ORIGINAL PAPER

The origin of shallow lakes in the Khorezm Province,Uzbekistan, and the history of pesticide use around theselakes

Michael R. Rosen . Arica Crootof . Liam Reidy . Laurel Saito . Bakhriddin Nishonov .

Julian A. Scott

Received: 4 March 2016 / Accepted: 22 September 2016

� Springer Science+Business Media Dordrecht (outside the USA) 2016

Abstract The economy of the Khorezm Province in

Uzbekistan relies on the large-scale agricultural pro-

duction of cotton. To sustain their staple crop, water

from the Amu Darya is diverted for irrigation through

canal systems constructed during the early to mid-

twentieth century when this region was part of the

Soviet Union. These diversions severely reduce river

flow to the Aral Sea. The Province has [400 small

shallow (\3 m deep) lakes that may have originated

because of this intensive irrigation. Sediment cores

were collected from 12 lakes to elucidate their origin

because this knowledge is critical to understanding

water use in Khorezm. Core chronological data indicate

that the majority of the lakes investigated are less than

150 years old, which supports a recent origin of the

lakes. The thickness of lacustrine sediments in the cores

analyzed ranged from 20 to 60 cm in all but two of the

lakes, indicating a relatively slow sedimentation rate

and a relatively short-term history for the lakes.

Hydrologic changes in the lakes are evident from loss

on ignition and pollen analyses of a subset of the lake

cores. The data indicate that the lakes have transitioned

from a dry, saline, arid landscape during pre-lake

conditions (low organic carbon content) and low pollen

concentrations (in the basal sediments) to the current

freshwater lakes (high organic content), with abundant

freshwater pollen taxa over the last 50–70 years.

Sediments at the base of the cores contain pollen taxa

M. R. Rosen (&)

Water Science Field Team, U.S. Geological Survey, 2730

North Deer Run Road, Carson City, NV 89701, USA

e-mail: mrosen@usgs.gov

A. Crootof � J. A. ScottGraduate Program of Hydrologic Sciences, University of

Nevada Reno, 1664 N. Virginia Street,

Mail Stop 0175, Reno, NV 89557, USA

L. Reidy

Department of Geography, University of California,

Berkeley, 507 McCone Hall #4740, Berkeley, CA 94720,

USA

L. Saito

Department of Natural Resources and Environmental

Science, University of Nevada Reno, Mail Stop 186, 1664

N. Virginia Street, Reno, NV 89557, USA

B. Nishonov

Laboratory of Surface Water Quality Investigation,

Hydrometeorological Research Institute, 72 K.

Makhsumov Str., Tashkent, Uzbekistan 700052

Present Address:

A. Crootof

School of Geography and Development, University of Arizona,

PO Box 210076, Tucson, AZ 857017, USA

Present Address:

J. A. Scott

U.S. Forest Service, Rocky Mountain Research Station, 2150

Centre Ave, Bldg. A, Suite 368, Fort Collins, CO 80526, USA

Present Address:

L. Saito

The Nature Conservancy, One E. First Street, Suite 1007,

Reno, NV 89501, USA

123

J Paleolimnol

DOI 10.1007/s10933-016-9914-2

dominated by Chenopodiaceae and Tamarix, indicating

that the vegetation growing nearby was tolerant to arid

saline conditions. The near surface sediments of the

cores are dominated by Typha/Sparganium, which

indicate freshwater conditions. Increases in pollen of

weeds and crop plants indicate an intensification of

agricultural activities since the 1950s in the watersheds

of the lakes analyzed. Pesticide profiles of DDT

(dichlorodiphenyltrichloroethane) and its degradates

and c-HCH (gamma-hexachlorocyclohexane), which

were used during the Soviet era, show peak concentra-

tions in the top 10 cm of some of the cores, where

estimated ages of the sediments (1950–1990) are

associated with peak pesticide use during the Soviet

era. These data indicate that the lakes are relatively

young (mostly \150 years old) and that without

irrigation and canal inputs from the Amu Darya, the

lakes would not exist as freshwater lakes.

Keywords Lake formation � DDT � HCH �Limnogeology � Central Asia � Sediment cores

Introduction

In arid regions of the world, perennial freshwater

resources are scarce: high rates of evaporation often

lead to the salinization of fresh water, particularly of

shallow lakes and groundwater. Where these arid

landscapes support agriculture, it is important to

understand the sustainability of freshwater resources.

One such arid landscape where water resource

sustainability is a growing concern is the Khorezm

Province of Uzbekistan. Khorezm is in the Aral Sea

Basin and lies along the Amu Darya (‘‘darya’’ means

‘‘river’’ in Persian) in southwestern Uzbekistan

(Fig. 1). The Amu Darya has a watershed of about

1,385,045 km2 and flows from glacial melt water in

the Pamir Mountains and Tian Shan historically to

the Aral Sea, but no longer reaches the Aral Sea due

to withdrawals for irrigation. Khorezm covers an area

of 6050 km2 and is home to approximately 1.7 mil-

lion residents (The Governmental Portal of the

Republic of Uzbekistan 2016). With annual precip-

itation averaging approximately 92 mm, Khorezm

predominately has an arid continental climate

(Letolle and Chesterikoff 1999; Wehrheim et al.

2008).

Despite the arid climate, Khorezm is one of the

most agriculturally productive regions in Uzbekistan.

