mu¨hlhauser et al., 1995). Water chemistry is of Na–Mg–HCO3–SO4 type with an average pH of 9. At present, oxic conditions extend to the lake bottom. Primary productivity in the lake is mainly governed by diatoms and chlorophyceans. During four sampling periods from 1998 to 1999, biomass values fluctuated from 0.34 to 8.74 mg Chlorophyll a l1 (Dorador et al., 2003). Oscillations in both, phytoplanktonic biomass and phytoplanktonic community structure seem to be mainly due to changes in water column temperature and salinity. 3. Materials and methods In November 2002, 15 sediment cores (6.6 cm inner diameter and up to 8 m long) were retrieved from Lago Chungara´ along the NW–SE and NE–SW transects of the lake (Fig. 1b). The core locations were selected to sample the different depositional environments using a modified Kullenberg piston corer from the Limnological Research Center, University of Minnesota (LRC). The retrieved cores were shipped to the LRC where physical properties (GRAPE-density, p-wave velocity and magnetic susceptibility) were non-destructively measured every cm using a GEOTEKTM Multi-Sensor Core Logger (MSCL). The cores were then split in two halves, scanned for color pictures using a DMT CoreScan digital color line scan camera system, and macroscopically described in terms of

texture, color and sedimentary structures (Core 11, Fig. 2). Smear slides were described using a Nikon polarizing microscope to estimate the biogenic, clastic and endogenic mineral content of the defined sedimentary facies. Subsamples were taken every 5 cm for mineralogical, chemical and biological analyses. After a detailed lithological correlation of all cores (Fig. 3 and Sa´ez et al., 2007), cores 10 and 11 were selected for paleoclimatic and paleoenvironmental reconstructions. Both cores recorded almost the entire sedimentary infill of the offshore zone, allowing reconstruction of a composite sequence. Total carbon (TC) and total inorganic carbon (TIC) contents were determined by a UIC model 5011 CO2 Coulometer, with TOC content then calculated. Samples for X-ray diffraction (XRD) were dried at 60 1C over 24 h and manually ground using an agate mill. XRD analyses were performed using an automatic Siemens D-500 X-ray diffractometer: Cu ka, 40 kV, 30 mA and graphite monochromator. Identification and quantification of the different mineralogical species present in the crystalline fraction were carried out following a standard procedure (Chung, 1974). The area of the amorphous fraction was calculated as total counts using the XRD software. The sample that showed the highest amorphous area, formed mostly by diatomaceous oozes, was progressively mixed with increasing quantities of pure calcite (5%, 10%, 20%, 40% and 60% of the total weight of the sample) and a logarithmic

Fig. 2. Digital image, lithological column and magnetic susceptibility profile from core 11. Key layers M-1 to M-11 and available 14C dates for this core (not calibrated) are indicated (see Table 2). A detailed facies legend is shown. See text for explanation.

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

7

Fig. 3. NW–SE stratigraphic cross-section including the six long cores retrieved from Lago Chungara´ (cores 12, 13 and 15 are projected as indicated in the figure). Stratigraphic correlations are based on lithostraphic and sedimentological criteria (limits between units, key levels M-1 to M-11) and magnetic susceptibility profiles. A small fault detected from seismic profiles (Sa´ez et al., 2007) is also shown. WAF: White ash flow.

function was adjusted. This function allowed expressing the area of the amorphous fraction as a percentage of the total sample weight. Core 11 was analyzed for opal (biogenic silica) content following an alkaline leaching technique (Mortlock and Froelich, 1989). After leaching, the dissolved silica concentration of the resulting extract was measured by the molybdate blue colorimetric method (Hansen and Grashoff, 1983) using an AutoAnalyser Technicon II. In addition to MSCL core logging methods, the XRF core scanner was applied. XRF data used in this study were produced by the new-generation XRF core scanner at the University of Bremen. The core archive halves were measured with 2 mm resolution for light (Al, Si, S, K, Ca, Ti, Mn and Fe) and 1 cm resolution for heavy (Sr, Zr, Sn and Ba) elements. The measurements were produced using 60 s count time, 10 kV X-ray voltage (50 kV for heavy elements) and an X-ray current of 1 mA to obtain statistically significant measurements. The analyzed sections conformed to a composite sequence of cores 10 and 11 (Fig. 3). A detailed description of the applied XRF analysis and system configuration of the XRF core scanner at the University of Bremen are provided by Jansen et al. (1998) and Ro¨hl and Abrams (2000). The data obtained by the XRF core scanner are expressed as element intensities in counts per second (cps). A comparison with other methods employed to convert the XRF core scanner measurements to absolute

elemental concentrations failed. Sixty samples were analzsed by inductively coupled plasma-optical emission spectrometry (ICP-OES, Perkin Elmer Optima 3300 RL) and 30 samples by powder XRF at the University of Bremen (Portable Energy Dispersive Polarization X-ray Fluorescence analyzer, Spectro Xepos). Although there was a high correlation between both methods and the scanner data for some elements, the overall correlation was not of sufficient quality to calibrate all elemental data. Consequently, and consistent with most published studies (e.g. Jansen et al., 1998; Ro¨hl and Abrams, 2000), the original XRF core scanner data are expressed in cps. Finally, the statistical treatment of the dataset was performed using the R software package (R Development Core Team, 2004). The AMS 14C dates from the Chungara sediments were obtained from (1) bulk organic matter from the central plain cores and (2) aquatic organic macrorests picked from littoral cores (Table 2). Several carbonate samples were also dated with U/Th techniques. 4. Results and interpretation 4.1. Lithostratigraphy Most facies in Lago Chungara´ sediments (Fig. 2) are massive or laminated diatomaceous ooze (A, B, C, D, E), while carbonate-rich facies occur in thin layers or laminae

ARTICLE IN PRESS 8

A. Moreno et al. / Quaternary International 161 (2007) 4–21

(F). Volcaniclastic facies (G and H) are particularly abundant in the upper part of the cores. The presence of these tephra layers and their continuity along the basin allows a detailed correlation of the retrieved cores. Tephra layers are labelled from M-1 to M-11 (from bottom to top) and correspond to peaks in magnetic susceptibility (Figs. 2 and 3). Additionally, the presence of a white ash-flow (WAF), in spite of the low magnetic susceptibility signal, was used for correlation since it occurs in all cores. From stratigraphical correlation and seismic stratigraphy (Sa´ez et al., 2007), two main lithostratigraphic units were defined. Unit 1 was deposited after the volcanic event that created the lake and is composed of laminated diatomaceous ooze (Facies A, B and C). Only one 2 cm thick glass-dominated tephra layer (M1) occurs in Unit 1 (Figs. 2 and 3). These lamination sets comprise rhythms and cycles of different mm-sized layers (average 2.6 mm) consisting of diatomaceous ooze with variable types (calcite, aragonite) and quantities of carbonates and amorphous organic matter. Unit 1 is divided in two subunits: Subunit 1a with green and white laminations and no carbonate (Facies A) and Subunit 1b with brownish to white laminations (Facies B) where endogenic carbonates occur in low concentrations. Carbonates occur in the lighter laminae (Facies F). Towards the upper part of Subunit 1b, laminated facies (Facies B) alternate with intervals of dark green, massive, organic-rich diatomaceous ooze (Facies C). Facies C layers are continuous throughout the basin and were detected in all cores from the central plain (lithological keybeds in Fig. 3). Unit 2 is mainly composed of massive to slightly banded diatomaceous ooze (Facies D and E) frequently intercalated by tephra layers (coarse- and fine-grained ashfalls, Facies G and H). This unit is also divided into two subunits. Subunit 2a is composed of brownish-red massive to slightly banded sapropelic diatomaceous ooze (Facies D) with common calcite crystals (silt grain-sized) and carbonate-rich layers (Facies F). Facies F (25–55% carbonate) represents up to 5% of the total thickness of Subunit 2a. Specifically, carbonate-rich layers are grouped in discrete whitish to pinkish thin layers, which are composed of calcite (fibrous crystals, fusiform aggregates, rice-shaped and euhedral crystals), magnesium calcite, needlelike aragonite and some traces of dolomite. Tephra layers are mainly composed of Ca-rich feldspars and volcanic glass. In particular, the presence of three lapilli layers (Facies G) is highlighted since their lateral continuity allowed considering them as key layers for correlating the cores (M-4–M-6 in Fig. 3). Subunit 2b consists of dark gray diatomaceous ooze with frequent macrophyte remains (Facies E) alternating with massive black tephra layers, mainly composed of plagioclase, glass and mafic minerals (Facies H). Volcaniclastic deposits represent 50% of the total thickness of this unit (levels from M-8 to M-11 in Figs. 2 and 3). A 1-cm-thick rhyolitic WAF in (Figs. 2 and 3) occurs in all cores and is used for correlation.

