Introduction

The modern regression in the Aral Sea region, which started in 1960, has received great attention (e.g., Aral’skij krizis 1991; Létolle and Mainguet 1996; Micklin and Williams 1996; Nihoul et al. 2004; Nurtaev 2004; Ferguson 2003). Due to extensive agricultural cultivation in this area, the abusive exploitation of water resources has not only reduced the lake level but the system reached a critical stage in the water and soil pollution. As a result, a broad scope of studies is required to improve our knowledge on ecosystem functioning, which will help to attenuate environmental and socio-economic risks in future decades. In the context of a policy of appeasement, sustainability rather than searching for new ways to exploit the system should be in the focus of upcoming studies.

But which have been the links between climate variability and ecology in the Aral Sea basin in the past? Due to the climatically exposed continental conditions (semi-arid or even arid over large areas), western Central Asia is potentially vulnerable to minimal environmental change. Therefore extensive deserts in this area are a prominent feature of the region. They spread north of the Kopet Dagh – the Karakum Desert – and west of the mountain ranges of Pamir and Tien Shan, – the Kyzylkum Desert. For settlements, today and in the past, active fresh water resources are restricted to the spatially very constrained flood plains and riverbanks along the Syr Darya and Amu Darya, both of them discharging into the endorheic Aral Sea. However, water resources feeding the rivers seem almost unlimited as, in the headwaters of the rivers, extended glacier systems of the Tien Shan and Pamir Mountains are steadily delivering fresh water. Similar climate conditions not only apply for today, but also have been dominating the landscape throughout most of the Holocene though the Amu Darya has transiently bypassed the Aral Sea and directly discharged into the Caspian Sea (e.g., Tolstov 1962).

The Aral Sea basin hosting one of the Silk Road branches is, therefore, an excellent study area for tracing human agricultural activities but also for highlighting human measures and reactions to past climate changes. To shed light on the past climatic and especially hydrological conditions we launched the INTAS-Project “CLIMAN” (Holocene climatic variability and evolution of human settlement in the Aral Sea basin; http://climan.gfz-potsdam.de/). The interdisciplinary study was designed to analyse natural climatic variations and anthropogenically controlled environmental changes in the past (Boroffka et al. 2003–2004, 2005, 2006; Sorrel et al. 2006, 2007a, b; Austin et al. 2007). In a geomorphologic survey, we focused on previous lake-level changes as recorded in shoreline marks during the last 5,000 years (Fig. 1; Krivonogov et al. 2003; Reinhardt and Wuennemann 2005) and related these observations to archaeological findings (Boroffka et al. 2006).

Fig. 1
figure 1

Reconstruction of past lake levels of the Aral Sea (a) and a close-up of the Pulzhaj area SW of the Aral Sea (b; after Krivonogov et al. 2003). Coloured lines outline different lake levels: blue 54 m asl.; rose 53 m asl.; red 46 m asl.; dashed green and red lines are hypothetical sea levels above 54 m asl. earlier reported by Boomer et al. (2000)

Full size image

However to date, relatively little attention has been granted to connections between local and regional climate changes over the Eurasian region. At a regional scale, past climate variability in the arid Aral Sea basin represents an important key for understanding future climate change, which may affect even more drastically such arid and semi-arid regions. Besides, understanding past climate change it is of great importance to evaluate the anthropogenic impact on present-day and future climates in this highly sensitive semi-arid region.

In this paper we focus on past climatic variations as expressed in humidity and temperature changes and evaluate irrigation activities as inferred from archaeological data. Natural climatic changes evidenced by temperature and precipitation variations during the last 2,000 years are inferred from vegetation remains (pollen grains) studied in a core retrieved from Chernyshov Bay, in the northwestern Large Aral Sea (Fig. 2). We then use geochemical proxies, the relative Ca and Sr abundances (mostly contained in gypsum and authigenic carbonates), as a measure of salinity change and an indication of the origin of fresh water input. Salinity indirectly reflects the river discharge during the last 2,000 years, thus tracing lake level changes (Fig. 2). By looking at the displacement of human settlements we try to rate and estimate the controlling factors, e.g., human versus natural climate forcing. Both, though at different extent, might influence the water balance.