An extensive network of canals (Fig. 2) supplies the

Province with irrigation water throughout most of the

year (Wehrheim et al. 2008; Veldwisch 2008). Since

Uzbekistan gained independence in 1991, the country

has remained dependent on the cash flow from their

agricultural exports, which make up approximately

37 % of the country’s gross domestic product (Martius

et al. 2009). With over 16,000 km of irrigation canals

in Khorezm, 3.5–4.0 km3 of water is diverted from the

Amu Darya during the growing season and

1.0–1.5 km3 during the winter season to produce cash

crops such as cotton, wheat and rice. Furrow irrigation

is used for cotton, while flood irrigation is used for

wheat and rice. Irrigation water that is not lost to

evapotranspiration or groundwater is channeled into

9000 km of drainage canals throughout the Province

(Oberkircher et al. 2011).

Over 400 lakes have been identified in the Province

of Khorezm (Kaiser 2005). These lakes are located in

natural depressions in the landscape and some, but not

all, have irrigation canals that supply water to the lakes

in the summer. The north–south linear orientation of

the lakes in the south Khorezm is due to their location

in the depressions between dunes in the regional

desert. Local residents believe the lakes have existed

for thousands of years and the lakes currently have

both cultural significance and prestige in the commu-

nity (Oberkircher et al. 2011). Therefore, changes to

agricultural practices that could affect the water

balance of these lakes must take into account the use

of these lakes and their sustainability. Water balance

studies on two lakes in Khorezm showed that without

the addition of surface water from irrigation canals,

the lakes were likely to desiccate after irrigation ends

(Scott et al. 2011). Although irrigation with Amu

Darya water has existed in the region for more than

2000 years (Tolstov 2005; Boroffka 2010), intensive

irrigation likely began in the early to mid-1900s, so it

is possible that the lakes are less than 100 years old.

Twelve lakes (‘‘kul’’ in the Uzbek language) were

chosen for this study (Fig. 1): Ata (ATA), Dovud

(DOV), Eshan Rabat (ESH), Gauk (GAU), Khod-

jababa (KHO), Koshkopyr (KOS), Nayman (NAY),

Sapcha (SAP), Shur (Koshkopyr) (SHK), Shur (Yan-

giaryk) (SHY), Tuyrek (TUY) and Xitoy (XIT). These

study lakes are well distributed across the Province

and are characteristic of the small (0.02–1.6 km2

J Paleolimnol

123

area), shallow (\3 m deep) lakes found in Khorezm.

Table 1 contains details about the size and land use

around each lake.

In addition to water availability, water quality is

also a concern in Khorezm because it can deviate from

acceptable standards by as much as 80 % (Severskiy

et al. 2005; Fayzieva et al. 2004). Pesticide pollution

was presumed to be prevalent in the region because

crops were historically sprayed with organochlorine

pesticides such as dichlorodiphenyltrichloroethane

8

7

53

4

12

1

9 11

2

6

10

Aral Sea

Caspian Sea

Study Lakes

LakesPolitical Boundaries

0 10 20

Kilometers

Uzbekistan

Kazakhstan

Turkmenistan

Amu Darya

West Central East

1 XIT2 KOS3 SHK4 KHO5 DOV6 SAP7 GAU8 ESH9 TUY10 SHY11 ATA12 NAY

Study Lake Abbreviations

Study Area

N

Fig. 1 Map of the Khorezm region of Uzbekistan showing

lakes in the region (grey areas) and the locations of the lakes

studied (black areas). Lakes are arbitrarily divided into West,

Central, East locations for plotting purposes and to illustrate

how geographic location affects analysis of these lakes. Lake

names corresponding to the abbreviations in the figure are: Ata

(ATA), Dovud (DOV), Eshan Rabat (ESH), Gauk (GAU),

Khodjababa (KHO), Koshkopyr (KOS), Nayman (NAY),

Sapcha (SAP), Shur (Koshkopyr) (SHK), Shur (Yangiaryk)

(SHY), Tuyrek (TUY) and Xitoy (XIT)

J Paleolimnol

123

(DDT) and gamma-hexachlorocyclohexane (c-HCH,better known as Lindane) (Galiulin and Bashkin

1996). While DDT and HCH will degrade by micro-

bial and abiotic decomposition, these pesticides are

known to be persistent and can still be found in water

bodies, soils and the atmosphere worldwide long after

application (Smol 2008; Rosen and Van Metre 2010;

Korosi et al. 2015). In soils, DDT has a recorded half-

life of 2–20 years depending on local temperature, soil

type, pH, redox potential, moisture and organic carbon

content. HCH, which is much less persistent, has an

average half-life in topsoil of 2 months (Manz et al.

2001). Because DDT is highly insoluble, it is not

usually present in high concentrations in the water

column, but will accumulate in bed sediments (Zweig

et al. 1999; Hill and McCarty 1967). The hydrolysis of

DDT in water is pH and temperature dependent.Wolfe

et al. (1977) found DDT to have a half-life of 8 years

with a pH of 7 and a temperature of 27 �C. Degrada-tion of both DDT and HCH occurs primarily by

biological processes; Sharom et al. (1980) found DDT

and HCH to be degraded much more quickly in water

when microbes were present.

Several studies have investigated the presence of

DDT and HCH in humans, water, and surface

sediment samples in the lower Amu Darya region.

Beginning in 1987, the Central Asian Science

Research Institute of Irrigation conducted a study to

assess the ecological and toxicological situation in

lakes and reservoirs in the lower Amu Darya region.

