Available online at www.sciencedirect.com R
Earth and Planetary Science Letters 212 (2003) 225^239 www.elsevier.com/locate/epsl
Cyclic water level oscillations of the KaraBogazGol^Caspian Sea system S. Giralt a; , R. Julia' b , S. Leroy c , F. Gasse d a b
Department of Geology, Royal Museum for Central Africa, Leuvensesteenweg 13, B-3080 Tervuren, Belgium Institute of Earth Sciences ‘Jaume Almera’ (C.S.I.C.), Lluis Sole i Sabaris s/n, E-08028 Barcelona, Spain c Department of Geography and Earth Sciences, Brunel University, Uxbridge, Middlesex UB8 3PH, UK d CNRS - CEREGE, Europo“le Me¤diterrane¤en l’Arbois, BP 80, 13545 Aix-en-Provence Cedex 04, France Received 12 December 2002; received in revised form 16 April 2003; accepted 6 May 2003
Abstract The KaraBogazGol (KBG) water level oscillations were reconstructed in the last 200 years using the geochemical evolution of the uppermost meter of its sedimentary infill. High-resolution studies of the mineralogical composition of the KBG sediments show alternating periods of high concentration brines followed by periods of more dilute waters. The relative water level reconstruction was based on statistical models (factor analysis (FA), correspondence analyses (CA) and detrended correspondence analyses (DCA)) whilst the chronological framework was established using the 210 Pb technique. This reconstruction was compared with the instrumental water level record of the Caspian Sea, showing a high degree of correlation (r2 = 0.83). The agreement between the reconstructed and the measured water level oscillations of the Caspian Sea indicates that environmental changes can be reconstructed from closed lakes provided that an accurate chronological control of sedimentary processes is available. The water level reconstruction shows that the Caspian Sea water level fluctuations follow a cyclical pattern rather than an ‘erratic’ one, as has been suggested in the literature. Two main periodicities of 62.5 and 38.46 years have been found. ; 2003 Elsevier Science B.V. All rights reserved. Keywords: Caspian Sea; KaraBogazGol; mineralogy; water level reconstruction; cyclical pattern
1. Introduction Lakes are excellent sensors for recording re-
* Corresponding author. Present address: Institute of Earth Sciences ‘Jaume Almera’ (CSIC), Lluis Sole¤ i Sabaris s/n, E-08028 Barcelona, Spain. Tel.: +34-934-095-410; Fax: +34-934-110-012. E-mail addresses:
[email protected] (S. Giralt),
[email protected] (S. Giralt),
[email protected] (R. Julia'),
[email protected] (S. Leroy),
[email protected] (F. Gasse).
gional environmental changes [1,2], which are recorded in lake sediments through variations in the biological assemblages and mineral species [3^7]. The study of lacustrine sediments allows us to characterize and model these changes (see [1,8] for further details). Such paleoenvironmental information allows us to determine and to evaluate the most recent long-term water level oscillations. This information should enable us to adopt appropriate shortand mid-term environmental policies in order to predict and mitigate their possible impact on so-
0012-821X / 03 / $ ^ see front matter ; 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0012-821X(03)00259-0
EPSL 6681 19-6-03
226
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Fig. 1. Mean annual Volga River discharge increments (dotted line) and mean annual Caspian Sea water level increments (continuous line) for the period 1901^1987. Modi¢ed from [12].
ciety. These data are especially valuable in the Caspian Sea^KaraBogazGol (KBG) system given that the historical water level records of both aquatic ecosystems show a dramatic water level decrease during the 1930s (up to 3 m), with signi¢cant environmental (concentration of heavy metals, pesticides and diverse organic compounds in the epilimnion of the Caspian Sea water body, sharp decrease in the sturgeon population) and economic (shallowing of the main northern harbors of the Caspian Sea, possible inundation of the nuclear power plants located in the delta plain of the Volga River) consequences for the surrounding countries (see [9] for further details). The causes of the Caspian Sea water level oscillations remain unclear and a number of possible mechanisms triggering them have been suggested. Some authors have suggested that these water level oscillations were forced by natural £uctuations in the climate of the Northern Hemisphere, mainly related to North Atlantic oscillation (NAO) variations [10], but other authors have pointed out the possible relationship with either the sunspot period [11] or the ENSO phenomenon [12]. Other works have modeled these water level oscillations suggesting that there are non-climatological factors such as tectonically induced variation of the basin con¢guration, which also play an important role [13]. Despite the poorly understood mechanisms,
these studies indicate that the Caspian Sea water level, and thus the sedimentary in¢ll of the KBG, would record these climate oscillations. This assumption seems to be correct since the annual mean values of the Caspian Sea level increments and the annual mean values of the Volga River discharge increments show the same pattern (Fig. 1). Nevertheless, superimposed on this climatic signal is the imprint of the progressive increase in anthropogenic pressure upon this ecosystem. Thus, it is noticeable that from 1950 onwards, despite the fact that both curves show the same trend, the Volga River discharge increment values are clearly below the annual increments of the Caspian Sea water level. This could be related to the regulation of the rivers that feed the Caspian Sea during the 1950s and 1960s [14]. In any case, these water level oscillations have often been described as ‘erratic’ [15]. One way to accurately characterize these water level oscillations using a long-term perspective could be through the sediments of the KBG lacustrine system. The KBG is an evaporitic bay that acts as a terminal ecosystem when the waters of the Caspian Sea evaporate (Fig. 2). Thus, this bay should record periods of high water level stands, when the geochemical composition of the water would be fresher, and periods of low water level stands, where the geochemical composition of the water would be more saline. These water geo-
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
227
Fig. 2. Geographical location of the Caspian Sea^KBG system. Location of the KBG-02-01 and KBG-08-01 cores in the KBG.