4.2. Mineralogical composition The sediments of Lago Chungara´ are mainly composed of two fractions: one crystalline (highlighted by sharp diffraction peaks), the other amorphous (characterized by the presence of a broad peak centered between 201 and 251 2y angles). In accordance with the previously described calculations, the amorphous component was quantified. The amorphous fraction percentages range from 40% (in volcaniclastic-rich deposits of Unit 2) to almost 100% (in Unit 1). This fraction represents organic matter, amorphous silica (from diatoms) and volcanic glass. On the other hand, the percentages of the total crystalline fraction range between 60% and less than 1% of the total weight of the samples, with this fraction composed of Ca-plagioclase, carbonates (calcite and dolomite), biotite, pyrite, quartz and amphibole. In Fig. 4 the main minerals from Lago Chungara´ sediments are represented versus the composite depth. From the bottom to the top of the sequence, and based on the dominance of the fraction and the mineral species, three zones were defined. These zones broadly corresponded to the lithostratigraphic units (Fig. 4). Unit 1 is dominated by amorphous material (more than 95% of the total weight of the samples). Plagioclase, quartz and pyrite are the main mineral species that comprise the crystalline fraction (o 5%). The presence of silt-sized calcite crystals is the main characteristic of Subunit 2a. From a depth of 350 cm upwards, the plagioclase percentages start to increase whereas the amorphous fraction percentages decrease. This change represents the increase of volcanic activity in the nearby Parinacota Volcano. Subunit 2b ranges from a depth of 100 cm to the top of the sequence. This subunit is characterized by constant, as well as the highest, percentages of plagioclase minerals. Plagioclase, amphibole, quartz and biotite minerals are related to the erosion of catchment rocks and the synsedimentary direct input from eruptions of the Parinacota Volcano. 4.3. Geochemical composition The XRF Core Scanner provided a record of geochemical variations for the Chungara´ sediments. A composite record joining the analyzed sections from cores 10 (Sections 1–5) and 11 (Sections 4–6) was generated to cover almost the entire sedimentary sequence (Fig. 5). Although above this sedimentary sequence up to 95 cm of the Chungara´ topmost sediments are recorded in other cores (e.g. core 14) (Fig. 3 and Sa´ez et al., 2007), we regarded the top (0 cm) of the studied sequence as the top of core 10 for simplification. The downcore profiles of heavy and light elements clearly delineated three different units in terms of their geochemical composition: an upper tephra-rich unit with maximum values in all elements (Subunit 2b), an intermediate unit with the highest Ca values (Subunit 2a) and a lower unit characterized by the highest Si content (Unit 1).

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

Amorphous (%)

Dolomite (%)

Calcite (%)

Plagioclase (%)

Amphibole (%)

Quartz (%)

9

Pyrite (%)

Biotite (%)

Fig. 4. Mineralogical profiles measured by X-ray diffraction (XRD) for the composite sequence of Lago Chungara´ in percentages versus depth. Lithological subzones are indicated by dashed lines.

Fig. 5. Light and heavy elements measured by the XRF Core Scanner for the composite sedimentary sequence of Lago Chungara´. All the measurements are in counts per second (cps). Lithological units are indicated.

ARTICLE IN PRESS 1.00 1.00 0.27 1.00 0.26 0.97 1.00 0.94 0.30 0.94 Correlations that are in plain text (not bold style) are not significant after application of the Bonferroni test (see text). MS refers to magnetic susceptibility.

1.00 0.86 0.94 0.17 0.89 1.00 0.90 0.87 0.94 0.23 0.89 1.00 0.27 0.19 0.24 0.24 0.06 0.21 1.00 0.26 0.98 0.89 0.89 0.96 0.25 0.93 1.00 0.20 0.18 0.20 0.14 0.48 0.20 0.20 0.23 1.00 0.19 0.95 0.19 0.91 0.88 0.87 0.92 0.26 0.93 1.00 0.27 0.06 0.33 0.00 0.41 0.36 0.17 0.27 0.30 0.23 1.00 0.60 0.08 0.12 0.02 0.23 0.06 0.10 0.08 0.01 0.37 0.00 1.00 0.32 0.44 0.90 0.32 0.83 0.11 0.80 0.78 0.80 0.79 0.12 0.81 1.00 0.19 0.06 0.27 0.27 0.48 0.24 0.14 0.22 0.25 0.02 0.20 0.09 0.17 1.00 0.85 0.17 0.02 0.18 0.23 0.28 0.20 0.11 0.19 0.21 0.07 0.17 0.13 0.14 1.00 0.29 0.31 0.68 0.07 0.01 0.79 0.14 0.78 0.16 0.74 0.74 0.76 0.79 0.33 0.78 1.00 0.63 0.17 0.11 0.43 0.30 0.06 0.59 0.19 0.64 0.31 0.65 0.62 0.64 0.64 0.25 0.59 1.00 0.18 0.09 0.17 0.59 0.08 0.23 0.27 0.12 0.47 0.09 0.11 0.08 0.13 0.11 0.08 0.10 0.08 1.00 0.04 0.87 0.72 0.22 0.27 0.47 0.19 0.02 0.62 0.07 0.68 0.29 0.69 0.65 0.62 0.68 0.23 0.61 1.00 0.88 0.42 0.87 0.60 0.10 0.04 0.38 0.28 0.11 0.50 0.28 0.57 0.30 0.58 0.53 0.61 0.57 0.26 0.52 Amorphous Plagioclase Calcite Opal MS TOC TC Al Si S K Ca Ti Mn Fe Rb Sr Zr Sn Ba

Sn Zr Sr Rb Fe Mn Ti Ca K S Si Al TC TOC MS Opal Calcite Plagioclase Amorphous

4.3.1. Correlation analyses and definition of main sediment components Statistical correlation analysis was performed in order to discern the similarities among the entire set of proxies (Table 1). The sampling interval of the different proxies was not identical (e.g. 2 mm for XRF and 5 cm for XRD). Subsequently, to highlight the main coarse relationships between the dataset, all proxies were linearly interpolated with a regular spacing of 5 cm, resulting in a dataset of 17 variables (proxies) and 171 cases (samples). The significance of the correlation analysis (p-values) was calculated and the p-values adjusted by applying the Bonferroni test (R Development Core Team, 2004). Many of the significant correlations among the variables have low values, highlighting that they do not have univocal relationships, since these variables could have more than one origin. As an example, Ca has a correlation with calcite and TC contents only around 0.48. This is explained by the volcanic source of Ca in the upper part of the sequence (Ca-plagioclase), as well as the sedimentary source of Ca in the middle part of the sequence (endogenic carbonates). Therefore, although the three main components of Chungara´ sediments discussed above (siliceous biogenic, volcanic and carbonates) can be shown via the higher correlation values among the related proxies (Table 1), several patterns arise when the records are examined more closely. Al, K, Ti and Fe are associated with the allochthonous component (here, volcanic), increasing with the presence of tephra layers as observed by lithology; heavy elements like Zr, Sr, Sn and Ba are related to tephra layers as well. On the other hand, while Ca behaved in similar fashion to the previous group of volcanic-originated elements along the upper part of the sedimentary succession, it began to behave independently starting from the uppermost carbonate-rich layer (190 cm in depth) and extending towards the bottom of the sequence (Fig. 5). Accordingly, Ca behaved similar to Al, K, Ti and Fe from 0 to 190 cm depth, while this element exhibited a distinctly different pattern from 190 cm to the bottom of the sequence (Fig. 5). The best explanation is the increasing abundance of Ca in the composition of volcanic minerals such as plagioclase, which dominates volcanic eruptions, thereby influencing Chungara´ sediments mainly in the upper part of the record. The Ca profile changes drastically in Subunit 2a: Ca appears as very sharp spikes instead of smooth peaks common in Subunit 2b. Ca content in the lower part of the record is very low but it is still possible to recognize abrupt Ca spikes associated with some white laminae. In the same manner as Ca, Si is influenced by different components, with volcanic and biogenic silica (diatoms) being the most important. The Si signal thus represents a mixture of both influences. Other elements, as S or Mn, are not strongly related to any of the described patterns. Therefore, the S profile does not closely mimic any other profile, but shows a slight correspondence with Si values (0.6 correlation in Table 1) indicating that this pattern most likely reflects an