Jensen et al. (1997) found high concentrations of

metals and pesticides in the blood of children living in

Lakes

Canals

Political Boundaries

0 10 20Kilometers

Amu Darya

N

Fig. 2 Map of the Khorezm region of Uzbekistan showing the

densely constructed irrigation and drainage canals in the region

(lakes are in grey areas). The map is the same area as shown in

Fig. 1. The lakes studied are not plotted on this map because for

the most part they are not visible through the irrigation canals.

Location of irrigation canals are from the Zentrum fur

Entwicklungforschung (ZEF)

J Paleolimnol

123

the Aral Sea Basin of Kazakhstan that had signs of

disease. The UNESCO Aral Sea Project (1992–1996)

and INTAS 1039 Project (2002–2003) continued this

research and found that between 1989 and 2002 there

was a sharp decrease in RDDT and RHCH concentra-

tions in water and bottom sediment samples from the

Amu Darya delta (Karimov et al. 2005). These studies

provide information about water quality in the Amu

Darya delta region, but do little to quantify pesticide

concentrations in Khorezm even though this region

was also subjected to large quantities of agrochemicals

when it was part of the Union of Soviet Socialist

Republics (USSR or Soviet Union). The rapid

decrease in pesticide concentrations in Amu Darya

delta sediments (Karimov et al. 2005) may have also

occurred in the lake sediments in Khorezm. Quanti-

fying pesticide pollution in the lakes, particularly for

RDDT and RHCH, provides valuable information

about the biological functions and human health

concerns associated with the lakes. Throughout this

paper RDDT refers to the sum of DDT and its major

metabolites dichlorodiphenyldichloroethylene (DDE)

and dichlorodiphenyldichloroethane (DDD). Simi-

larly, RHCH is defined as the sum of c-HCH and a-HCH isomers.

Although irrigated agriculture has existed in the

Khorezm region for at least 2000 years (Tolstov

2005), intensive irrigation in Khorezm began in the

early to mid-1900s when the region was part of the

Soviet Union. Understanding the origin of these lakes

is important to provide guidance for the development

of water management strategies in this water-scarce

region, as changes to land and water management

could change the hydrology and water balance of the

lakes (Scott et al. 2011), potentially leading to

desiccation and introduction of potentially toxic lake

bed material to the air. This study is the first, to our

knowledge, to assess the origin of these lakes, and the

historical role they play in the hydrology of the region.

We hypothesized that the lakes are young (\100 years

old) based on the amount of irrigation that is used in

the region, and that non-polar pesticides such as

RDDT and RHCH should be present in sediment cores

at concentrations reflecting the Soviet use of these

pesticides in the 1950s to 1980s. The objectives of this

study were to use sediment cores to determine the age

of the lakes using 210Pb and 137Cs dating and to

analyze the lake sediment cores for RDDT and RHCHat small enough intervals to determine the timing of

peak pesticide application.

Materials and methods

Twelve lakes (Fig. 1) in Khorezm, Uzbekistan were

cored from 2008 to 2010 using a home-made coring

Table 1 General information about the lakes studied, including percent agricultural land around each lake determined from satellite

imagery from 2002 to 2006 supplied by C. Conrad

Lake name and

abbreviation used

in text

Area of fields

within 0.5 km

of lake (m2)

Lake area

(m2)

Buffer

size (m2)

Cotton

(%)

Rice

(%)

Wheat–

rice (%)

Wheat–

other

crop (%)

Wheat–

Fallow

(%)

Agriculture

(total %)

Eshan Rabat, ESH 4,318,997 1,584,779 5,903,776 6.60 1.89 0.00 0.00 0.00 8.49

Sapcha, SAP 1,437,084 66,708 1,503,792 22.22 0.00 0.00 0.00 0.00 22.22

Koshkupyr, KOS 1,321,057 71,343 1,392,400 24.00 8.00 0.00 4.00 0.00 36.00

Gauk, GAU 2,925,062 1,196,442 4,121,504 25.68 4.05 2.70 6.76 1.35 40.54

Shur (Koshkupyr), SHK 2,520,212 487,372 3,007,584 16.67 0.00 3.70 16.67 3.70 40.74

Shur (Yangiaryk), SHY 1,286,637 50,067 1,336,704 25.00 12.50 0.00 4.17 0.00 41.67

Tuyrek, TUY 1,756,635 192,725 1,949,360 14.29 11.43 2.86 11.43 2.86 42.86

Xitoy, XIT 2,981,545 304,519 3,286,064 32.20 10.17 1.69 10.17 5.08 59.32

Khodjababa, KHO 1,768,577 236,479 2,005,056 36.11 13.89 5.56 5.56 0.00 61.11

Dovud, DOV 2,412,139 539,749 2,951,888 32.08 7.55 7.55 18.87 1.89 67.92

Nayman, NAY 2,177,317 161,915 2,339,232 23.81 40.48 2.38 16.67 7.14 90.48

Ata, ATA 1,631,098 95,478 1,726,576 48.39 25.81 12.90 6.45 0.00 93.55

J Paleolimnol

123

platform and a custom engineered percussion corer.

While this technique may have smeared the sediment–

water interface, we are confident that the down-core

record is representative of the history of each lake due

to consistent trends seen in the sedimentological and

pesticide records discussed below. TUY and KHO

were cored on two occasions, in March and July 2008.