chemical changes should be recorded in the sediments of this bay through changes in their mineralogical composition. The aim of this paper is to provide geochemical evidence of the contraction/expansion water phases in the KBG sedimentary record and hence to infer past Caspian Sea water level oscillations. The enlargement of the Caspian Sea instrumental water level record using this technique allowed us to highlight the cyclical nature of the water level oscillations.
2. Site description The Caspian Sea is the largest inland water body on the planet [16,17]. It is bordered by Russia, Azerbaijan, Kazakhstan, Iran and Turkmeni-
stan (Fig. 2). It is made up of three sub-basins (north, central and south) with an average water depth of 210 m, a maximum depth of 1024 m, an estimated surface area of about 374 000 km2 and a catchment area basin surface of about 3.5U106 km2 (see [18] for further details). The present-day water level of the Caspian Sea is situated at 327.5 m below the water level of Baltic Sea. The water inputs to the Caspian Sea are represented by river in£ows, by precipitation and by groundwater. River in£ow represents approximately 80% of the total water input to the Caspian Sea. The Volga River accounts for more than 82% of the total river discharge, and the second largest water supplier is the Ural River, with a contribution of 3%. River water in£ow ranges between 459 and 200 km3 , depending on
EPSL 6681 19-6-03
228
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Fig. 3. Instrumental water level record of the Caspian Sea (continuous line) and of the KBG (discontinuous line) for the period 1900^1994. Values in m below the Baltic Sea water level. Modi¢ed from [24].
the year. The discharge volume of the rivers is not constant throughout the year. Moreover, the main water discharge to the Caspian Sea occurs mainly in June, July and August, and seems to be attributed to the rainfall that feeds the catchment area and to the melt of snow accumulated in this area during winter. The mean water input from direct precipitation over the Caspian Sea is estimated to be about 130 km3 per year [18]. The humid air masses are primarily carried by midlatitude cyclonic air currents from the Atlantic Ocean and the Mediterranean Sea in spring and autumn, although secondary centers are also found over the Caspian Sea [19,20]. Groundwater inputs range between 2 and 9% of the total water in£ow and it seems that this has a signi¢cant impact on the Caspian Sea water chemistry [21]. The main water outputs of the Caspian Sea are related to evaporation (380 km3 per year) and water £ow to the KBG basin (ranging between 30 and 1.6 km3 , depending on the year). The instrumental record of the water level oscillations covers the last 100 years and shows that the Caspian Sea has undergone considerable water level oscillations (Fig. 3). These oscillations are clearly visible in some coastal features, such as depositional coastal bodies or erosional scarps [22] and in the dynamics of the Volga delta [23].
The KBG, a bay on the eastern side of the Caspian Sea, is located in the western part of the Republic of Turkmenistan (Fig. 2). It is connected to the Caspian Sea through a narrow strait, from which it receives its main water input. The KBG is more than 160 km long, 137 km wide, has a mean water depth of 6 m, and covers approximately 18 000 km2 . The climate of this region is severe-continental and extremely dry (less than 90 mm of rainfall), with important monthly mean temperature oscillations, ranging from 5‡C in December to 29.8‡C in July. The present-day water level of the KBG is 28 m below Baltic Sea level, but a long instrumental record, over 76 years, shows that it has undergone considerable £uctuations (Fig. 3) [24]. The KBG has traditionally been exploited for salt, especially sulfates, such as mirabilite, epsomite and bischo¢te. Historical sources indicate that this extraction started long before the 20th century. Although exploitation of these salts on an industrial scale began in 1929, the construction of a factory commenced in 1963 and was completed in 1973. This factory continues to operate today. Owing to the progressive and dramatic decrease in the water level of the Caspian Sea during the 40s (which continued up to 1977), the strait of the KBG was dammed in 1980 to protect the salt industry [9]. However, this resulted in the unplanned drying out of the KBG in 1983. In an attempt to restore the ecosystem in 1984, a number of pipes pumped water into the KBG and the dam was completely demolished in 1992. From 1992 to 1995 the KBG re¢lled up to its presentday water level.