Ba

A. Moreno et al. / Quaternary International 161 (2007) 4–21

Table 1 Correlation coefficients among different proxies used in this study

10

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

interference of influences and origins. In the same way, Mn values mark the occurrence of tephra layers and correlate with some Ca peaks. The presence of Mn in these sediments is related to changes in redox conditions at the sediment–water interface since Mn forms a highly insoluble oxide where oxic conditions prevail. Thus, solid phases of Mn in lake sediments appear as a result of the upward diffusion of dissolved Mn and its posterior precipitation at oxic horizons in the form of Mn oxides (Aguilar and Nealson, 1998). In Chungara´ sediments, Mn background is very low, usually below 0.5 g kg1 (1000 cps), although it increases to 2.5 g kg1 (5000 cps) during discrete peaks that are more abundant during Subunit 1b. This behavior may indicate a close relation with changes in the sedimentary oxygen content most likely related to biogenic processes or changes in water circulation due to lake-level variations. The strong volcanic influence in the geochemical composition of Chungara´ sediments makes it necessary to normalize the elemental data to ‘‘volcanic’’ elements like Ti to unravel the environmental or climatic meaning of the different profiles. Such high-resolution profiles constrain the evolution of the three main inputs contributing to the Chungara´ record: siliceous biological remains (mainly diatom skeletons), volcanic minerals and endogenic carbonates. 4.3.2. Biogenic component In Fig. 6, the Si/Ti ratio is represented to discern the biogenic silica amount compared to the percentage of opal

11

extracted by alkaline digestion, as well as with the amorphous fraction obtained from XRD. In addition, TOC and S/Ti profiles are plotted to infer the variability of the organic matter content in Lago Chungara´ since the S content reflects the occurrence of pyrite, which generally increases when organic matter content is high. The Mn/Ti ratio is plotted in the same figure. From a comparison of the whole dataset (Fig. 6), some striking similarities can be observed. Thus, the Si/Ti, as well as the opal and amorphous profiles follow similar trends at some intervals: low values in the upper part tend to increase from 300 to 550 cm and remain at intermediate levels until the bottom of the sequence. These results point to a higher lake diatom productivity (or better, silica preservation) in the laminated deposits of Subunit 1b and, to a lesser extent, in Subunit 1a deposits (Fig. 6). In addition, the Si/Ti record helps to identify intervals where the biogenic lacustrine signal is dominant over volcanic layers. This pattern is not so clear in opal or amorphous profiles because of lower spatial resolution. The S/Ti, TOC and Mn/Ti data require careful interpretation (Fig. 6). Microscopic observation of smear slides and SEM-EDS determinations (Sa´ez et al., 2007) show the presence of pyrite associated with organic-rich sediments. Thus, the S/Ti ratio could reflect variations in pyrite that could be correlated to TOC. Pyrite was detected by XRD although always at very low percentages (Fig. 4). However, the very low values of S obtained in some samples by XRF (below 2 g kg1) prevent an accurate comparison among

Fig. 6. Biogenic influences on Lago Chungara´ sediments: Si/Ti ratio, biogenic opal, amorphous minerals obtained by XRD, S/Ti ratio, TOC and Mn/Ti ratio. All proxies are plotted versus core depth (cm), lithological subunits are indicated.

ARTICLE IN PRESS 12

A. Moreno et al. / Quaternary International 161 (2007) 4–21

these records. In addition, the TOC record obtained from upper Chungara´ sediments is highly variable due to the frequent macrophyte remains and volcanic input. Organic matter content increases from the bottom of the sequence (4%) to the top of laminated deposits of Subunit 1b, where it reaches 8%. The maximum in Mn/Ti is associated with the maximum in organic matter and diatomaceous productivity (Subunit 1b). Although an accurate explanation requires an understanding of diatom assemblages during this interval, since diatoms contribute about 80% of the sediment, we would suggest changes in redox conditions as a significant factor in Mn variability. More oxygen in the water column caused either by photosynthetic processes (related to maximum diatom productivity) or increased bottom water ventilation would account for the increase in Mn. In such a scenario, Mn would be precipitated as oxides and then, preserved in sediments. 4.3.3. Volcanic components Elements directly related to the volcanic influence in the Chungara´ sequence are compared with magnetic susceptibility and volcanic minerals, such as plagioclase or amphibole (Fig. 7). It is evident that Unit 1 deposits are almost free of volcanic input, while Unit 2 is undoubtedly dominated by tephra ashfalls. Mineralogical results indicate that the dominant volcanic mineral is plagioclase, likely andesine, with percentages ranging from 5% to 60%.

Values of Fe, Ti and K are excellent indicators of volcanic deposits in the Lago Chungara´ record (Table 1, Fig. 7). 4.3.4. Carbonate components To investigate carbonate production variability, the Ca/ Ti ratio versus TIC and calcite percentages are plotted (Fig. 8). In addition, the Sr/Ti ratio was plotted to verify whether the Sr variations were related to carbonate precipitation (i.e. in aragonite) or to volcanic input. The high correlation between Ca/Ti and Sr/Ti confirms the relation of both elements (Ca and Sr) to the carbonate precipitation (Fig. 8). Normalization with respect to Ti is a valuable tool for unravelling previously undetected patterns. The high-resolution record shows that carbonate in the offshore central plain of the lake was produced during very short intervals and deposited as thin layers. This spiky character of carbonates was hidden when working with discrete samples at lower sampling resolution (see calcite and TIC profiles, Fig. 8). It is also evident that carbonate deposits are concentrated in Subunit 2a, thus confirming the presence of a carbonate-rich interval (250–450 cm), reaching up to 40% of carbonate. Microscopic (optical and scanning) observation of these carbonate layers indicates that they are mainly composed of calcite, in the form of rice-shaped or euhedral crystals. These types of carbonate crystals are formed in the offshore central plain epilimnium (without any transport from the vegetated littoral areas or

Fig. 7. Volcanic influences on Lago Chungara´ sediments: Fe, Ti, K, plagioclase and amphibole obtained by XRD and magnetic susceptibility (SI units). All proxies are plotted versus core depth, lithological subunits are indicated and the correlation lines (M-1 to M-11) marked by arrows (see also Fig. 3).

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

13

Fig. 8. Carbonate in offshore zones at Lago Chungara´: Ca/Ti ratio, Sr/Ti ratio, calcite obtained by XRD and TIC. All proxies are plotted versus core depth, lithological subunits are indicated.

any relation to organisms, such as bivalves or gastropods). Two possible origins of these calcite crystals are: (1) they are linked to algal blooms capable of altering water CO2 concentration (Kelts and Hsu¨, 1978; Teranes et al., 1999) or (2) they are associated with alterations in water chemistry, either of volcanic influence or from changes in water concentration. The apparent absence of correlation between the long-term evolution of Ca/Ti (carbonates) and of Si/Ti (mainly diatoms) allowed us to propose that the productivity blooms are not the main promoter of the carbonate production. This relation seems to be true only in carbonate layers where rice-shaped carbonate crystals are present (Table 1, Figs. 6 and 8). On the other hand, a volcanic influence cannot be totally discarded since Subunit 2a is characterized by the presence of the thickest lapilli layers along the sequence (Fig. 2). However, calcite precipitation starts earlier (in Subunit 1b) than does volcanic activity, which calls into question volcanic influence as the main causative force (Fig. 8). Therefore, the most likely mechanism stems from changes in water concentration. In current climatic and limnological conditions, water concentration is mainly related to the

precipitation–evaporation balance (Valero-Garce´s et al., 2003). More evaporation or less precipitation (arid period) would imply more concentrated waters and lead to precipitation of offshore carbonates. 4.4. Chronology Dating the sediments from Lago Chungara´ is complicated due to the scarcity of terrestrial organic rests resulting from the low vegetation cover. Therefore, AMS 14 C dates from Chungara´ sediments were obtained from (1) bulk organic matter from the central plain cores and (2) aquatic organic macrorests picked from littoral cores (Table 2). In Fig. 3, the correlation panel with the cores used to construct the chronological model is presented. The similarities in the sedimentary facies among the cores and the presence of key tephra layers allow transferring the obtained dates to a single composite depth (Table 2). As observed in Table 2, there are five ages that were discarded (shaded samples) and not included in the final age model. Two of them are clearly reversed (15A-5, 76 and 13A-4, 66), probably due to depositional reworking of the