Cores were taken from the edge of the lakes in March

and from the lake centers in July. The lake edge cores

were taken to determine whether the lake centers’ core

profiles were consistent across the entire lake. All

other lakes were cored once near the deepest area of

the lake. Penetration into lakebed sediment was

generally greater than 1 m. Cores were cut to the

sediment–water interface in the field and packed with

sorbent material to prevent movement of the sediment

within the core barrels. Cores were kept cold and

vertical at all times during storage and transport to

Tashkent, Uzbekistan, for analysis. In Tashkent, the

cores were sliced into 2.5 cm sections and visual

identification of clay and sand units were made during

slicing. As described in detail below, all cores were

analyzed for 210Pb and 137Cs at Eawag, ETH,

Switzerland. Loss on ignition (LOI550), and pesticides

were analyzed at the Laboratory of Surface Water

Quality Investigation, Hydrometeorological Research

Institute, Uzbekistan. Cores from ESH, KOS, and XIT

were analyzed in detail for pollen at the Paleocology

Laboratory, University of California, Berkeley, USA,

with supplemental data from the top and bottom of the

cores from DOV, SHK, and ATA used to verify

interpretations from the detailed examination. All data

used in this assessment of the lakes are available at

Rosen et al. (2016).

210Pb and 137Cs age dating of the cores

Individual 2.5 cm thick slices of core sediments from

each core (5.0 cm thick slices were used for the March

2008 TUY core) were analyzed at ETH-Eawag

Aquatic Research labs (Switzerland) from the top

down until no detection or low detection of 210Pb or137Cs was encountered. Data from 210Pb analyses did

not always show consistent exponential decay in some

of the lakes sampled. These results could be due to

(a) physical or biological mixing at the surface;

(b) changes in the 210Pb supply due to changes in the

patterns of sediment focusing (Appleby 2001) or

(c) core slices that were vertically too thick to get

accurate results. However, 210Pb could be used to

check ages calculated using 137Cs in 10 of the 15 cores.

One lake core (XIT) had no 137Cs peak or useable210Pb profile (there wasn’t logarithmic decay of the210Pb data), and theMarch 2008 KHO core had a small137Cs peak near the surface of the lake sediments,

indicating that sediment from the top of XIT and one

of the KHO lake cores was likely eroded. The age of

XIT and its sedimentation rate could not be deter-

mined because of the lack of useable 137Cs peak and210Pb data (see Rosen et al. 2016), but the data are

included to show patterns with depth. However, 210Pb

data for GAU gave reasonable sedimentation rates,

and these were used for calculating the age of the lake.

Based on the difference between 210Pb and 226Ra,

background 210Pb activities are around 20 Bq/kg.

Constant rate of supply (CRS) method was used for

simplicity and comparability between 137Cs and 210Pb.

Where both 137Cs and 210Pb sedimentation rates could

be calculated for the same lake, the average rate was

used to calculate ages. The global 1964 above-ground

nuclear testing 137Cs peak was used as the primary

marker for the 137Cs model, with the time the lake core

was collected as the other marker. In 3 cores (SHY,

NAY, and KHO), a small spike in the 137Cs activity

above the 1964 peak may be the result of the 1986

Chernobyl accident (Appleby 2001) that spread 137Cs

over much of Europe and parts of Asia (see Rosen

et al. 2016). These peaks were used as additional

controls in these cores. Samples were dried, homog-

enized, and analyzed using a gamma counting spec-

trometer using methods presented in Appleby (2001).

Results of 137Cs and 210Pb counting are presented in

Rosen et al. (2010) and Table 2. Dates are plotted on

diagrams as averages of 137Cs and 210Pb calculated

ages where both methods allowed calculation of ages.

Loss on ignition

For loss on ignition (LOI550), sediment samples were

dried at 105 �C for 6 h until constant weight was

obtained to remove water. The weight was recorded

and the sample was then placed in a muffle furnace at

550 �C for 6 h (Heiri et al. 2001). The difference in

weights between the sample dried at 105 �C and the

weight after the sample was placed in the muffle

furnace determined the LOI550 percentage. Samples

were not treated with HCl because this could remove

some organic material (Beasy and Ellison 2013) and

J Paleolimnol

123

Table

2Agedeterminationsforthelakes

studiedusing

137Csand

210Pbconstantrate

ofsupply

methods

Date/locationofcoressamples

137CsAgedating

210PbagedatingwithCRSmodel

Meanof137Cs

and

210Pb

Datecore

was

collected

Lakeabbreviation

(see

textforlake

nam

es)

Locationof

core

inthe

lake

Mud

depth

(cm)

Depth

of

137Cspeak

incore

(cm)

Sed.rate

(cm/year)

Number

of

years

to

bottom

of

lakesedim

ent

Dateof

lakeorigin

(calendar

years)

Sed.rate

(cm/year)

Number

of

years

to

bottom

of

lakesedim

ent

Dateof

lakeorigin

(calendar

years)

Mean

sedim

entation

rate

(cm/year)

Mar-08

KHO1

Edge

37.5

20

0.45

82

1926

??

?

Jul-08

KHO2a

Center

25

Cesium

peakat

topofcore

0.22

114

1894

Jul-08

SHK

Center

50.0

22.5

0.51

98

1910

0.73

70

1938

0.61

Mar-08

TUY1

Edge

30.0

15

0.34

88

1920

0.35

86

1922

0.35

Jul-08

TUY2

Center

30.0

7.5

0.17

176

1832

0.35

86

1922

0.26

Jul-08

TUY3

Center

35.0

7.5

0.17

205

1803

0.27

130

1878

0.22

2009

ATA

Center

27.5

12.5

0.28

99

1909

0.34

80

1929

0.31

2009

GAUb,c

Center

[70

Cesium

peakat

topofcore

0.27

259

1750

0.14

2010

NAY

Center

45.0

32.50

0.71

64

1945

0.64

70

1940

0.67

2010

SAPb

Center

[62.5

12.5

0.27

230

1780

??