Fig. 2
figure 2

Sediment core from Chernyshov Bay: a Lithology with relative Ca and b Sr concentrations through the core reflecting salinity change within the lake water body; lake level changes related to c human and/or d climate causes e climatic trends as represented by dry versus humid periods

Full size image

Main features of the hydrology in the Aral Sea basin

Today the average annual rainfall ranges from 100 to 140 mm/year (Bortnik and Chistyaeva 1990). The net river discharge, a major fresh water source of the endorheic Aral Sea has changed a lot during the last 100 years. Both rivers Amu Darya and Syr Darya discharged in the range of 56–58 km3/year ± 14 km3 between 1911 and 1960 (Bortnik and Chistyaeva 1990). After 1960 the hydrological cycle of western Central Asia has been disturbed fundamentally by (1) the abusive usage of river water for irrigation and, (2) the building of numerous reservoirs and wide uncovered channels along the two tributaries and into the Karakum Desert. Beginning of 2002 the discharge had been reduced to less than 10 km3/year (Zavialov 2005) and actually Amu Darya contributes less than 1 km3/year (Zavialov, personal communication). The rivers feeding into the lake load in the Tien Shan and Pamir Mountains to the east with meltwater. Lake level changes in the endorheic Aral Sea are preferentially controlled by river discharge of Amu Darya and Syr Darya, which are affected by two main factors (1) climatic change in the headwater system of the rivers and/or by (2) men-controlled irrigation activities. Both play at times a dominating role. Today Global Climate Change affects melting rates in the headwater region but also the precipitation rates and both are increasing due to globally raising average annual temperatures. Coevally evapo-transpiration at high altitudes might increase too (Aizen, personal communication). Besides the riverine drainage, the annual hydrology budget of the endorheic Aral Sea is further controlled by evaporation, which is more than seven times higher than precipitation.

Geomorphologic settings

Geomorphologic surveys and analyses of satellite images reveal distinct shoreline fluctuations in the past (Fig. 1) by tracing positions of ancient littoral terraces, beach ridges, and wave-cut escarpments, and by mapping features such as channel alignments to paleoshorelines and abandoned delta areas. For characterisation of the past local drainage network and lake level reconstruction we combined field observations, and used LANDSAT satellite images together with a digital elevation model (DEM; e.g., Krivonogov et al. 2003; Reinhardt 2005; Reinhardt and Wuennemann 2005). The highest shoreline along the northern Small Aral Sea (Tastubek Bay, Fig. 1) was observed at 53–54 m a.s.l (Boroffka et al. 2006). Due to very smooth topography adjacent and within the delta area, the perimeter at different lake stages shown in Fig. 1a,b documents that lowering of lake level by 1 or 2 m releases large surfaces. For settlement activities even the slightest lake level changes might opened new perspectives as will be shown in the discussion below.

Materials and methods

Sediment coring was performed at Chernyshov Bay, a 22-m deep basin that today, is still attached to the Large Aral Sea located at the northwestern shore of the Aral Sea (Fig. 1). With a Usinger piston corer ( http://www.uwitec.at ), we retrieved sediment cores of 6 and 12 m in length. The description of the core lithology is based on macroscopic and smear slide observations (Sorrel 2006). The relative calcium and strontium abundance in sediment cores was measured by scanning cores with the XRF Core Scanner (®AVAATECH; Jansen et al. 1998) at a 1-cm resolution, representing less than a decadal time resolution. The chronology for the sediment core is based on AMS 14C dates performed on the green alga Vaucheria sp., which are reported as calibrated ages (Nourgaliev et al. 2003; Sorrel et al. 2006).