14

Table 2 14 C AMS radiocarbon ages measured in Chungara cores Units

Composite depth (cm)

Laboratory ID

Sample (core and depth)

Type of sample

Uncalibrated 14C age (2s) (yr BP)

Calibrated age (cal. yr BP) after Heegaard’s method

Beta 188745

DIC measured in surface water

2320740

Subunit 2b

42 67 95

Poz-8720 AA56904 Poz-8721

11A-2, 39 15A-2, 48 11A-2, 84

Bulk organic matter Aquatic organic macroremains Bulk organic matter

4850740 6635739 7290750

Subunit 2a

258 344 436

Poz-8723 AA56903 Poz-8724

11A-3, 2 15A-4, 27 11A-3, 86

Bulk organic matter Aquatic organic macroremains Bulk organic matter

8920750 9999750 10,860760

Subunit 1b

490 550 615

Poz-7170 Poz-8647 Poz-7171

11A-3,123 11A-4, 10 11A-4, 63

Bulk organic matter Bulk organic matter Bulk organic matter

8570750 9860760 11,070770

Subunit 1a

665 675 697 743 785 827 865

AA56905 Poz-8725 Poz-11891 Poz-13032 Poz-11982 Poz-13033 Poz-7169

15A-5,76 13A-4, 66 11A-4,145 11A-5, 41 11A-5, 84 11A-6, 41 11A-6. 79

Aquatic organic macroremains Bulk organic matter Bulk organic matter Bulk organic matter Bulk organic matter Bulk organic matter Bulk organic matter

43857100 8810750 11,460760 10,950780 11,180770 12,120780 13,100780

28657900 44407500 59707600

—

—

28657900 44407500 59707600

12.97 0.4 25.46 14.870.2 16.170.1 No data 16.970.1

7325790 83807450 92807690

974571000 1011071190 1077571000 1172071225 127107800

9550790 114507370 127707390

16.870.1 14.170.3 13.670.2

130007420 130507420 134107410 140957570 148757910

No data 22.9 16.270.4 22.772.3 28.773.7 19.671.7 23.170.2

Depth transposed to composite sequence is indicated. Lithological subunits are also shown. Italic samples were not used for the age model construction (reversed dates, error too high). DIC: dissolved inorganic carbon (apparent age of surface water). Model 1: constant reservoir age correction; Model 2: no reservoir age correction for Unit 1. See text for details about calibration procedures and Fig. 9.

ARTICLE IN PRESS

0

Model 2

A. Moreno et al. / Quaternary International 161 (2007) 4–21

Model 1

d13C (%)

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

sediments, since one sample comes from a volcanic sandy layer and the second from a reworked peaty interval. The other three samples (11A-3, 2, 15A-4, 27 and 11A-3, 86) belong to the volcanic-rich Subunit 2a (Fig. 2) and do not show evidence of depositional reworking. One possibility is that the strong volcanic influence may have altered the CO2 balance among the different sources (atmosphere, soil, runoff and groundwater, volcanic), thus modifying the 14 C value of these three samples. A similar effect has been documented in other lakes from the Altiplano (Valero-Garce´s, et al., 1999, 2000) and Easter Island (Butler et al., 2004) with high volcanic influence. Obtaining reliable radiocarbon dates in the sediments of Lago Chungara´ is also problematic because of an assumed large and variable radiocarbon reservoir effect. Similar problems occur in the majority of lake deposits from the Andean Altiplano, with very few studies able to properly evaluate the variations in reservoir effect over time (Geyh et al., 1999; Geyh and Grosjean, 2000; Grosjean et al., 2001). The modern reservoir effect for Lago Chungara´ was obtained by dating the dissolved inorganic carbon (DIC) of the surface lake water. This resulted in 2320740 14C yr BP (Table 2), a value very similar to that obtained by Geyh et al. (1999) after analyzing DIC lake water (17457160 14C yr) and living macrophytes (25607245 14C yr) in Lago Chungara´. However, the reservoir effect in the Altiplano lakes has proved to be highly variable over time, with the influence of lake water volume being one of the most significant factors (Geyh and Grosjean, 2000). Since no terrestrial organic remains were found in Lago Chungara´ sediments, we could not evaluate the variation of the reservoir effect with time applying the methodology of Geyh and Grosjean (2000). Therefore, our approach to correct the dates for the variable reservoir effect has been based on two assumptions: (1) the Lago Chungara´ system during deposition of Subunit 2b is very similar to that currently found there and (2) the present-day lake level is at the highest in its history. Accordingly, the correction of the reservoir effect was achieved differently in Unit 1 and Unit 2. It is worth noting the very different sedimentary patterns recorded when comparing Units 1 and 2: while the latter is composed of rather homogenous facies (dark gray massive diatomaceous oozes in Subunit 2b) characterized by abundant

15

volcanic layers, the former represents a clearly distinct stage of the lake system (laminated sediments, no volcanic input). This pattern may allow us to consider a constant reservoir effect for the uppermost subunit, since the average lake characteristics (depth, water volume) most likely did not vary much over time. Therefore, a constant reservoir effect of 2320 years (the present-day one) is subtracted from the Subunit 2b dates. After removing the reversed ages, no 14C AMS data were available from Subunit 2a (Table 2). It is only possible to hypothesize about the variations over time of the reservoir effect in Unit 1. In accordance with the hypothesis set forward by Geyh and Grosjean (2000), the reservoir effect was most likely lower in Unit 1 than in Unit 2 since the lake was, on average, shallower than during the deposition of the upper unit. This is supported by a seismic study, by the absence of emerged former lake terraces, and other indicators (sedimentary facies, diatoms, carbonate content) (Sa´ez et al., 2007). We propose to correct the Unit 1 dates for two possible extreme reservoir age values: a minimum value of 0 years and a maximum of 2320 years. Once corrected by considering these two extreme reservoir effect values, an age range of variation for every Unit 1 date is obtained. These dates were calibrated using the most updated calibration curve (INTCAL04), which is provided by the CALIB 5.02 software package (Reimer et al., 2004) selecting the mid-point of the 95.4% of the distribution (2s probability interval). In order to fill the gaps of the age model constructed only by 14C AMS dates, four 238U/230Th measurements were carried out on calcite crystals that appeared in some thin layers from Chungara´ cores. These dates were done via the standard method using an ICP-IRMS multicollector at the University of Minnesota (Table 3). For a summary of this method see Edwards et al. (1986). Only one 238U/230Th date was finally acceptable for the chronological model due to their high content of 232Th in the other three samples. Fortunately, this 238U/230Th date belongs to Subunit 2a, the only interval without sound 14C AMS dates. To construct a reliable age-depth model with the remaining 12 dates (11 14C AMS and 1 U/Th), we used the software described in Heegaard et al. (2005) as a useful interpolation tool. This software provides a procedure to

Table 3 238 U/230Th ages measured in Chungara cores Depth (cm)

Sample ID

Carbonate material

238

232

d234U

230

230 Th/232Th (ppm)

Error

Calendar age BP

280 285 344 374

14A-3, 6 13A-2, 45 13A-2, 105 15A-4, 77

Crystal Crystal Crystal Shell

Out of scale 576.4 467.4 717.4

— 53.9 35.2 203.1

323.3 335.1 413.5 320.2

— 0.0774 0.1036 0.1607

0 14 23 9

— 1244 974 3445

— 4450 6730 7720

U (ppb)

Th (ppm)

Th/238U

Depth transposed to composite sequence is indicated. Italic samples were not used for age model construction (reversed dates, error too high or high values of 232Th). See text for details about calibration procedures and Fig. 9.