3. Materials and methods In November 1999 eight cores (6 cm in diameter and 0.80^1.00 m long) were retrieved from the northwest corner of the KBG (at about 3.5 m of water depth) using a piston core from a raft. Two of these cores (KBG-02-01 and KBG-08-01) were split longitudinally for lithological description, digitalization, and sampling. After their visual correlation, KBG-02-01 was used for X-ray dif-
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
fraction and thin sections whereas KBG-08-01 was employed for 210 Pb measurements. X-ray di¡ractions were done with an automatic Brucker D-5005 X-ray di¡ractometer: Cu radiation (KK = 1.5405), 40 kV, 30 mA and graphite monochromator. The core was sampled every 1.5 cm and a total of 59 samples were characterized. The mineralogical zones were established using the Edwards and Cavalli^Sforza chord distance method of TILIA software [25]. Thin sections were obtained after freeze-drying and balsam hardening the samples, and 25 thin sections, from 0.10 to 0.95 m depth, were studied. The thin sections were digitalized with a CCD camera, and images were employed for measurements, using Scion Image software in accordance with the technique described in [26]. The water level oscillations of the KBG were reconstructed using the mineralogical composition of sediments, and a statistical approach. A detailed description of this approach can be found in [27]. The reconstructed water level record was investigated with wavelet analysis in order to highlight the possible cyclicities. Two di¡erent wavelets (haar 5 and db 6 level 5) were applied. A total of 762 chemical analyses of the KBG waters covering the period 1900^1976 were obtained from Russian literature [24] and were studied for the degree of water saturation with respect to calcite, aragonite, hydromagnesite, gypsum and mirabilite. The degree of water saturation with respect to these mineral species in the 762 chemical analyses was calculated using the computer software PHREEQC [28,29]. This long geochemical record was used in order to determine the present-day mineralogical precipitation. The 210 Pb dating method applied in core KBG08-01 is based on the technique of [30]. 210 Pb activity was determined through the measurement of its granddaughter 210 Po (proxy method) using K-spectrometry. A known solution of 209 Po was used as spike. Activity of 210 Pb was calculated for each sample allowing an activity^depth graph to be plotted. Eleven samples were measured. The instrumental record of the KBG water level oscillations was obtained from [24] whereas that of the Caspian Sea was obtained from the International Atomic Energy Agency (IAEA).
229
4. Results 4.1. Present-day geochemical evolution of the KBG ecosystem The water chemistry of this bay records a predominance of Cl3 among the anions and Naþ and Mg2þ among the cations, with a high density ( s 1.2 g/cm3 ). pH values range between 7.2 and 9. The temperature of the KBG water also records considerable oscillations, ranging from 6 4‡C in December to 25‡C in July. These temperature £uctuations are also evidenced by the precipitation of di¡erent mineral phases during the year. Calcite and aragonite usually precipitate in spring, gypsum forms in summer, and large amounts of mirabilite and halite precipitate on the shores of the bay in winter (Fig. 4) [24,32,33]. 4.2. Lithological description of cores KBG-02-01 and KBG-08-01 The sediments of both cores are mainly composed of a centimetric to decimetric alternation of dark grayish silty-clay and light sandy layers, with some intercalations of whitish layers of salt and several levels rich in Cerastoderma sp. shells. The uppermost 7 cm of core KBG-02-01 are composed of a sandy level, mainly made up of homometric ¢ne quartz grains. This sandy level was not observed in core KBG-08-01 and could be attributed to the coring location within the bay ^ core KBG-02-01 is closer to the shore than core KBG08-01. 4.3. Lithological facies Two main microfacies can be distinguished on the basis of petrographical observations : 4.3.1. Massive micritic silty-clays Petrographical observations provide evidence of the micritic character of the silty-clay layers. These layers are massive, without any visible internal structure. The upper and lower boundaries of these layers are sharp and are usually parallel to the surface of sedimentation (Fig. 5a) although some of the layers have a crinkled or domal mor-
EPSL 6681 19-6-03
230
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Fig. 4. Monthly calcite, gypsum and mirabilite saturation index oscillations in the KBG water column for the period ranging between May 1974 and November 1976. Shaded areas indicate oversaturation in the mineralogical species, whereas white areas depict undersaturation. Depths are expressed in negative mm. Missing months have been interpolated. The scale from 0 to 900 indicates time in Julian days. Gray crosses depict the locations of the chemical analyses. Data extracted from [24].
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
231
Fig. 5. (a) Microphotograph of the massive micritic silty-clay facies (84 cm depth). (b) Microphotograph of the massive sandy facies (44 cm depth). (c) Microphotograph of the coarsening upwards sandy layers (65 cm depth). The white bar of the three microphotographs represents 1 mm.
phology. In some levels there are large and isolated euhedral crystals of gypsum, denoting their diagenetic origin. The micritic levels were attributed to the chemical precipitation of carbonates during the diluted phases of the bay corresponding to high water levels. Layers with a crinkled morphology could correspond to microbial mats. 4.3.2. Sandy layers The sandy layers are mainly formed by gypsum. These layers present two main microfacies on the basis of the grain size and grain distribution : 4.3.2.1. Massive sandy layers. These gypsum sandy layers are constituted by anhedral and homometric crystals of gypsum, with no internal structure (Fig. 5b). The base of these layers is
parallel to the surface of sedimentation, whereas the upper boundary is irregular. The thickness of these layers is relatively constant. Some fragments of biota (mainly ostracods and gastropods) are present at the base of some of these layers. This sandy microfacies was interpreted as reworked primary gypsarenites resulting from the wave action during low lake level episodes. 4.3.2.2. Coarsening upwards layers. This second gypsum microfacies is composed of gypsum crystals (ranging from 10 to 50 Wm) forming coarsening upward sequences. At the bottom homometric gypsum crystals progressively change upwards to elongated and swallow tail twinned gypsum crystals (Fig. 5c). The lower boundary of these layers is di¡use, whereas the contact with the upper
EPSL 6681 19-6-03
232
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Fig. 6. Mineralogical composition of core KBG-02-01. Values expressed in percentage. The four zones were established according to the cluster results.