For tracing moisture and temperature variations in the past (Fig. 3) we studied the vegetation changes. Pollen slide preparation followed the Cour’s method (Cour 1974). A transmitting light microscope using 400× and 1,000× magnifications was used for pollen identification. This was performed using the pollen photograph bank and several atlases of the ‘Laboratoire PaléoEnvironnements et PaléobioSphère’ (Lyon) as well as its pollen database “Photopal” ( http://medias.obs-mip.fr/photopal ). Concentration in palynomorphs varies from <500 to >45,000 grains/g. Minimums of 100 pollen grains, excluding Amaranthaceae-Chenopodiaceae and Artemisia, which are usually over-represented in arid environments, were counted in each sample. Generally more than 25 different taxa were found in each sample, but a total of 79 different taxa have been identified (Sorrel et al. 2007a).

Fig. 3
figure 3

Climatic variations and western Central Asia based on reconstruction from pollen data (Sorrel et al 2007a). MTC: mean temperature of the coldest month (°C); MAT: mean annual temperature (°C); MTW: mean temperature of the warmest month (°C), and MAP: mean annual precipitation (mm/year) in the Aral Sea basin during the last 2,000 years. These climatic parameters are compared to the δ18O record from carbonate deposits in the Soreq Cave, Israel (Schilman et al. 2002)

Full size image

Pollen grains transported either by air or by rivers reflect the local to regional vegetation. Here we present the results inferred from the quantification of pollen data. For the quantification of palaeoclimate signals recorded in plant assemblages, the “probability mutual climatic spheres” (PCS) method described in detail by Klotz and Pross (1999) and Klotz et al. (2003, 2004) was used. The main restriction in applying modern analogue methods in this study is the general poorness of the available database of surface pollen spectra from the Aral Sea region (only 91 in Kazakhstan, Tarasov et al. 1998), which may serve as modern analogues for reliable climate reconstructions. Hence the use of the PCS method is more suitable than modern analogues for the reconstruction of climate change in the Aral Sea basin during the last 2,000 years. For more compelling information on the application of the PCS method to our case study, we refer to Sorrel et al. (2007a).

Besides, we investigated the northern and southern shores of the Aral Sea for archaeological remains during two expeditions in 2002 and 2003. During these studies, archaeological traces were positioned by geographic positioning system (GPS) and dated by (1) conventional archaeological methods and, (2) by comparing archaeological assemblages to radiocarbon-dated material from analogous sites (Boroffka et al. 2006). The complete repertoire of archaeological finds (the toolkit) was studied to reconstruct the economic basis through time. Previously collected archaeological data is re-evaluated in light of new informations, and placed within the context of results from both palaeontological and geological investigations.

Lithology: Gypsum and Sr content as hydrological indicators

From sediments retrieved at Chernyshov Bay, in the northwestern Aral Sea, we used dinoflagellate cysts and diatoms, both sensitively reacting on salinity changes, for tracing variations in lake levels (Sorrel et al. 2006; Austin et al. 2007). Furthermore, the sediments, mostly consisting of dark clayey and silty muds with changing organic contents (Sorrel 2006), contain four levels of gypsum-rich mud, forming distinct light-coloured beds or dark clayey muds with abundant idiomorphic gypsum crystals (Fig. 2, in Sorrel et al. 2006). The observed lithological changes are coeval to the increase of the relative content in calcium (Boroffka et al. 2006). Gypsum (CaSO4·2H2O) precipitates when salinity reaches concentrations beyond a certain level; it is, therefore, a typical indicator of progressive salinization, and we use the element scans of calcium to keep track of the lake’s desiccation state. In the Aral Sea, the stage of over saturation for gypsum was reached in the early 1960s when salinity exceeded 26–28 ppt (Bortnik and Chistyaeva 1990). Thus, we can use the relative abundance of Ca to estimate beginning of changes in the river inflow, which control, together with the annual evaporation (approximately 100 cm/year), the water balance and at a longer time range the lake level of the Aral Sea. In Fig. 2, the relative calcium abundance indicates distinct changes through the last ca. 2,000 years. High relative calcium contents are taken as indication for rising salinities and thus lowering lake levels. The acme of the lake level lowering, with precipitation of chloride salts is however following the Ca–sulphate maximum and thus not reported in Fig. 2a. According to the age model, these events culminate between ca. around a.d. 600, a.d. 1220, a.d. 1500 and slightly elevated values occur at about a.d. 750, a.d. 1000, and a.d. 1800. Between a.d. 0 and a.d. 300 (?) we observe an increased variability with average values similar to the event at a.d. 600. Most of these events match with salinity changes as reflected by dinoflagellate cyst assemblages. With dinoflagellate cysts we also have two phases of higher salinity at ca. 1500 and ca. a.d. 1600/1650, which is matching well with results obtained on ostracods by Boomer et al. (2003), though time resolution of the latter is coarser (Sorrel et al. 2006). Lake-level low stands from the eastern basin dated to approximately a.d. 350–450 and a.d. 1550–1650 (Rubanov et al. 1987; Maev and Karpychev 1999) are also concurrent with our observations. Comparing the salinity record (Fig. 2a) with the precipitation as reconstructed from pollen, lower precipitation falls into intervals when salinity is elevated (Figs. 2 and 3).