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

16

estimate the age–depth relationship by setting the midpoint value of the calibrated ages in relation to the central distributional range. The software provides a final corrected age for every calibrated date (Table 2, column ‘‘Calibrated age after Heegaard’s method’’). These ages are then employed for the age-depth model (Fig. 9). This applied correction has led to slight age differences in the boundaries of the subunits with respect to those presented in Sa´ez et al. (2007) and incorporates two additional tie points. In summary, the age model for the sediment sequence of Lago Chungara´ presented here is constructed with (1) the available dates along Subunit 2b corrected by a constant reservoir effect similar to that currently found, (2) a unique date in Subunit 2a acquired by the 238U/230Th procedure and (3) the mid-point value of the obtained range taking into account the maximum and the minimum reservoir age corrections for the existing dates along Unit 1. Thus the Lago Chungara´ sequence covers the last 14,000 cal. yr BP

Calibrated age (cal. years BP) after Heegaard's model 0

2000

4000

6000

8000

10000 12000 14000 16000

0 Subunit 2b 100

Composite depth (cm)

200 Subunit 2a 300 U/Th age 400 500 Subunit 1b 600 700 800

Subunit 1a

900 Model 1 and 2: common reservoir age correction for Unit 2 Model 1: constant reservoir age correction Model 2: no reservoir age correction for Unit 1

Fig. 9. Age control points used for the age model of Lago Chungara´. We represent the ‘‘Calibrated age (cal. yr BP) after Heegaard’s method’’ data from Table 2. The error bar displayed for every age is a consequence of the interpolation carried out following Heegaard et al. (2005). Two different approaches were carried out to correct the reservoir effect in the Lago Chungara´ sequence: model 1 (application of a constant reservoir age correction along the entire sequence, white dots) and model 2 (application of a constant reservoir age correction for Subunit 2b and no correction for Unit 1 dates, gray dots). Therefore, for both models the uppermost interval (Subunit 2b) is common (reservoir age is considered constant, 2320 years, black dots). In Subunit 2a only one U/Th date is available. Along Unit 1, the calibrated dates range from model 1 to model 2 values, representing the maximum and minimum reservoir age correction, respectively. Although the final age model was constructed with the intermediate values from the represented range (dashed line), the broad interval is taken into account in the paleoclimate interpretations.

with a sedimentation rate in offshore zones between 0.47 and 0.78 mm yr1. 5. Paleoclimate implications Selected proxies from the Chungara´ sequence were plotted versus age and compared with published records from the nearby Nevado Sajama ice core and Lake Titicaca (Fig. 10). Accordingly, any discussions addressing the paleoclimate implications should take into account the range of possible ages along Unit 1 (Fig. 9). Additional improvements to the age model of this sequence (more 238 U/230Th dates or dating tephra layers) will allow more extensive and refined interpretations of paleoclimate dynamics. Close correlation among Fe elemental intensities and the amount of coarse particles in the Sajama ice core indicates that the volcanic signal detected in the Chungara´ record has a regional character and that most of the dust particles in the Sajama ice core are of volcanic origin. Very likely, this volcanic signal is related to the Parinacota Volcano eruptions since it is the only one in the area with recent eruptive activity (Wo¨rner et al., 1988). Due to the short distance (30 km) between Lago Chungara´ and Nevado Sajama (Fig. 1), one can easily infer that the source for volcaniclastic particles in both records was the same. Therefore, we suggest that the variations found in the coarser particles in the Sajama ice core are mainly controlled by the main volcanic eruptions of Parinacota Volcano and are not due to (1) atmospheric dust input or (2) the greater availability of salt minerals during dry periods in the Altiplano, as was proposed by Thompson et al. (1998). The Fe record from Lago Chungara´ and the Sajama record indicate that explosive activities of Parinacota Volcano increased dramatically at 5800 cal. yr BP (Fig. 10). The period with less explosive volcanic activity is also coincident in both records from 14,000 to 7000–8000 cal. yr BP (Fig. 10). A comparison of the Lago Chungara´ record with other paleoclimate reconstructions from the nearby Cotacotani basin becomes necessary to assess the Chungara´ record within a regional context. Pollen stratigraphy obtained from an outcrop at Laguna Seca (Fig. 1) indicates a gradual transition towards drier and warmer climates since the late Pleistocene (Baied and Wheeler, 1993). An increased desiccation between 8000 and 6500 14C yr BP, as well as warmer conditions until about 5000 14C yr BP are also suggested by that study. These paleoclimate interpretations are supported by an isotopic study carried out in a sediment core obtained from Laguna Seca (Schwalb et al., 1999). Moreover, a transition towards higher lake levels from the mid-to-late Holocene was also postulated (Schwalb et al., 1999). The high-resolution geochemical study of Lago Chungara´ provides new data on two main paleoclimatic topics: the glacial–interglacial transition and the aridity crises during the Holocene.

A. Moreno et al. / Quaternary International 161 (2007) 4–21

17

Fig. 10. Selected records (Si/Ti, Ca/Ti, Mn/Ti and Fe) from Lago Chungara´ plotted versus age. Represented tie points (black rhombuses) are the midpoint values of model 1 and 2 dates (Table 2). The d18O and total particles coarser than 63 mm from the Nevado Sajama ice core record (data from the World Data Center for Paleoclimatology, Boulder and the NOAA Paleoclimatology Program) and the benthic diatoms from Lake Titicaca (Baker et al., 2001a) are plotted for comparison. Summer insolation at 181S is indicated (reversed axis). An arrow marks the productivity increase along the early Holocene. Both maximums in aridity in Lago Chungara´ and Titicaca are indicated. The complete arid phase for Lake Titicaca is also indicated).

5.1. Glacial–interglacial transition—early holocene The glacial–interglacial transition can be observed in several northern Andean Altiplano records as a sharp change towards drier conditions associated with minimum summer insolation (E11,000 cal. yr BP, Fig. 10) following the wet Tauca phase when the largest lakes and the highest lake levels occurred (Argollo and Mourguiart, 2000; Baker et al., 2001a; Mourguiart and Ledru, 2003). Although the end of the Tauca phase has not been well established, it is believed to have ended about 14,900 cal. yr BP (Baker et al., 2001b; Fornari et al., 2001). In some records, a short wet event (13–11.5 kyr BP; ‘‘Coipasa phase’’) has been observed and correlated with the Younger Dryas in the northern hemisphere (Baker et al., 2001a). One of the main transitions in the Lago Chungara´ record corresponds with the boundary between Subunits 1a and 1b, which depending on the age model occurred between 9500 and 12,500 cal. yr BP (Fig. 9). In spite of the chronological uncertainty, this boundary is likely to represent the glacial–interglacial transition and the onset of the Holocene. During that transition, Chungara sediments reflect the change from green-white laminations (Subunit 1a) to brown-white laminations (Subunit 1b).

Although the meaning of these laminations cannot be discerned with our geochemical proxies, we suggest an increase in productivity, probably related to favored diatom productivity, at the beginning of the Holocene (11,500 cal. yr BP) lasting until 7500 cal. yr BP. A detailed diatom study (Bao et al., in preparation) and organic geochemical analyses are in progress to understand the significance of the different laminations and their cyclicities. As well as the sedimentary facies changes, several productivity proxies point to a general increase (TOC, Si/ Ti and S/Ti) after the end of the glacial period. High values are maintained throughout the early Holocene (Figs. 6 and 10). This change correlates with the Sajama ice core isotopic change towards warmer temperatures (Thompson et al., 1998). In contrast to the glacial–interglacial transition, the relevant geochemical indicators do not show any clear indication of climate changes associated with the ‘‘Coipasa phase’’ or Younger Dryas event (Fig. 10). However, in the Sajama ice core (Thompson et al., 1998) and the Titicaca record (Baker et al., 2001a), a decrease in temperature and an increase in humidity have been postulated for that period. The high sampling resolution attained in this study allowed focusing on short and dramatic climate events that