silty-clays is sharp. This microfacies was interpreted as non-disturbed primary gypsarenites formed during low lake water level episodes. 4.3.3. Facies organization The two main microfacies described above alternate throughout the core and can be interpreted as high-frequency oscillations of the water level. They alternate with the carbonate layer at the base of the cycle and with the sandy layer at the top. 4.4. Mineralogical composition X-ray di¡ractions show that the mineralogical assemblage of the KBG sediments is mainly made
up of sulfates (gypsum and glauberite), carbonates (calcite, aragonite, high-magnesian calcite and hydromagnesite) and terrigenous minerals (quartz, illite, chlorite and feldspars) (Fig. 6). The sediments are also composed of hydrated magnesian (epsomite MgSO4 W7H2 O) and sodium sulfates (mirabilite Na2 SO4 W10H2 O). These mineralogical phases were not quanti¢ed owing to their high solubility and were not considered in the statistical study. The main mineral is gypsum, with a mean value of more than 45% of the total weight composition, ranging from 0 to approximately 100% of the total sediment composition. Calcium sulfate is present throughout the sequence. The carbonates show di¡erent patterns. Hydro-
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
233
magnesite is present in two discrete levels (between 74 and 56 cm and between 26 and 7 cm of depth) and calcite occurs in three discrete levels (between 92 and 80 cm, between 71 and 54 cm, and between 29 and 16 cm of depth) in contrast to aragonite and magnesian calcite, which are more widespread. Total carbonates have a mean percentage of 4.5% of the total weight, ranging between 17.3 and 0%, depending on the depth. The main terrigenous mineral species is quartz (mean percentage about 20%) and the other terrigenous minerals are present with approximately constant values (clinochlorite 3%, microcline 9.3%, illite 10%, and albite 11%), except for the layers where gypsum predominates. Cluster analysis shows four mineralogical zones. Zones 4 and 2 are mainly composed of terrigenous minerals, whereas zones 3 and 1 are dominated by gypsum (Fig. 5). Fig. 7. Total 210 Pb values measured in core KBG-08-01. Units in Bq/kg of dry sediment.
4.5. Chronology 210
Pb is a naturally occurring radionuclide formed as a decay product of the 238 U decay series. The 210 Pb dating technique is well established for dating marine and lacustrine sediments deposited over the last 100^150 years and is based on a measurement of the ‘excess’ or unsupported 210 Pb activity incorporated into the sediments [31,34^ 36]. This dating method has also been successfully applied to recent dried lakes [37]. The 210 Pb pro¢le (Fig. 7) shows a decreasing curve, from 6.5 Bq/kg at the top of the core to 0.3 Bq/kg at 80 cm depth. At 51 cm depth the 210 Pb content increases to 4.1 Bq/kg, oscillates slightly, and at 73 cm depth the 210 Pb content abruptly decreases to 0.6 Bq/kg. The lowermost two samples have a 210 Pb content of 0.3 Bq/kg. These lowest values were considered as supported 210 Pb, and thus the background for the KBG. These 210 Pb £uctuations recorded at 61 and 53 cm depth may be attributed to lithofacies changes. The sediments change downwards from a centimetric silty-clay alternation to ¢ne gypsum silty layers rich in biota remains. Thin sections through the silty sediments do not indicate the presence of erosive discontinuities. The water level oscillations of the hypersaline
ecosystems trigger the chemical precipitation of endogenic minerals. This chemical precipitation is mainly produced during the water level decline phases, when there is the progressive concentration of the water allowing the progressive supersaturation of the di¡erent mineral species. During these phases, the 210 Pb fallout is ¢xed within the bottom sediments. But during the re¢lling phases the input of more diluted water (or brackish water in the case of the Caspian Sea^KBG ecosystem) provokes the partial dissolution of the uppermost salt pan, which is composed of highly soluble salts, such as halite, or mirabilite and bischo⁄te in the case of the KBG [38^40]. The partial dissolution of the salt pan implies that the 210 Pb previously deposited is concentrated in the insoluble remaining fraction, giving an ‘anomalous’ high value. This 210 Pb enrichment is only possible with sediments younger than 150 years. With sediments exceeding 150 years only 210 Pb background values will be obtained since these are in equilibrium. Thus, the upper 70 cm of the KBG sedimentary in¢ll are younger than 150 years. Despite these 210 Pb content £uctuations, a sedimentation rate model can be constructed using
EPSL 6681 19-6-03
234
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