Another element, strontium, may further be used for reconstructing the river discharge in the Aral Sea basin. Indeed recent studies on river water have shown that only Syr Darya, due to the specific rock composition in the watershed, discharges dissolved Sr into the Aral Sea. Therefore we use the relative Sr content in the sediment core to discriminate between the water discharge of Amu Darya and Syr Darya. In Fig. 2 the relative abundance of Ca and Sr show enrichments of six to eight times above average levels within the time window at around 1200 a.d. followed by a second spike at about a.d. 1220. During the second half of the past century the increase of Sr is by far higher than for Ca too in between Sr shows only some minor fluctuations, like around a.d. 600 and a.d. 1400. We may therefore infer, that during short intervals between ca. a.d. 1200 and a.d. 1250 and since a.d.1970 onwards, the Syr Darya discharge into the Aral Sea has increased disproportionately.

Vegetation changes as climate indicator

The pollen-based temperature and precipitation reconstructions document shifts in vegetation development that reflect climate changes in western Central Asia linked to the atmospheric circulation in the eastern Mediterranean region (Fig. 3). In the Aral Sea basin, the climate during the first few centuries a.d. according to our chronology was featured by cold winter temperatures, relatively cool summers and arid conditions (mean annual rainfall <300 mm). Similar conditions were reported from Syria with reduced winter / spring rains (Bryson 1996), while a pronounced decrease in humidity and lake levels occurred in Lake Van (Turkey) between ca. 1500 b.c. and a.d. 0 (Landmann et al. 1996; Lemcke and Sturm 1996). In Soreq Cave (Israel), the time span a.d. 0–400 is characterized by decreasing precipitation (Schilman et al. 2002) implying a diminishing activity of cyclones in the eastern Mediterranean region. This arid phase was followed by a general increase in temperature and moisture conditions between ca. a.d. 400 and a.d. 900, although some fluctuations in the precipitation (MAP) and temperature (MTW) records are evident especially around a.d. 550–600 and a.d. 650–700. This milder period (cool winters, warm summers), however, was probably favourable for the development of some arboreal vegetation in the less dry edaphic areas and the replacement of subdesertic herbs by steppe vegetation (Sorrel et al. 2007a). This is coevally a period of elevated temperatures and favourable conditions for the growth of trees in Israel (Lipschitz et al. 1981), linked to increased rainfall over the eastern Mediterranean (Bar-Matthews et al. 1998; Schilman et al. 2002). While by ca. a.d. 900–1150, climate switched back to drier and somewhat cooler conditions in the Aral Sea basin, with cold winter temperatures (−7 to −10°C), cool summers (15–21°C) and reduced precipitation rates (<250 mm/year). We attribute this increase in aridity to reduced humidity transported from the eastern Mediterranean to western Central Asia during a period of lowered rainfall in winter and early spring in Israel (Issar et al. 1991), as inferred by higher δ18O values in Soreq Cave carbonate deposits (Schilman et al. 2002).