ARTICLE IN PRESS 18

A. Moreno et al. / Quaternary International 161 (2007) 4–21

otherwise would not be observable by discrete sampling. In the Chungara´ record, a very prominent peak in the Mn/Ti profile occurs at around 10,000 cal. yr BP ranging from 8400 to 11,400 cal. yr BP (Fig. 10). As outlined above, Mn is assumed to be associated with redox front variations when oxic conditions develop following an interval in which an anoxic environment dominated. Therefore, we propose that an increase in oxygen in the bottom waters occurred at that time. The most probable mechanism that would produce such an increase in oxygen is a lack of long periods of water stratification, e.g., sufficiently low lake levels could cause winds to mix the entire water column. This mechanism may be reinforced by an increase in diatom productivity, as a consequence of such a mixing, and thus in oxygen supply. This potentially more arid scenario (lower water levels), when compared with previous sediment deposits would be supported by an increase in benthic diatoms during this interval as indicated by preliminary data (Sa´ez et al., 2007; Bao et al., in preparation). In addition, peaks in calcite contents, Ca/Ti and Sr/Ti would further strengthen this hypothesis (Figs. 8 and 10). 5.2. Aridity crisis during the early and mid-Holocene Apart from the event detected in the Lago Chungara´ sediments around 10,000 cal. yr BP in the Mn/Ti record, other intervals were observed during the early-to-mid Holocene that may indicate lower lake levels as well (Fig. 10). In general, periods of high lake levels in the Andean Altiplano are interpreted in terms of increasing summer precipitation in the southern hemisphere, with the ITCZ occupying a more southern position, while dry periods are related to a northward displacement of the ITCZ (i.e. Argollo and Mourguiart, 2000). In the nearby Laguna Seca record, a transition from carbonate-rich, laminated lacustrine sediments to peaty sediments occurred at about 70307245 14C yr BP (Baied and Wheeler, 1993). In spite of the uncertainties with the Laguna Seca age model, this change from a lake towards a peatbog points to arid conditions for the mid-Holocene in the Lago Chungara´ area. Enhanced carbonate precipitation in the offshore zones of Lago Chungara´ was observed along Subunit 2a (Figs. 8 and 10). Carbonate production in the lake requires the presence of Ca in the water, which is assured throughout nearly the entire sequence: Ca is provided by leaching of Ca-rich volcanic ashes or by the increase of volcanic input that becomes more important in Subunit 2a (Sa´ez et al., 2007). Some of the carbonate-rich levels occur in intervals where Ca-rich tephra layers are more frequent, suggesting that volcanism plays a role. However, carbonate formation predates the Ca-rich volcanic interval (Subunit 2b), and carbonate production halts before the Ca-rich volcanic input to the lake stops, suggesting that other factors could also control carbonate formation. Therefore, the presence of endogenic carbonate deposits in offshore cores is a potential indicator of more concentrated waters, e.g., a

scenario involving lower lake levels with an increase in littoral areas. The increase in benthic diatoms in the carbonate-rich levels supports such a hypothesis (Sa´ez et al., 2007; Bao et al., in preparation). However, fluctuations in biological activity, as well as changes in inflowing water composition could also have played a role in making the lake waters more conducive to carbonate formation. Several periods of increased carbonate formation in littoral areas from Lago Chungara´, related to increased charophyte productivity, have occurred during the last 4000 years (Valero-Garce´s et al., 1996, 2003). In nearby Laguna Seca, periods of carbonate formation seem to have occurred throughout the Holocene (Baied and Wheeler, 1993; Schwalb et al., 1999), although in this case they seem to be related to travertine deposition and spring activity. Detailed diatom studies and statistical analyses to more precisely determine the volcanic input are in progress. Considering the offshore formation of carbonate as an indicator of more concentrated lake water, the period ranging from 8600 to 6400 cal. yr BP could be proposed as the most arid one (taking into account the established age ranges, the starting point of this interval varies from 7300 to 9500 cal. yr BP). This period does not precisely coincide with the dry period detected in Lake Titicaca using benthic diatom abundances which is dated between 8.5 and 4.5 cal. kyr BP, with an extremely dry period from 6 to 5 cal. kyr BP (Baker et al., 2001a; Tapia et al., 2003, Fig. 10). The main difference with the Lake Titicaca record is the timing of the dry period, ending after the corresponding period of the Chungara´ record (Fig. 10). As detected in other studies, Holocene aridity crises were not synchronous in the Andean Altiplano (Betancourt et al., 2000; Holmgren et al., 2001; Abbott et al., 2003; Grosjean et al., 2003; Latorre et al., 2003). According to the age model, these findings are consistent with the proposed N–S gradient, and therefore support the complex timing and structure of the early-to-mid Holocene aridity crisis. Finally, it is worth noting that dry conditions were not constant during the arid period in the Chungara´ region, but characterized by a series of short and rapid dry spells (Fig. 10). Considering the Ca/Ti record as an aridity indicator, several abrupt arid periods (of less than 100 years duration) some of them coincident with Mn/Ti peaks are detected (Fig. 10). As stated before, both Ca/Ti and Mn/Ti ratios point towards climate scenarios with less water availability in the Chilean Altiplano. The fact that several arid events alternate with less arid periods during a relatively short time interval indicates that conditions were unstable and highly changeable. Although these events cannot be precisely correlated with other paleoclimate archives in the region due to age model uncertainties, they support the argument that the aridity postulated to have occurred since the early Holocene in South American tropical records is not continuous but rather represents a succession of short

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21

dry/wet events. Despite the fact that the forcing mechanisms of these abrupt changes in moisture availability remain partially unknown (Ruter et al., 2004), model reconstructions should take into consideration all possible factors that could account for the presence of abrupt and short dry/wet events over this generally arid period.

6. Conclusions High-resolution geochemical profiles were produced for the sedimentary succesion recovered from Lago Chungara´ at 181S. These results, compared with physical, mineralogical and sedimentological data obtained from the same cores, provide the first high-resolution lacustrine sequence of the Chilean Andean Altiplano during the last 14,000 cal. yr BP. The exhaustive analysis of the meaning of the elementary signatures, taking into account their potential origins before inferring a paleoclimatic signal, allows recognition of three main components in Lago Chungara´ sediments: (1) lacustrine biological remains mainly from diatoms, (2) volcanic minerals, and (3) endogenic offshore carbonates. Siliceous productivity was deduced from Si/Ti, opal and amorphous material while the volcanic supplies were inferred from Fe, Ti and K elemental intensities, variation in magnetic susceptibility and occurrences of plagioclase and amphibole minerals. Volcanic contributions were closely correlated to the total amount of coarse particles recovered in the Nevado Sajama ice core, thus pointing to the same volcano source for both, the Lago Chungara´ and Sajama records, most likely the Parinacota Volcano main explosive eruptions. Offshore carbonate production was asserted from Ca/Ti and Sr/Ti ratios as well as from calcite and TIC contents. The transition from greenish to brownish laminated diatomites identifies the glacial–interglacial transition in Chungara´ sediments that should be synchronous with the Sajama temperature increase. In addition, productivity proxies show a tendency towards higher values, suggesting an increase in diatoms that was maintained throughout the early Holocene. The use of continuously measured properties allowed investigating the timing of abrupt and short events that would remain undetectable with a discrete sampling methodology. The interval between 8600 and 6400 cal. kyr BP appears to have been the driest episode, starting and ending before the Titicaca extreme dry period (6–5 cal. kyr BP). Additionally, several short and abrupt arid intervals were detected from the carbonate content, with the ca. 10,000 yr event being the most pronounced one. For that period, an increase in oxic conditions in the lake bottom waters is postulated to have occurred synchronously to the sharp Mn/Ti peak. We suggest that a lowered lake level may have caused more efficient deepwater ventilation and, consequently, the formation of Mn oxides. A study of biological proxies, mainly diatoms, is currently underway to further strengthen our hypothesis regarding lake-level variability during the last 14,000 years.