?

2010

SHY

Center

20.0

10.00

0.22

92

1918

??

?

2010

XIT

aCenter

47.5

Cesium

peakat

topofcore

??

?

2010

ESH

Center

45.0

12.5

0.27

166

1844

0.36

125

1885

0.32

2010

DOV

Center

50.0

12.50

0.27

184

1826

??

?

2010

KOS

Center

45.0

7.50

0.16

276

1734

0.10

450

1560

0.13

‘‘?’’indicates

inconsistent210Pbactivitiesobtained

downcore,so

nosedim

entationrate

could

begenerated

aMay

beeroded

atthetop

bAgeto

bottom

ofcore–clay

alltheway

tobottom

cMay

beeroded

atthetopofthecore

orthecesium

peakisbelow

70cm

depth

J Paleolimnol

123

all lakes except ESH, which had calcium carbonate

precipitated around macrophyte stems in the near

surface sediments, were undersaturated with respect to

carbonate minerals. LOI550 is used here as a qualita-

tive assessment of organic carbon content of the

sediments because LOI550 percentages can be affected

by lithology (Santisteban et al. 2004). However, trends

down core can be useful and comparable using the

same LOI550 method for lakes from the same geologic

setting with a similar lithology, as is the case for the

sediments from the lakes studied. Replicates were

analyzed on 10–20 % of samples analyzed. Agree-

ment was generally within 10 %. In figures included

here, replicate data are plotted as the average of the

replicates.

Pollen analysis

Standard pollen extraction and preparation procedures

were followed (Faegri and Iversen 1989). One tablet

containing approximately 20,848 Lycopodium spores

was added to each sample for control purposes

(Stockmarr 1971). Samples were sieved with a

125 lm mesh filter to remove large debris and

processed with the following treatments: hydrochloric

acid, potassium hydroxide, hydrofluoric acid, iso-

propanol wash, nitric acid, glacial acetic acid, and

acetolysis. The residues were stained with safranin,

dehydrated with tertiary butyl alcohol, suspended in

silicone oil, and mounted on microscope slides. Pollen

was counted at the 4009magnification with 10009 to

determine fine detail. Pollen grains and fern spores

were identified to the lowest possible taxonomic level

using the UC Berkeley Museum of Paleontology’s

collection of over 10,000 modern pollen samples and

published pollen keys (McAndrews et al. 1973; Kapp

et al. 2000). Unknown grains were noted and drawn.

Damaged, torn or crumpled grains that were uniden-

tifiable were counted as ‘‘Indeterminate’’. A minimum

of 400 grains were counted for at least 9 samples on

cores from KOS, XIT, and ESH. Pollen counts were

compiled and plotted using the CALPALYN program

(Bauer et al. 1991).

Pesticide analyses

The methodology used to analyze pesticide residues in

sediments followed the standard techniques used by

Uzhydromet (RD 52.24.71-88 1988). Core sample

slices were homogenized as wet sediment and then air-

dried at 20 �C. Double extraction of pesticides from

the sediment was done using n-hexane ? acetone.

Sample cleanup was conducted using concentrated

sulfuric acid for elimination of organic substances.

Extracts were dried with Na2SO4, reconcentrated, and

injected into an Agilent 6890 chromatograph with a

capillary column. In addition, the absolute dried bulk

density of the sediment was determined by weighing

homogenized and air-dried samples after they were

heated in a muffle furnace at 105 �C for 6 h. Samples

were checked to ensure that the recorded weights were

constant. Replicates were analyzed on 1–3 samples in

most cores and results are generally within 25 % of

each other. A blank, using bottled water, was run on

one occasion and no RHCH or RDDT was detected,

but because some RHCH and RDDT were detected in

many samples of prelacustrine sands, where detections

would not be expected, low concentrations of RHCHand RDDT (below 5 lg/kg for RDDT and below

2.5 lg/kg for RHCH) were considered as background

and not environmentally relevant. In figures, replicate

data are plotted as the average of the replicates.

Results

Age dating

The lacustrine sediment in all of the lakes except GAU

and SAP is B50 cm. Below 50 cm depth, all cores

penetrated relatively clean sand that formed from

either riverine or eolian processes. Cesium-137 pro-

files for all the lakes show that all of the 1964 above-

ground testing 137Cs peak activities are shallower than

22.5 cm from the sediment–water interface (Table 2),

except for NAY core, which is at 32.5 cm depth. Lakes

SHY, NAY, and KHO (lake center core), show a small137Cs peak between 2.5 and 5.0 cm depth, 17.5 and

20 cm depth, and 7.5 and 10 cm depth, respectively,

which are interpreted to be the spike from the 1986

Chernobyl accident. The small Chernobyl peak is

almost exactly half-way between the surface (2008 or

2010) and the 1964 137Cs peak in all three cores. The

maximum age of the lakes calculated using the 137Cs

sedimentation rates is 276 years before collection of

the cores, and the minimum age is 64 years.