After ca. a.d. 1150, reconstructed climatic parameters suggest the onset of moister conditions in western Central Asia with increased humidity (370–500 mm year−1) and warmth (mean annual temperature: 7–11°C), which were maintained until the beginning of the fifteenth century a.d. (“Medieval Warm Period”). These conditions probably favoured the establishment of a riparian forest vegetation and the regression of both steppe herbs and shrubs in the Aral Sea region, coevally with the development of some trees along river valleys. At the onset of more humid conditions in the Aral Sea basin, the climate signal from the Eastern Mediterranean indicates a conspicuous decline in terrestrial δ18O values from the Soreq Cave. This is interpreted as strong positive rainfall anomalies in Israel (Schilman et al. 2002), reflecting enhanced cyclonic activity in the eastern Mediterranean during winter and early spring. It also coincides with high-stand levels of the Dead Sea (Issar et al. 1991) and the Sea of Galilee (Frumkin et al. 1991) implying thus elevated moisture conditions. From ca. a.d. 1450 to a.d. 1550, the Aral Sea region probably experienced a brief aridification with colder and drier conditions. This short-term interval has been identified in other regions within Eurasia (Mackay et al. 2005), and corresponds to the Little Ice Age. In the eastern Mediterranean region, it is contemporaneous with higher δ18O values from the foraminifera G. ruber (Schilman et al. 2002) from marine deposits off Israel (Bar-Matthews et al. 1998; Schilman et al. 2002), indicating lower sea surface temperatures and thus lowered moisture transported to western Central Asia. From a.d. 1550 onwards (to ca. a.d. 1970), the pollen reconstruction document a progressive warming of the climate and corroborate evidence of enhanced aridity and higher temperatures during recent decades. For the last ca. 50 years, our reconstructed values are in accordance with present-day instrumental data from Central Asia (Sorrel et al. 2007a).

The climate reconstruction provides evidence that centennial-scale events are recorded for the last 2,000 years, implying that the precipitation pattern in the Aral Sea basin is strongly linked to atmospheric changes in the eastern Mediterranean region, which modulates the moisture distribution towards the Middle East and western Central Asia. The climatic reconstruction is in agreement with former paleoenvironmental, historical and archaeological studies from Russian scientists who attributed the decline of the classical and medieval civilization of Turkestan to an enhanced aridity (Shnitnikov 1969; Doluhanov 1985; Varuschenko et al. 1987). For the last 2,000 years, however, no human activity exerting control on vegetation change has been detected from the pollen record of Chernyshov Bay. Hence a pervasive and strong impact of human-induced processes on climatic and environmental changes in western Central Asia during the last millennia, as suggested by Kharin et al. (1998), is unlikely.

Archaeological features and water availability

As to the settlement history there is a tight link between climate and hydrology. Levina (1998) and Boroffka et al. (2003–2004, 2005, 2006) have recently demonstrated that ancient settlement development is quite well understood for the region south of the lake. Although the CLIMAN expeditions have filled many chronological gaps in the human settlement history, the knowledge of human life along the northern shore is still patchy. Human settlement around lake Aral begins in the middle Palaeolithic (50,000–35,000 b.c.). After a long period of missing information, a gap that is not yet well understood, Neolithic (5,000–3,000 b.c.) sites of the Kel’teminar culture are well known north and south of the lake. During Eneolithic and Early Bronze Age (3,000–2,000 b.c.) the Kel’teminar culture remains restricted to the south. Along the northern shore settlers change their cultural and economic orientation. They used large projectile points for hunting big game which shows analogies to finds from the forest and forest-steppe zones of the Tobol-Ishim-Irtysh River system and the eastern Ural Mountains, therefore indicating that forest and forest-steppe zones had extended further to the south. This vegetation change heralds more humid conditions, as humans did not apply irrigation systems in this region at that time (Boroffka et al. 2003–2004, 2005, 2006).

From the classical Bronze Age (beginning around 2,000 b.c.), the earliest canals for irrigation agriculture are known from the southern foreland of the Aral Sea (Fig. 4). It is therefore from this period onwards, that the human factor needs to be taken into account when studying hydrological changes in the Aral Sea basin.