19

Acknowledgements We are indebted to the Limnological Research Center staff who participated in the field expedition (D. Schnurremberger, M. Shapley and A. Myrbo) and collaborated during the initial core descriptions, as well as the CONAF (Corporacio´n Nacional Forestal, Chile) for the facilities provided in Chungara´. We acknowledge C. Herrera (Universidad Cato´lica del Norte, Chile) for his help during the field expedition. We are very grateful to L. Edwards (University of Minnesota) and R.O. Gibert (University of Barcelona) for the ICP-IRMS U/Th dating and to S. Fritz for providing the Titicaca diatom data. The University of Bremen, particularly U. Ro¨hl, F. Lamy, M. Ko¨lling, H. Pfletschinger and H. Kuhlmann are acknowledged for technical assistance with the XRF-Core Scanner, powder XRF and ICP-OES analyses. We also acknowledge M. Grosjean and N. Piotrowska for their advice in age model construction and R. Rycroft for the English correction. The Paleostudies programme (European Science Foundation) provided the necessary funding to carry out the analyses at the University of Bremen. This study is supported by the projects BTE2001-3225 and BTE20015257-E and CGL2004-00683/BTE funded by the CICYT, the Spanish Ministry of Science and Technology. A. Moreno and P. Gonza´lez-Sampe´riz are the recipients of a CSIC research contract (I3P postdoctoral programme). S. Giralt acknowledges the Spanish Ministry of Science and Technology for his postdoctoral contract in the Ramo´n y Cajal programme. References Abbott, M.B., Seltzer, G.O., Kelts, K., Southon, J., 1997. Holocene paleohydrology of the tropical Andes from Lake Records. Quaternary Research 47, 70–80. Abbott, M.B., Wolfe, B.B., Wolfe, A.P., Seltzer, G.O., Aravena, R., Mark, B.G., Polissar, P.J., Rodwell, D.T., Rowe, H.D., Vuille, M., 2003. Holocene paleohydrology and glacial history of the central Andes using multiproxy lake sediment studies. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 123–138. Aguilar, C., Nealson, K.H., 1998. Biogeochemical cycling of manganese in Oneida Lake, New York: whole lake studies of manganese. Journal of Great Lakes Research 24, 93–104. Argollo, J., Mourguiart, P., 2000. Late Quaternary climate history of the Bolivian Altiplano. Quaternary International 72, 37–51. Baied, C.A., Wheeler, J.C., 1993. Evolution of High Andean Puna ecosystem environment, climate and culture change over the last 12,000 years in the Central Andes. Mountain Research and Development 13, 145–156. Baker, P.A., Seltzer, G.O., Fritz, S.C., Dunbar, R.B., Grove, M.J., Tapia, P.M., Cross, S.L., Rowe, H.D., Broda, J.P., 2001a. The history of South American tropical precipitation for the past 25,000 years. Science 291, 640–643. Baker, P.A., Rigsby, C.A., Seltzer, G.O., Fritz, S.C., Lowenstein, T.K., Bacher, N.P., Veliz, C., 2001b. Tropical climate changes at millennial and orbital timescales on the Bolivian Altiplano. Nature 409, 698–701. Betancourt, J.L., Latorre, C., Rech, J.A., Quade, J., Rylander, K.A., 2000. A 22,000-year record of monsoonal precipitation from Northern Chile’s Atacama Desert. Science 289, 1542–1546.

ARTICLE IN PRESS 20

A. Moreno et al. / Quaternary International 161 (2007) 4–21

Butler, K., Prior, C.A., Flenley, J.R., 2004. Anomalous radiocarbon dates from Easter Island. Radiocarbon 46 (1), 395–405. Chung, F.H., 1974. Quantitative interpretation of X-ray diffraction patterns of mixtures: II. Adiabatic principles of X-ray diffraction analysis of mixtures. Journal of Applied Crystallography 7, 526–531. Clavero, J.E., Sparks, R.S.J., Huppert, H.E., Dade, W.B., 2002. Geological constraints on the emplacement mechanism of the Parinacota debris avalanche, northern Chile. Bulletin of Volcanology 64, 40–54. Clavero, J.E., Sparks, R.S.J., Polanco, E., Pringle, M.S., 2004. Evolution of Parinacota Volcano, Central Andes, Northern Chile. Revista Geolo´gica de Chile 31, 317–347. Dorador, C., Pardo, R., Vila, I., 2003. Variaciones temporales de para´metros fı´ sicos, quı´ micos y biolo´gicos de un lago de altura: el caso del Lago Chungara´. Revista Chilena de Historia Natural 76, 15–22. Earle, L.R., Warner, B.G., Aravena, R., 2003. Rapid development of an unusual peat-accumulating ecosystem in the Chilean Altiplano. Quaternary Research 59, 2–11. Edwards, R.L., Chen, J.H., Wasserburg, G.J., 1986. 238 U–234U–230Th–232Th systematics and the precise measurements of time over the past 500,000 years. Earth and Planetary Science Letters 81, 175–192. Fornari, M., Risacher, F., Fe´raud, G., 2001. Dating of paleolakes in the central Altiplano of Bolivia. Palaeogeography, Palaeoclimatology, Palaeoecology 172, 269–282. Garreaud, R.D., 2001. Subtropical cold surges: regional aspects and global distribution. International Journal of Climatology 21, 1181–1197. Garreaud, R.D., Vuille, M., Clement, A.C., 2003. The climate of the Altiplano: observed current conditions and mechanisms of past changes. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 5–22. Geyh, M.A., Grosjean, M., 2000. Establishing a reliable chronology of lake level changes in the Chilean Altiplano: a result of close collaboration between geochronologists and geomorphologists. Zentralblatt fu¨r Geologie und Pala¨ontologie, Teil I 7/8, 985–995. Geyh, M.A., Grosjean, M., Nu´n˜ez, L., Schotterer, U., 1999. Radiocarbon reservoir effect and the timing of the Late-Glacial/early Holocene Humid phase in the Atacama desert (Northern Chile). Quaternary Research 52, 143–153. Grosjean, M., 2001. Mid-Holocene climate in the South-Central Andes: humid or dry? Science 292, 2391–2392. Grosjean, M., van Leeuwen, J.F.N., van der Knaap, W.O., Geyh, M.A., Ammann, B., Tanner, W., Messerli, B., Nu´n˜ez, L., Valero-Garce´s, B.L., Veit, H., 2001. A 22,000 14C year BP sediment and pollen record of climate change from Laguna Miscanti (231S), northern Chile. Global and Planetary Change 28, 35–51. Grosjean, M., Cartajena, I., Geyh, M.A., Nu´n˜ez, L., 2003. From proxy data to paleoclimate interpretation: the mid-Holocene paradox of the Atacama Desert, northern Chile. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 247–258. Hansen, H.P., Grashoff, K., 1983. Automated chemical analysis. In: Grashoff, M., Ehrhardt, M., Kremlin, K. (Eds.), Methods of Seawater Analysis. Verlag Chemie, Weinheim, pp. 368–376. Heegaard, E., Birks, H.J.B., Telford, R.J., 2005. Relationships between calibrated ages and depth in stratigraphical sequences: an estimation procedure by mixed-effect regression. The Holocene 15, 612–618. Herrera, C., Pueyo, J.J., Sa´ez, A., Valero-Garce´s, B.L., 2006. Relacio´n de aguas superficiales y subterra´neas en el a´rea del lago Chungara´ y lagunas de Cotacotani, norte de Chile: un estudio isoto´pico. Revista Geolo´gica de Chile 33 (2), 299–325. Holmgren, C.A., Betancourt, J.L., Rylander, K.A., Roque, J., Tovar, O., Zeballos, H., Linares, E., Quade, J., 2001. Holocene vegetation history from fossil rodent middens near Arequipa, Peru. Quaternary Research 56, 242–251. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Xiaosu, D., 2001. Climate Change 2001: The Scientific Basis. Cambridge University Press, Cambridge, p. 944.