the ¢ve uppermost samples and the background values. This equation allows us to calculate the depth at which the unsupported 210 Pb appears. Given this equation, and considering the background value of the KBG as 0.3 Bq/kg, the increase in 210 Pb should be situated at 68 cm depth. This assumption ¢ts in with the previous working hypothesis. The 210 Pb £uctuation does not allow us to accurately calculate the age of the core depth ranging between 41 and 70 cm. But the age of the upper and lower limit of this zone can be established from the sedimentation rate model calculated for the upper 41 cm and from the core depth of the onset of the unsupported 210 Pb. The upper limit of this core depth would be approximately 1919 AD whereas the lower limit would be approximately 1868 AD. In order to calculate the age equivalence of each sample for this 51 year zone a linear sedimentation rate approach was applied. 4.6. Statistical approach 4.6.1. Correlation matrix In order to describe the relationships between the mineral species, a factor analysis (FA) was carried out over a matrix formed by 59 samples (rows) and 11 mineral species (columns). The following factors may be hightlighted from the correlation matrix (Table 1): 1. the negative correlation existing between the
gypsum and the rest of the mineral species, especially with the terrigenous ones. This suggests that the origin of the gypsum is endogenic, which is also corroborated by observations in thin sections. 2. The positive correlation existing between the illite and the chlorite indicates the same terrigenous source for both mineralogical species. 3. The positive correlation existing between the albite and the microcline denotes the same rock source (granitic rocks commonly found in the catchment area). 4. The positive correlation existing between the magnesian calcite and the hydromagnesite suggests a possible common origin. 4.6.2. Ordination analyses Statistical analyses are useful tools for handling, summarizing and interpreting paleoenvironmental data [41]. Correspondence analyses (CA) and detrended correspondence analyses (DCA) were performed to describe the main sedimentological processes that determine the mineralogical composition of the KBG sediments. CA was initially applied but, since the second eigenvector de¢ned by this analysis is regarded as a quadratic function of the ¢rst [42], DCA were carried out. The ¢rst two eigenvectors of DCA account for more than 55% of the total variance (Table 2). The distribution of the mineral species in the plane de¢ned by the ¢rst two eigenvectors of the DCA (Fig. 8) clearly shows the contrasting
Table 1 Correlation matrix of the mineral species that compose the sediment
Alb Arag Calc Chlr Glau Gyp Hydr Illt MgCa Micr Qtz
Alb
Arag
Calc
Chlr
Glau
Gyp
Hydr
Illt
MgCa
Micr
Qtz
1 0.03 30.13 0.46 30.06 30.75 30.36 0.46 30.15 0.56 0.39
1 30.06 0.09 30.06 30.03 30.17 0.09 30.17 0 30.13
1 30.11 0.15 30.04 0.49 30.15 30.15 30.09 0.18
1 30.08 30.61 30.25 0.89 30.13 0.26 0.21
1 0.09 30.07 30.07 30.08 30.1 0.01
1 0.07 30.68 30.06 30.58 30.74
1 30.3 0.63 30.13 30.06
1 30.07 0.34 0.29
1 0.00 0.00
1 0.13
1
In bold the signi¢cant ( s 0.6) correlations. Alb, albite; Arag, aragonite; Calc, calcite; Chlr, chlorite; Glau, glauberite; Gyp, gypsum; Hydr, hydromagnesite; Illt, illite; MgCa, magnesian calcite; Micr, microcline; and Qtz, quartz.
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239 Table 2 Eigenvectors de¢ned by DCA and their percentage of explained variability Factor
Eigenvalue
% of variability
% cumulated
1 2 3 4
3.87 2.23 1.3 1.16
35.2 20.3 11.8 10.6
35.2 55.5 67.3 77.9
behavior of the terrigenous and the endogenic minerals. Furthermore, the endogenic mineralogical species are ordered in the same way as the theoretical sequence obtained from a concentration process [43,27]. The second eigenvector focuses on the details within the carbonate species ; the magnesian mineralogical species (magnesian calcite and hydromagnesite) are present at the positive end of this eigenvector whereas aragonite occurs at the negative end. According to the mineral species distribution with respect to the ¢rst eigenvector and from right to left, they can be grouped in four ‘mineral groups’: the ‘clay group’ (clinochlorite and illite), the ‘coarser terrigenous group’ (albite, microcline and quartz), the ‘carbonate endogenic group’ (aragonite, calcite, magnesian calcite and hydro-
235
magnesite) and the ‘sulfate endogenic group’ (gypsum). Two main patterns can be identi¢ed when plotting these mineral groups with respect to their depth (Fig. 9): b Behavior 1: Relative low percentages of the sulfate group with respect to the terrigenous ones. Also, a covariance between the sulfate endogenic and the clay groups whereas the coarser terrigenous group shows an inverse mode with respect to these groups. This behavior was observed in zones 4 and 2, which were de¢ned by the cluster analyses. b Behavior 2: Main predominance of the sulfate group over the other groups. This behavior can be observed in zones 3 and 1, which were de¢ned by the cluster analyses.
5. Discussion 5.1. The KBG water level reconstruction According to the chronological framework established by the 210 Pb method, the top meter of the sedimentary in¢ll of the KBG records the environmental history of the last 200 years.
Fig. 8. Plot of the mineralogical species in the plane de¢ned by the ¢rst two eigenvectors obtained from the DCA analysis.
EPSL 6681 19-6-03
236
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Fig. 9. Plot of the three mineral groups de¢ned by the DCA analysis: sulfate endogenic group (upper diagram), coarser terrigenous group (middle diagram), and clay group (lower diagram) for the period 1811^1998. Comparison of these groups with the instrumental Caspian Sea water level record (bold continuous line).