Fig. 4
figure 4

Settlement and irrigation systems of the Bronze Age (4,000–3,000 b.p.) at Kokcha, site 15: 1 pit-dwellings, 2 surface dwellings, 3 irrigation canals (adapted from Itina 1977)

Full size image

Settlement economy and irrigation

There is general agreement that, before the incising human involvement beginning in 1960, fluvial discharge by Syr Darya and Amu Darya was sufficient to maintain the endorheic Aral Sea during several thousand years. However, from historical descriptions we understand that already in the past some cultures populating Turkestan used the river water in a more abusive than in a sustainable way (Tolstov 1962). Hence depending on the population density and the economic power, the irrigation in the riverbanks and flood plains of Amu Darya and Syr Darya may, besides the climate, have put additional pressure on the regional hydrology already in the past, a mechanism we are going to highlight below.

For the Neolithic Kel’teminar culture (5,000–3,000 b.c.) the toolkit (with harpoon insets and small projectile points) and the site distribution along ancient river courses and around former lakes indicate an economy in which hunting and fishing played a major role (Boroffka et al. 2003–2004, 2006). For the following period, especially in the north, we observed large projectile points, which indicate a reorientation towards hunting of larger animals (e.g., horses). This proves that economic strategy has been reverted and adapted to a forest steppe environment. The expansion of the forest steppe zone towards the south reflects a more humid climate between 3,000 and 2,000 b.c. By this time men did not influence the environment to any significant point, but rather reacted to natural environmental changes adapting settling and economic strategies (Boroffka et al. 2003–2004, 2005, 2006).

During the Bronze Age (2,000–1,000 b.c.) man became more independent of the environment, and the first irrigation channels have been documented near the southern Aral Sea (Itina 1977; Fig. 4). However these systems were of local scale and did not, therefore, have any major effect on the hydrology of the Aral Sea. Besides, water level at this time was low, as could be observed in a coeval settlement located at the southeastern part of the Aral Sea, which was later buried under lake sediments. Moreover, a high density of Neolithic settlements may be observed along the Uzboj channel. We conclude that, due to a change in the river course before the Neolithic, the Amu Darya had been diverted from the Aral Sea and drained to a great extent the Caspian Sea. A lack of coeval sites along the Akcha Darya channel to the north of the Khorezm basin indicate that this channel was not active at that time, hence confirming this interpretation. Conditions changed during the Bronze Age, when intense settling occurred along the Akcha Darya (now clearly carrying water into the Aral Sea), while river flow almost ceased along the Uzboj (Tolstov 1962; Itina 1977; Boroffka et al. 2005, 2006). These shifts in water-courses may have been due to tectonic uplift in the lower Amy Darya river valley, as observed in other regions of Central Asia (Gentelle 1989). However there are no early historical sources referring to earthquakes, which could help to unravel the chronology of geodynamic events. Further research is needed before this aspect can be evaluated at large.

Irrigation agriculture was intensified during the first Millennium b.c. and reached a first maximum extension during Classical Antiquity (fourth Century b.c.–a.d. fourth Century; Figs. 2d and 5). Traces of ancient canals, some more than 20 m wide and stretching over many kilometres, have been observed. According to Tolstov (1962), Andrianov (1969, 1991), Gerasimov (1978), Baipakov and Groshev (1991) and Levina and Ptichnikov (1991) the network of canals would allow to irrigate 5–10 million ha an area equivalent to the northern and southern banks of the Amu Darya, the Khorezm Basin, the Sarykamysh Basin and the entire area between the Syr Darya and the Zhana Darya (Fig. 6). A regression of the Aral Sea has been reconstructed for this period (Aladin and Plotnikov 1995; Boomer et al. 2000; Sorrel et al. 2006). The increased irrigation activity possibly helped to accelerate the desiccation of the Aral Sea during Classical Antiquity, as increased aridity at this time has been observed in several regions of the Old World (Bryson 1996; Landmann et al. 1996; Lemcke and Sturm 1996; Schilman et al. 2002; Schmidt and Gruhle, 2003). The irrigated surface in this region surprisingly was of similar extent, 6.5 million ha, when the Soviet Union collapsed (Boomer et al. 2000). Today we know that irrigation is indeed one of the factors controlling the extent of lake regression.