Lead-210 CRS models were constructed for 10 of

the cores, but only eight have 137Cs profiles that can be

J Paleolimnol

123

compared to the 210Pb profiles. All of the 210Pb-

calculated sedimentation rates are similar to those

calculated using 137Cs peaks. Sedimentation rates

range from 0.10 to 0.73 cm/yr and those sedimenta-

tion rates that can be compared by using duplicate

cores from the same lake are B0.22 cm/yr different

(Table 2). Given this potential error in sedimentation

rate, each centimeter of core could be in error by

5 years in each direction, so dates are a minimum of

±10 years per centimeter of core, and more likely

±20 years, if other errors such as mixing are incor-

porated. The maximum age of the lakes calculated

using the 210Pb models is 450 years prior to core

recovery and the minimum age is 70 years. Some

lakes have younger ages with sedimentation rates

calculated from 137Cs peaks compared to ages derived

from 210Pb others have younger ages calculated with210Pb models compared to 137Cs derived ages (see

Table 2).

There appear to be three groups of lake ages for the

onset of lacustrine sedimentation. Older lakes (SAP,

KOS and possibly GAU) appear to have formed before

1800. Other lakes appear to have begun forming

between 1880 and 1920 (ESH, SHK, KHO, and TUY),

and still others appear to have formed between 1930

and 1960 (NAY, DOV, ATA, and SHY). One lake

(XIT) could not be assigned to any groups because age

models could not be constructed for this lake because

there was no real 137Cs peak and 210Pb did not decay

exponentially (see Rosen et al. 2016). It is possible that

sediment at the top of XIT was eroded during past low

lake levels. However, XIT transitioned from sand to

lake muds by 50 cm depth, indicating a time of

formation that is similar to the other lakes that were

dated, assuming erosion did not remove more than

20 cm of sediment from the lake.

Loss on ignition

Loss on ignition data show dramatic increases in the

organic matter component of the sediment once

lacustrine sedimentation begins in all lakes (Fig. 3a–

d). LOI550 values range from 5 to 23 % at the tops of

the cores, and between 3.7 and 0.6 % in the sand at the

bottom of the cores. The GAU core consisted of mud

down to 70 cm depth, however, the LOI550 decreased

from 7 % at the top of the core to 1.7 % at the bottom

of the core, indicating that the transition to sand was

likely not far below this depth. Similarly, the SAP core

also had LOI550 decreasing to 3.7 % (from[17 % at

the top of the core), indicating that the transition to

non-lacustrine sediments was not far below the bottom

of the core. Several cores show a slowdown of

increasing LOI550 around 1970–1990, with a final

increase in LOI550 in the last 20 years or so.

Pollen

Selected major pollen and spore profiles were devel-

oped for three of the lakes (KOS, ESH and XIT) and

are displayed in Fig. 4a–c and used to discuss the

results of the record. Pollen concentration is measured

as grains/cm3 (see Rosen et al. 2010). Preliminary

pollen data from the bottom, middle and top sections

of two additional lakes (ATA, DOV and SHK) are

presented in Rosen et al. (2010).

Koshkopyr

Two pollen zones were defined for the KOS record

(Fig. 4a) based on changes in the abundance of certain

aquatic and riparian pollen types and the first appear-

ance of several agricultural and weedy pollen types.

Changes and reversals in the aquatic and riparian

record reflect local environmental change and natural

plant succession in the lake. The overall pollen record

shows variation through time, especially for the last

one hundred years.

Zone 2: ca. 1840–1900 (22–15 cm) The high

percentages of Chenopodiaceae (Goosefoot family)

([40 %) and Tamarix (salt-cedar) (*20 %) at the

bottom of this zone (Fig. 4a) indicate that plants

adapted to dry saline conditions were locally common.

Low percentages (\1 %) of tree pollen Juglans

(walnut), Pinus (pine) and Populus (aspen) suggests

an open environment with few trees. The presence of

large Poaceae (grass family) pollen with a size range

of 42–51 lm (Chaturvedi et al. 1998) reflects the

cultivation of Oryza sativa (Asian rice) locally. Pollen

concentrations in this zone average 30,000 grains/

cm3.

Zone 1 ca. 1901–2010 (15–0 cm) The pollen record

changes dramatically in zone 1. It is characterized by

sharp decreases in both Chenopodiaceae and Tamarix

and the first appearance of weedy and additional crop

pollen types. Rumex acetosella and Plantago

J Paleolimnol

123

lanceolata both appear at 15 cm. Solanum spp.

(potato) pollen appears at the 12.5 cm level while

Malvaceae (cotton) first appears at the 10 cm interval.

Typha/Sparganium (cattails) steadily increase from

*10 % at the base of this zone to over 40 % in the

near surface sediments. There is a slight increase in

Populus between 12 and 5 cm to 3 % of the pollen

sum but it drops again towards the top of the zone to

\1 %. Pollen concentrations average close to

64,000 grains/cm3 in this upper section.

Eshan Rabat The general ESH pollen record

(Fig. 4b) does not change significantly during the

past 100 years. Weed (Rumex spp., Plantago

lanceolata) and crop pollen (Solanum spp.,

Malvaceae, and Oryza sativa) types are present

throughout the record from the basal sediments

(30 cm) to the top of the core. There is a sharp

increase in Poceae pollen in the upper 2.5 cm of the

record. Chenopodiaceae is the dominant pollen type

throughout the core and averages between *30 and

40 % between 30 and 0 cm. Pollen concentrations are

variable and range between 10,000 and 70,000 grains/

cm3.