The behavior of the mineral groups may be interpreted from the chronological framework and from the comparison with the instrumental water level record of the Caspian Sea (Fig. 9). Slight water level decreases, such as those recorded in 1905^1910 and in 1920^1925, are recorded in the KBG sedimentary record as small sulfate and clay group increases (Behavior 1) whereas the drastic water level decrease that occurred during the 1930s and 1940s up to the late 1970s is recorded by the dominance of the sulfate group (Behavior 2). Moreover, the recent facies distribution reported in the Aral Sea provides evidence that the gypsum clayish sediments (Behavior 1) correspond to the deepest points of several sub-basins [44]. Coarser sediments are always located in the margins of the lake. The four mineral groups were established ac-
cording to the ¢rst eigenvector of the DCA analyses. Thus this eigenvector re£ected changes in the endogenic terrigenous mineral phases. The fact that they matched the Caspian Sea water level oscillations indicates that these mineral changes are triggered by these oscillations. Thus, the ¢rst eigenvector re£ects £uctuations of the water level in the KBG. According to the technique established previously [27], the distance of every sample to the ¢rst eigenvector will represent a relative water level. The stratigraphical position of the whole set of samples will allow us to reconstruct the evolution of the relative water level of the KBG over time (Fig. 10). High DCA values indicate relative high lake level, and vice versa. The high degree of correspondence between the reconstructed and the instrumental water level oscillations (r2 = 0.83) is remarkable.
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
237
Fig. 10. Comparison of the reconstructed water level oscillations using the mineral record (thin line) with the instrumental Caspian Sea water level record (bold line) for the period 1811^1998. The reconstructed curve has arbitrary units whereas the measured one is in m below the Baltic Sea level.
This reconstruction shows a highly oscillating record, with long and short trends in terms of high and low water level periods. The reconstruction records 13 lake level minima in 1814, 1820^ 23, 1834, 1852^53, 1860, 1867^68, 1872^82, 1892^ 95, 1901^02, 1920^26, 1940^67, 1975^80, and in 1986^1995, recording the lowest lake level in 1953, and 15 lake level maxima in 1818, 1830, 1850, 1856^57, 1862, 1869, 1888^90, 1898^00, 1911, 1918, 1927^28, 1950, 1972, 1982 and 1988^89, recording the highest lake level in 1928. The reconstructed water level also records longterm oscillations. From 1812 to 1819 the KBG record indicates a high water level phase. This phase was followed by a period (1820^1839) when the KBG water level was generally low. Between 1840 and 1863 the water levels were high again, and between 1864 and 1863 the KBG water levels were low, although there was a short highstand of 5 years (1886^1890). From 1897 up to 1942 the lacustrine system records an important highstand and from 1943 up to the present-day (1998) the reconstructed water level has been low. Wavelet analysis was performed in order to highlight the possible periodicities present in the reconstructed water level record (Fig. 11). From this analysis, and despite the short length of the analyzed series (187 years), the existence of two clear periodicities of 62.5 and 38.46 years should be pointed out. The ¢rst periodicity matches the one already described of 65 years in the Caspian
Sea [45], whereas the second one is close to that of V35 years commonly recognized in Europe and Central Asia [46]. Although the origin of this 33 year periodicity is unclear, recent works have tentatively attributed it to multidecadal interactions between the NAO and the North Atlantic thermohaline circulation [47]. It should be noted that these water oscillations show an evident cyclic pattern rather than an ‘erratic’ one despite the fact that the mechanisms triggering such water level oscillations remain unclear.
Fig. 11. Periodogram of the reconstructed Caspian Sea water level record for the period 1811^1998.