Fig. 5
figure 5

Settlements and irrigation systems near Bazar-kala. a general plan, 1 Bronze Age canals, 2 canals of the seventh to fifth Century b.c., 3 canals of Antiquity (fourth Century b.c. – a.d. fourth Century), 4 Earthworks connected to irrigation, 5 rivers, 6 Bronze Age settlements, 7 settlements of the seventh to fifth Century b.c., 8 settlements of earlier Antiquity (fourth Century b.c. – a.d. first Century), 9 settlements of later Antiquity (a.d. second – fourth Century); b Aerial photograph of Bazar-kala with one of the major canals; c cross section of a major canal at Bazar-kala (point 423). a–c adapted from Andrianov 1969

Full size image
Fig. 6
figure 6

Areas with archaeological traces of ancient irrigation systems (shaded) around Aral Sea, with location of some important sites

Full size image

During the medieval period (ninth to fourteenth centuries) irrigation (Figs. 2d and 7) again reached roughly the extent it had during Antiquity (Tolstov 1962; Andrianov 1969, 1991; Gerasimov 1978; Baipakov and Groshev 1991; Levina and Ptichnikov 1991) and could have also influenced the water balance of the lake. These irrigation systems, controlling the river discharge of the Aral Sea to the north and the Sarykamysh basin to the west, were destroyed repeatedly. After the destructions assigned to the Mongol invasion in the year 1221, the Amu Darya appears to have changed its course towards the Uzboj and the Caspian Sea (Barthold 1910). Support comes from the Sr record when we observe increased Sr input indicating that possibly at this time mainly Syr Darya water was discharging into the lake. During the 1370s and 1380s, coeval to Timurs campaigns against Khorezmia, dams and weirs were again destroyed and the Amu Darya filled lake Sarykamysh. However this event did not have a deep impact on the lake’s chemistry like at the beginning of the thirteenth century. Enrichment of Sr was moderate to minor indicating that the Amu Darya was probably only partly deviated towards Lake Sarakamysh and still discharged a major amount of water into the Aral Sea (Fig. 2b). Based on archaeological data it was only during the a.d. fourteenth or early fifteenth centuries when the lake level was well below the present day lake level (Fig. 2c), though it is not recorded in both aquatic (dinoflagellate cysts, diatoms) and geochemical (Ca, Sr) proxies. This mismatch might be due to the actual age model available for this time period, which mostly relies on a correlation with the tree-ring record of Esper et al. (2002) (see Sorrel et al. 2006). This regression is, however, witnessed by the mausoleum at Kerderi situated at 31 m a.s.l. (Fig. 2c,d). Furthermore, it is during this regression when settling activities at Pulzhaj ceased.

Fig. 7
figure 7

1 Aerial view of irrigated field systems near Koi Krylgan Kala (fourth Century b.c. – a.d. fourth Century; adapted from Andrianov 1969), 2 irrigated field system of the a.d. eleventh – thirteenth Century north of Kokcha (CLIMAN Expedition 2003)

Full size image

Both Kerderi and Pulzhaj, until then unknown sites, are for the medieval period important, being only observed for the first time during the CLIMAN expeditions. The mausoleum of Kerderi is located in the northeastern part of the Large Aral Basin while Pulzhaj is situated at the southern end of the former Aibugir Bay on the southwestern edge of the Aral Sea (Figs. 1 and 6). At Pulzhaj the dated archaeological materials together with dated sediments from an adjacent outcrop allow to identify transgressions around the a.d. fourth Century and after a.d. 1400. Between these transgressions, people lived on the lowlands indicating that the water level (53 m a.s.l.) must have been lowered by at least a few meters. Lowering of lake levels, however, should be minimal and would corroborate other observations. According to the dinocysts, the Aral Sea experienced lower salinity levels between ca. a.d. 400 and a.d. 900, suggesting still relatively high lake levels. A short lasting lowering might have occurred at about 600 a.d. as also indicated by increased Ca contents (Fig. 2a).