Xitoykul The XIT core is undated but there are some

similarities with the KOS record (Fig. 4c).

Chenopodiaceae, Tamarix, and Poaceae are the

dominant pollen types in the lower part of the record

but all decrease towards the top. Typha/Sparganium is

low (7 %) in the basal sediments but increases to

almost 60 % of the pollen sum above the 7.5 cm level.

The pollen of Myriophyllum (water mil-foil) averages

8 % in the basal sediments and then declines all the

way to the surface. Myriophyllum is an emergent

aquatic that does well in shallow water environments.

Weedy pollen (Rumex acetosella; Plantago

lanceolata) first appear at or above 20 cm. Monolete

A B

C D

Fig. 3 Loss on ignition (LOI550) percentages for all lakes

studied plotted by calendar years, except XIT, which is plotted

by depth. a Lakes located in the western, b central, and c easternpart of Khorezm. d The 3 cores taken in TUY from the edge and

center part of the lake. TUY is located in the eastern part of

Khorezm. Dashed lines show where sand is located in the

bottom of the cores. Lacustrine sedimentation begins where the

line becomes solid. Due to uncertainties in the age models,

errors in age dates may be as high as ± 20 years (see text for

explanation)

J Paleolimnol

123

Koshkopyr

CHRONOLOGY

Years

A.D.

2000

1981

1961

1942

1923

1904

1885

1865

1846

Dep

th in

cm

.

0

2

5

8

10

12

15

18

20

22

Popul

us

0 2 % Total Pollen and Spores

Jugl

ans

0 1

Pinus

0 1

Salix

0 1

Tamar

ix

0 5 10 15 20

Typha

/

Spa

rgan

ium

0 15 30 45

CYPERACEAE

0 3

Myr

ioph

yllum

0 4

CHENOPODIACEAE

0 10 20 30 40

POACEAE

0 10

ASTERACEAE

0 1

Mon

olete

Spores

0 1

Orzya

sativ

a

0 2 4

MALVACEAE

0 1

Plant

ago

lanc

eola

ta

0 1

Sola

num

spp.

0 1

Rumex

aceto

sella

0 1

Pollen

Con

centr

ation

0 50,000

TREES RIPARIAN & AQUATICS HERBS & SHRUBS

3

AGRICULTURAL

Zone 1

Zone 2

grain

s/cm

Eshan Rabat

CHRONOLOGY

Years

A.D.

2006

1998

1990

1982

1974

1967

1959

1951

1944

1936

1928

1920

Dep

th in

cm

.

0

2

5

8

10

12

15

18

20

22

25

28

30

Popul

us

0 2 % Total Pollen and Spores

Jugl

ans

0 2

Pinus

0 3

Salix

0 2 4

Tamar

ix

0 5 10 15 20 25 30

Typha

/

Sp

arga

nium

0 5 10 15 20

CYPERACEAE

0 2 4

Myr

ioph

yllum

0 4

CHENOPODIACEAE

0 10 20 30 40

POACEAE

0 10 20

ASTERACEAE

0 1

Mon

olete

Spores

0 2 4

Orzya

sativ

a

0 1 2

MALVACEAE

0 1

Plant

ago l

ance

olat

a

0 2

Sola

num sp

p.

0 1 2

Rumex

aceto

sella

0 1 0 50,000

TREES HERBS & SHRUBSRIPARIAN & AQUATICS AGRICULTURAL

Pollen

Con

centr

ation

grain

s/cm

3

Xitoykul

Dep

th in

cm

.

0

2

5

8

10

12

15

18

20

22

25

28

30

32

35

Popul

us

0 2 % Total Pollen and Spores

Jugl

ans

0 1

Pinus

0 1

Salix

0 1

Tamar

ix

0 5 10 15

Typha

/Spa

rgan

ium

0 15 30 45 60

CYPERACEAE

0 3

Myr

ioph

yllum

0 4 8

CHENOPODIACEAE

0 10 20 30

POACEAE

0 10 20

ASTERACEAE

0 1

Mon

olete

Spores

0 1

Orzya

sativ

a

0 2

MALVACEAE

0 1

Plant

ago

lanc

eola

ta

0 1

Sola

num

spp.

0 1

Rumex

aceto

sella

0 1

Pollen

Con

centr

ation

grain

s/cm

0 35,000 70,000

TREES HERBS & SHRUBSRIPARIAN & AQUATICS

3

AGRICULTURAL

A

B

C

Fig. 4 Selected pollen and spore taxa for aKOS, b ESH, and cXIT. Due to uncertainties in the age models, errors in age dates may be

as high as ± 20 years (see text for explanation)

J Paleolimnol

123

spores (ferns) are found throughout the length of the

record along with the pollen of Orzya sativa.

Malvaceae only appears at the 25 cm level, while

Solanum appears above the 20 cm level. Pollen

concentrations average 20,000 grains/cm3 below

30 cm while concentrations increase towards the top

of the record to an average of about 60,000 grains/

cm3.

Pesticides

Pesticide concentrations in all cores are relatively low

compared to lakes worldwide (for example see Elpiner

1995; Rosen and Van Metre 2010), \10 lg/kg for

bothRDDT andRHCH, except for a few samples from

TUY, ESH, and DOV (Figs. 5a–d, 6a–d). Concentra-

tions were generally highest in the upper part of the

cores, generally from 1960 to 1990 for RDDT. Spikesof higher than background concentrations are partic-

ularly noticeable in ESH, TUY, and SHK for RDDT,and in DOV, TUY, SHY and SHK for RHCH. In 5 of

the lakes, the highest RHCH concentrations are at the

top of the cores (DOV, SAP, KOS, SHY, and SHK).