EPSL 6681 19-6-03
238
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
Acknowledgements The authors are grateful to G. Seret (RMCA, Belgium), I. Kalugin, V. Zykin (UIGGM, Russia) and P. Tucholka (Paris I, France) for organizing, performing and assisting the coring campaign, and G. Yennick and D. Lobrov (Ministry of Nature Use and Environmental Protection, Republic of Turkmenistan) for local assistance. We are indebted to Judith Firth (Brunel University) for performing the 210 Pb measurements, to Irina Overeem, Salomon Kroonenberg and Ian Boomer for their useful comments and suggestions that greatly helped to improve the manuscript, and to George von Knorring for revising the English. Financial support was provided by EU project (IC15-CT96-0112 ‘Understanding the Caspian Sea erratic £uctuations’).[AC]
[9]
[10] [11]
[12]
[13]
[14]
[15]
References [16] [1] R.W. Battarbee, Palaeolimnological approaches to climate change, with special regard to the biological record, Quat. Sci. Rev. 19 (2000) 107^124. [2] C. Bigler, R.I. Hall, Diatoms as indicators of climatic and limnological change in Swedish Lapland: a 100 lake calibration set and its validation for paleoecological reconstruction, J. Paleolimnol. 27 (2002) 97^115. [3] F. Gasse, S. Juggins, L.B. Khelifa, Diatom-based transfer functions for inferring past hydrochemical characteristics of African lakes, Palaeogeogr. Palaeolimnol. Palaeoecol. 117 (1995) 31^54. [4] S. Leroy, Climatic versus non climatic lake-level changes from a Plio-Pleistocene lacustrine complex of Catalonia (Spain): palynology of the Tres Pins sequences, J. Paleolimnol. 17 (1997) 347^367. [5] S. Leroy, S. Giralt, P. Francus, G. Seret, The high sensitivity of the palynological record in the Vico maar lacustrine sequence (Latium, Italy) highlights the climatic gradient through Europe for the last 90 ka, Quat. Sci. Rev. 16 (1996) 189^201. [6] R. Julia', F. Burjachs, M.J. Das|¤, F. Mezquita, M.R. Miracle, J.R. Roca, G. Seret, E. Vicente, Meromixis origin and recent trophic evolution in the Spanish mountain lake La Cruz, Aquat. Sci. 60 (1998) 279^299. [7] S. Giralt, F. Burjachs, J.R. Roca, R. Julia', Late Glacial to Early Holocene environmental adjustment in the Mediterranean semi-arid zone of the Salines playa-lake (Alicant Spain), J. Paleolimnol. 21 (1999) 449^460. [8] J.P. Smol, I.R. Walker, P.R. Leavitt, Paleolimnology and
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
hindcasting climatic trends, Verh. Int. Ver. Theor. Angew. Limnol. 24 (1991) 1240^1246. G.S. Golitsyn, The Caspian Sea Level as a problem of diagnosis and prognosis of the regional climate change, Atmos. Ocean. Phys. 31 (1995) 366^372. S.N. Rodionov, Global and Regional Climate Interaction, Kluwer Academic Press, Dordrecht, 1994, 241 pp. V.M. Lyatkher, Solar cycle length stochastic association with Caspian Sea level, Geol. Res. Lett. 27 (2000) 3727^ 3730. K. Arpe, L. Bengtsson, G.S. Golitsyn, I.I. Mokhov, V.A. Semenov, P.V. Sporyshev, Connection between Caspian Sea level variability and ENSO, Geol. Res. Lett. 27 (2000) 2693^2696. A.V. Kislov, G.V. Surkova, Simulation of the Caspian Sea Level changes during the last 20,000 years, in: G. Benito, V.R. Baker, K.J. Gregory (Eds.), Palaeohydrology and Environmental Change, Wiley, New York, 1998, pp. 235^244. A.V. Frolov, New methods for managing Caspian Sea level £uctuations, in: W. Asher, N. Mirovitskaya (Eds.), The Caspian Sea: A Quest for Environmental Security, ASI NATO Science Series, Environment 67, 2000, pp. 79^ 90. S. Leroy, F. Marret, F. Gasse, F. Chalie¤, Understanding the Caspian Sea erratic £uctuations: palynological results from the south basin, Terra Nostra 7 (2000) 45^49. R.K. Klige, M.S. Myagkov, Changes in the water regime of the Caspian Sea, GeoJournal 27 (1992) 299^307. F. Peeters, R. Kipfer, D. Achermann, M. Hofer, W. Aeschbach-Hertig, U. Beyerle, D.M. Imboden, K. Rozanski, K. Fro«hlich, Analysis of deep-water exchange in the Caspian Sea based on environmental tracers, Deep-Sea Res. 47 (2000) 621^654. I.A. Shiklomanov, V. Georgievski, Z.D. Kopaliani, Water balance of the Caspian Sea and reasons of the water level rise in the Caspian Sea, Workshop Report 28, Supplement, UNESCO, Intergovernmental Oceanographic Commission, Paris, 1995, pp. 1^27. P. Alpert, B.U. Neeman, Y. Shay-El, Climatological analysis of Mediterranean cyclones using ECMWF data, Tellus 42A (1990) 65^77. H.M. Cullen, P.B. DeMenocal, North Atlantic in£uence on Tigris^Euphrates stream£ow, Int. J. Clim. 20 (2000) 853^863. N. Clauer, S. Chaudhuri, T. Toulkeridis, G. Blanc, Fluctuations of Caspian Sea level: beyond climatic variations?, Geology 28 (2000) 1015^1018. P.A. Kaplin, A.O. Selivanov, Recent coastal evolution of the Caspian Sea as a natural model for coastal responses to the possible acceleration of global sea-level rise, Mar. Geol. 124 (1995) 161^175. S.B. Kroonenberg, G.V. Rusakov, A.A. Svitoch, The wandering of the Volga delta: a response to rapid Caspian sea-level change, Sediment. Geol. 107 (1997) 189^ 209. I.N. Lepeshevkov, D.V. Buynevich, N.A. Buynevich, G.S.