Hughen, K.A., Overpeck, J.T., Peterson, L.C., Trumbore, S., 1996. Rapid climate changes in the tropical Atlantic region during the last deglaciation. Nature 380, 51–54. Hughen, K.A., Eglinton, T.I., Xu, L., Makou, M., 2004. Abrupt tropical vegetation response to rapid climate changes. Science 304, 1955–1959. Jansen, J.H.F., Van der Gaast, S.J., Koster, B., Vaars, A.J., 1998. CORTEX, a shipboard XRF-scanner for element analyses in split sediment cores. Marine Geology 151, 143–153. Kelts, K., Hsu¨, K.J., 1978. Freshwater carbonate sedimentation. In: Lerman, A. (Ed.), Lakes: Chemistry, Geology, Physics. Springer, Berlin, pp. 297–323. Kuhlmann, H., Meggers, H., Freudenthal, T., Wefer, G., 2004. The transition of the monsoonal and the N Atlantic climate system off NW Africa during the Holocene. Geophysical Research Letters 31, L22204. Latorre, C., Betancourt, J.L., Rylander, K.A., Quade, J., Matthei, O., 2003. A vegetation history from the arid prepuna of northern Chile (22–231S) over the last 13,500 years. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 223–246. Mayewski, P.A., Rohling, E.J., Stager, J.C., Karle´n, W., Maasch, K.A., Meeker, L.D., Meyerson, E.A., Gasse, F., Van Kreveld, S.A., Holmgren, C.A., Lee-Thorp, J.A., Rosqvist, G., Rack, F., Staubwasser, M., Schneider, R., Steig, E.J., 2004. Holocene climate variability. Quaternary Research 62, 243–255. Montgomery, E., Rosko, M., Castro, S., Keller, B., Bevacqua, P., 2003. Interbasin underflow between closed Altiplano Basins in Chile. Groundwater 41 (4), 523–531. Mortlock, R.A., Froelich, P.N., 1989. A simple method for the rapid determination of biogenic opal in pelagic marine sediments. Deep-Sea Research 36, 1415–1426. Mourguiart, P., Ledru, M.P., 2003. Last Glacial Maximum in an Andean cloud forest environment (Eastern Cordillera, Bolivia). Geology 31, 195–198. Mu¨hlhauser, H., Hrepic, N., Mladinic, P., Montecino, V., Cabrera, S., 1995. Water quality and limnological features of a high altitude Andean lake, Chungara´, in northern Chile. Revista Chilena de Historia Natural 68, 341–349. Nu´n˜ez, L., Grosjean, M., Cartajena, I., 2002. Human occupations and climate change in the Puna de Atacama, Chile. Science 298, 821–824. Paduano, G.M., Bush, M.B., Baker, P.A., Fritz, S.C., Seltzer, G.O., 2003. A vegetation and fire history of Lake Titicaca since the Last Glacial Maximum. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 259–279. R Development Core Team, 2004. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria URL /http://www.R-project.orgS. Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Bertrand, C.J.H., Blackwell, P.G., Buck, C.E., Burr, G.S., Cutler, K.B., Damon, P.E., Edwards, R.L., Fairbanks, R.G., Friedrich, M., Guilderson, T.P., Hogg, A.G., Hughen, K.A., Kromer, B., McCormac, G., Manning, S., Ramsey, C.B., Reimer, R.W., Remmele, S., Southon, J.R., Stuiver, M., Talamo, S., Taylor, F.W., van der Plicht, J., Weyhenmeyer, C.E., 2004. IntCal04 terrestrial radiocarbon age calibration, 0–26 Cal Kyr BP. Radiocarbon 46, 1029–1058. Risacher, F., Alonso, H., Salazar, C., 2003. The origin of brines and salts in Chilean salars: a hydrochemical review. Earth-Science Reviews 63, 249–293. Ro¨hl, U., Abrams, L.J., 2000. High-resolution, downhole, and nondestructive core measurements from sites 999 and 1001 in the Caribbean Sea: application to the Late Paleocene Thermal Maximum. In: Leckie, R.M., Sigurdsson, H., Acton, G. D., Draper, G. (Eds.), Proceedings of the Ocean Drilling Program, Scientific Results, pp. 191–203. Ruter, A., Arzt, J., Vavrus, S., Bryson, R.A., Kutzbach, J.E., 2004. Climate and environment of the subtropical and tropical Americas (NH) in the mid-Holocene: comparison of observations with climate model simulations. Quaternary Science Reviews 23, 663–669. Sa´ez, A., Valero-Garce´s, B.L., Moreno, A., Bao, R., Pueyo, J.J., Gonza´lez-Sampe´riz, P., Giralt, S., Taberner, C., Herrera, C., Gibert, R.O., 2007. Volcanic controls on lacustrine sedimentation: the late

ARTICLE IN PRESS A. Moreno et al. / Quaternary International 161 (2007) 4–21 Quaternary depositional evolution of lake Chungara´ (Northern Chile). Sedimentology, in press. Schwalb, A., Burns, S., Kelts, K., 1999. Holocene environments from stable isotope stratigraphy of ostracods and authigenic carbonate in Chilean Altiplano Lakes. Palaeogeography, Palaeoclimatology, Palaeoecology 148, 153–168. Servant, M., Servant-Vildary, S., 2003. Holocene precipitation and atmospheric changes inferred from river paleowetlands in the Bolivian Andes. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 187–206. Tapia, P.M., Fritz, S.C., Baker, P.A., Seltzer, G.O., Dunbar, R.B., 2003. A Late Quaternary diatom record of tropical climatic history from Lake Titicaca (Peru and Bolivia). Palaeogeography, Palaeoclimatology, Palaeoecology 194, 139–164. Teranes, J.L., McKenzie, J.A., Bernasconi, S.M., Lotter, A.F., Sturm, M., 1999. A study of oxygen isotopic fractionation during bio-induced calcite precipitation in eutrophic Baldeggersee, Switzerland. Geochimica et Cosmochimica Acta 63, 1981–1989. Theissen, K.M., Zinniker, D.A., Moldowan, J.M., Dunbar, R.B., Rowe, H.D., 2005. Pronounced occurrence of long-chain alkenones and dinosterol in a 25,000-year lipid molecular fossil record from Lake Titicaca, South America. Geochimica, Cosmochimica Acta 69, 623–636. Thompson, L.G., Davis, M., Mosley-Thompson, E., Sowers, T., Henderson, K.A., Zagorodnov, V.S., Lin, P.N., Mikhalenko, V.N., Campen, R.K., Bolzan, J.F., Cole-Dai, J., Francou, B., 1998. A 25,000-year tropical climate history from Bolivian ice cores. Science 282, 58–64. Thompson, L.G., Mosley-Thompson, E., Henderson, K.A., 2000. Ice-core palaeoclimate records in tropical South America since the Last Glacial Maximum. Journal of Quaternary Science 15, 377–394. Valero-Garce´s, B.L., Grosjean, M., Schwalb, A., Kelts, K., Schreir, H., Messerli, B., 1996. Limnogeologı´ a de Laguna Chungara´ y cambio clima´tico durante el Holoceno superior en el altiplano chileno

21

septentrional. In: Grandal d’Anglade, A., Page´s Valcarlos, J. (Eds.), IV Reunio´n de Geomorfologı´ a, O Castro, A Corun˜a, Spain, pp. 271–280. Valero-Garce´s, B.L., Grosjean, M., Kelts, K., Schreir, H., Messerli, B., 1999. Holocene lacustrine deposition in the Atacama Altiplano: facies models, climate and tectonic forcing. Palaeogeography, Palaeoclimatology, Palaeoecology 151, 101–125. Valero-Garce´s, B.L., Grosjean, M., Messerli, B., Schwalb, A., Kelts, K., 2000. Late Quaternary lacustrine deposition in the Chilean altiplano (18–281S). In: Gierlowski-Kordesch, E.H., Kelts, K. (Eds.), Lake Basins Through Space and Time. AAPG Studies in Geology, pp. 625–636. Valero-Garce´s, B.L., Delgado-Huertas, A., Navas, A., Edwards, R.L., Schwalb, A., Ratto, N., 2003. Patterns of regional hydrological variability in central-southern Altiplano (18–261S) lakes during the last 500 years. Palaeogeography, Palaeoclimatology, Palaeoecology 194, 319–338. Vuille, M., 1999. Atmospheric circulation over the Bolivian altiplano during dry and wet periods and extreme phases of the Southern Oscillation. International Journal of Climatology 19, 1579–1600. Wo¨rner, G., Harmon, R.S., Davidson, J., Moorbath, S., Turner, D.L., McMillan, N., Nye, C., Lo´pez-Escobar, L., Moreno, H., 1988. The Nevados de Payachata volcanic region (181S/691W, N Chile). I: geological, geochemical and isotopic observations. Bulletin of Volcanolgy 50, 287–303. Wo¨rner, G., Hammerschmidt, K., Henjes-Kunst, F., Wilke, H., 2000. Geochronology (40Ar/39Ar, K–Ar and He-exposure ages) of Cenozoic magmatic rocks from northern Chile (18–221S): implications for magmatism and tectonic evolution of the central Andes. Revista Geolo´gica de Chile 27, 205–240. Wo¨rner, G., Uhlig, D., Kohler, I., Seyfried, H., 2002. Evolution of the West Andean Escarpment at 181S (N. Chile) during the last 25 Ma: uplift, erosion and collapse through time. Tectonophysics 345, 183–198.