The Medieval regression seems at least partly be related to human activity (Boroffka et al. 2006; Austin et al. 2007). Dinoflagellate cyst assemblages like the negative sheet-wash index (Sorrel et al. 2006) are rather in favour of a climatically driven Medieval regression starting around a.d. 900 and lasting until about a.d. 1220. A climatically driven decreased river discharge would indeed been confirmed by the tree ring record from the Tien Shan and northwestern Karakorum moutains (Esper et al. 2002), the high catchment areas of the Syr Darya and Amu Darya rivers. During this interval where lowered river discharge are reported, tree-ring growth was minimal, still an indication that summer temperatures were at a minimum. At about 1220 a.d. this regressive event terminates, which may match quite well with pollen data indicating that climate became transiently more humid. However it is at about the time when a drastic event changed the lake’s chemistry. It may be a combination of several events, which fall into this time window. At a.d. 1208 a major earthquake with magnitude above 6 occurred in this area (Nurtaev 2004). This might have caused major damages including destroying the riverbank thus changing the flow direction of Amu Darya towards the Caspian Sea. Furthermore, historical sources document another event, which might have affected the discharge of the Amu Darya. In a.d. 1221 the Mongols destroyed great parts of the irrigation system in the delta plain. As a result both events might have deviated the Amu Darya for some decades during which only the Syr Darya was feeding the lake as shown by the dramatic increase of Sr in the Aral Sea sediments (Fig. 2b).

After a.d. 1550 the pollen reconstruction document a steady warming, with only minor oscillations. Since the nineteenth Century reliable reports exist on irrigation in the Aral Sea region (Sarybaev 2002). At this time water management has been started. At about a.d. 1815–1816 construction of dams is well documented. The first dam was built on the Syr Darya, causing the Tanghi-Daria ( = Zhana Darya?) to fall dry (Murchison and de Khanikoff 1844; Wood 1875; Meiendorf 1975). A following dam cut off the Kuvan Darya (south of the Syr Darya) from the Aral Sea (Butakoff 1853; Michell 1868; Wood 1875). On the Taldyk several dams were constructed (Wood 1875) and remains of dams were observed on the Kuldun (Wood 1875). In a.d. 1857 the Laudan canal, branching west from the Amu Darya, was closed; before a.d. 1863 a dam was built near Khodzeili to close the Kok Uzak (Wood 1875). Some of the actions were politically intended. The Khan of Khiva shot down the Laudan and the Kok Uzak to impose a penalty for nomads or Turkmen raiders. This resulted in the drying up of Aigubir Bay (Butakoff and Michell 1867; Wood 1875 and map). However, this basin was already flooded again by a.d. 1891 (Blanc 1891) and at least the northern part remained under water until a.d. 1931 (Arkhangelskii 1931), which Kropotkin (1904) ascribed to a climatic change. Indeed there is no indication, that the Laudan canal was re-opened at that time. These well-documented records showing abundant manipulations on the irrigation system during the nineteenth Century, match with water levels changes (Sorrel et al. 2006; Austin et al. 2007) and vegetation changes in the sediment cores retrieved from the Aral Sea.

Conclusions

Reconstruction of past climate from proxy data is important for improving constraints on the role and the scope of natural climate variability onto environments that have been persistently influenced by anthropogenic activities. Mostly observations show that the changes in climate was a prominent factor controlling the environment and to a variable extent also human irrigation activities (most probably during the Antiquity and the Middle Ages) exerted some controlling function. But it is only for the last decades, we have compelling evidences that mankind is unequivocally controlling the environmental changes in the Aral Sea basin occurring since the 1960s. Given the different time resolution it is not always unequivocal to match single anthropogenically-influenced with climatically driven events. Highest time resolution is provided from chemical data (XRF scanning data), which allow a resolution better than a few years, while from archaeological toolkits, unless written documents are available, the spatial and temporal resolution is rather sketchy. Therefore the presented interpretations on mutual influences between climate and man are a first attempt to highlight past interaction/reactions between men and climate in this region.