EPSL 6681 19-6-03
S. Giralt et al. / Earth and Planetary Science Letters 212 (2003) 225^239
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
Sedel’nikov, Perspectives of Use of Salt Resources of Kara-Bogaz-Gol, Academia Nauka, Moscow, 1981, 273 pp. (in Russian). E. Grimm, A Fortran 77 program for stratigraphically constrained cluster analysis by the method of incremental sum of squares, Comp. Geosci. 13 (1987) 13^35. P. Francus, An image analysis technique to measure grain-size variation in thin sections of soft clastic sediments, Sediment. Geol. 121 (1998) 289^298. X. Rodo¤, S. Giralt, F. Burjachs, F.A. Com|¤n, R.G. Tenorio, R. Julia', High-resolution saline lake sediments as enhanced tools for relating proxy paleolake records to recent climatic data series, Sediment. Geol. 148 (2002) 203^220. D.L. Parkhurst, User’s guide to PHREEQC-a computer program for speciation, reaction-path, advective-transport, and inverse geochemical calculations, Water Resources Investigations Report 95-4227, US Geological Survey, 1995, 143 pp. S.R. Charlton, C.L. Macklin, D.L. Parkhurst, PHREEQCI ^ a graphical user interface for the geochemical computer program PHREEQC, Water Resources Investigation Report 97-4222, US Geological Survey, 1997, 9 pp. W.W. Flynn, The determination of low levels of Polonium-210 in environmental materials, Anal. Chim. Acta 43 (1968) 221^227. A.B. Cundy, I.W. Croudace, Physical and chemical associations of radionuclides and trace metals in estuarine sediments: an example from Poole Harbour, Southern England, J. Environ. Radioact. 29 (1995) 191^211. S. Giralt, R. Julia', F. Gasse, G. Seret, P. Tucholka, V.I. Ferronsky, S. Leroy, F. Chalie¤, A. Jelinowska, M.C. Pierret, V.Y. Lavrouchine, First results of the multidisciplinary study of the Kara-Bogaz Gol sediments (Republic of Turkmenistan, Central Asia), in: 8th International Symposium on Paleolimnology, 2000, pp. 31^32. S. Giralt, R. Julia', F. Gasse, S. Leroy, F. Chalie¤, Highresolution reconstruction of the KaraBogazGol^Caspian Sea water level evolution for the last 200 years based on mineralogical changes, PAGES-PEPIII: Past Climate Variability Through Europe and Africa, Aix-en-Provence, 2001. F. Old¢eld, P.G. Appleby, Empirical testing of 210Pbdating models for lake sediments, in: E.Y. Haworth, J.W.G. Lund (Eds.), Lake Sediments and Environmental History, University of Minnesota Press, Minneapolis, MN, 1984, pp. 93^124.
239
[35] P.G. Appleby, P. Nolan, D.W. Gi¡ord, M.J. Godfrey, F. Old¢eld, N.J. Anderson, R.W. Battarbee, 210 Pb dating by low background gamma counting, Hydrobiologia 141 (1986) 21^27. [36] J. Vesely¤, H. Almquist-Jacobson, L.M. Miller, S.A. Norton, P. Appleby, A.S. Dixit, J.P. Smol, The history and impact of air pollution at Certovo Lake, southwestern Czech Republic, J. Paleolimnol. 8 (1993) 211^231. [37] M.S. Al-Masri, A. Aba, H. Khalil, Z. Al-Hares, Sedimentation rates and pollution history of a dried lake: AlOteibeh Lake, Sci. Total Environ. 293 (2002) 177^189. [38] J.M. Bowler, Australian salt lakes: a palaeohydrologic approach, Hydrobiologia 82 (1981) 431^444. [39] W.B. Langbein, Salinity and hydrology of closed lakes, U.S.G.S. Prof. Paper, vol. 412, U.S. Gov. Print O⁄ce, Washington, DC, 1961, p. 220. [40] F.A. Com|¤n, R. Julia', M.P. Com|¤n, F. Plana, Hydrogeochemistry of Lake Gallocanta (Arago¤n, NE Spain), Hydrobiologia 197 (1990) 51^66. [41] H.J.B. Birks, Numerical zonation, comparison and correlation of Quaternary pollen-stratigraphical data, in: B.E. Berglund (Ed.), Handbook of Holocene Palaeoecology and Palaeohydrology, Wiley, New York, 1986, pp. 743^ 774. [42] R.H.G. Jongman, C.J.F. ter Braak, O.R.F. van Tongeren, Data Analysis in Community and Landscape Ecology, Pudoc, Wageningen, 1987, 299 pp. [43] H.P. Eugster, L.A. Hardie, Saline lakes, in: A. Lerman (Ed.), Lakes: Chemistry, Geology, Physics, Springer, New York, 1978, pp. 237^294. [44] V.A. Zhamoida, V.P. Butylin, E.A. Popova, N.V. Aladin, Recent sedimentation processes in the Northern Aral Sea, Int. J. Salt Lake Res. 6 (1997) 67^81. [45] S.B. Kroonenberg, E.N. Badyukova, J.E.A. Storms, E.I. Ignatov, N.S. Kasimov, A full sea-level cycle in 65 years: barrier dynamics along Caspian shores, Sediment. Geol. 134 (2000) 257^274. [46] D.E. Paulsen, H-Ch. Li, T-L. Ku, Climate variability in central China over the last 1270 years revealed by highresolution stalagmite records, Quat. Sci. Rev., in press. [47] P.E. O’Sullivan, R. Moyeed, M.C. Cooper, M.J. Nicholson, Comparison between instrumental, observational and high resolution proxy sedimentary records of Late Holocene climatic change ^ a discussion of possibilities, Quat. Inter. 88 (2002) 27^44.
EPSL 6681 19-6-03