There is no apparent correlation between pesticide

concentrations in the lakes and percentage of recent

agricultural lands around the lakes (Table 1). ESH,

which has the lowest agricultural land use around the

lake, has the highestRDDT concentration of any of the

lakes (Table 1). However, the only surface water

inflow to the lake is a drainage canal, which contains

the runoff and groundwater from upstream agricultural

fields.

Discussion

Age dating

The age models constructed from 137Cs and 210Pb

dating are relatively consistent, but are somewhat

simplistic (CRS model only). This is because the

decay curves of unsupported 210Pb are not always

exponential and so may have some error associated

with the dates. In some cases, 137Cs peaks are not sharp

and so may not provide an exact date for the 1964

global atmospheric testing peak. However, even with

these uncertainties, it is clear that the lakes have not

existed for thousands of years because the thickness of

lacustrine sediment in the lakes is\60 cm in all but

two of the lakes, and extrapolation of even these

poorly constrained ages does not make the lake

sediments older than a few hundred years. In addition,

the LOI550 data for both SAP and GAU are\5 % and

decline near the base of the cores, possibly indicating

that pre-lacustrine conditions are not far from the base

of these cores.

The reason for the variable ages of the lakes is

difficult to determine as there is no clear pattern. The

variability may reflect either poor age control in some

of the lakes caused by variable smearing of 210Pb or137Cs peaks, or other mechanical or chemical diffusion

or mixing within the lake. Klaminder et al. (2012)

found that cesium deposited during the Chernobyl

accident migrated down core, obscuring the 1964

global peak in a lake located near Chernobyl that

received high amounts of fallout from the accident.

While the lakes in Uzbekistan were not in the path of

high fallout from Chernobyl (Pollanen et al. 1997),

some of the lakes may show possible Chernobyl peaks

(e.g., NAY and SAP). This may indicate some

downward migration of cesium in these lakes. In

some lakes, the first deposited muds may represent

natural depressions that may have been playas or

ephemeral saline lakes. Salt was visually identified in

KOS sediments, which is an indication of a saline past

for this lake, and the calculated age for the start of

lacustrine sedimentation is older in KOS than the other

lakes. Other possible causes of the variability is

mixing of sediment due to human and/or animal

disturbance, or eroded sediment being washed into the

lakes in the suspended load via irrigation water and

being deposited on or mixed with younger aged

sediments. However, except for a few cases (i.e., TUY

cores), calculated 137Cs and 210Pb ages are consistent

(within 20–40 years) Alternatively, irrigation may

have a much longer history in the region. There is

limited knowledge of pre-Soviet irrigation canals that

could go back more than 2000 years (Tolstov 2005;

Boroffka 2010). However, the distribution of lakes

formed prior to 1900 appears random and does not

suggest any regional trend. In contrast, three of the

younger lakes, which formed post-1900 (NAY, ATA,

and SHY) cluster in the eastern area of Khorezm

(Fig. 1). Historical data on where and how much

irrigation was used in the region was not collected in

most parts of Khorezm because water meters were not

in place, and in some areas still are not used. Without

information on the historical use of irrigation in

J Paleolimnol

123

Khorezm, it is impossible to determine precisely the

date of origin of the lakes. However, as discussed

below, the 12 study lakes appear to have started out as

natural depressions, possibly with saline water con-

tributing to the depressions before irrigation began.

Two lakes (TUY and KHO) had multiple cores

dated and analyzed (Table 2). For lake TUY, one was

taken on the edge of the lake (TUY-1) and 2 cores were

taken in the center of the lake (TUY-2 and TUY-3).

The core from the edge of the lake has a younger age

calculated for the bottom of the core using a 137Cs

peak at 7.5 cm depth for both lake center cores.

However, TUY-2 core has a 210Pb age that is similar to

the age of the lake edge core using the 137Cs peak, and

TUY-3 has a 210Pb age that is about 40 years older. It

may be that the ages indicated by the 137Cs data of both

lake center cores are too old and that the lake more

likely began forming around 1920 as indicated by the210Pb, rather than at the earlier dates calculated from

the 137Cs of the lake center cores.

The 2 cores from KHO show different issues as

compared to the TUY cores. For the KHO cores, the

core taken near the center of the lake has no 137Cs

peak, indicating either erosion or dredging of the lake

because the lake muds are thin in this core (25 cm)

compared to the lake edge core that is 37.5 cm thick.

Although an age of the lake for the basin center core

can be determined from the 210Pb data, the age is likely

overestimated due to missing sediment and possible

mixing, although this age is used for plotting purposes.

The ages calculated for both cores differ by 35 years.,

Due to the missing top of the lake center core, the 1926

age for the lake edge core is more consistent with the

age of lacustrine deposition of muds in the majority of

A B

C D

Fig. 5 Concentrations of RHCH for all lakes studied plotted by

calendar years, except XIT, which is plotted by depth. a Lakes

located in the western, b central, and c eastern part of Khorezm.

d Concentrations of RHCH for the 3 cores taken in TUY from

the edge and center part of the lake. TUY is located in the eastern

part of Khorezm.Dashed lines show where sand is located in the

bottom of the cores. Lacustrine sedimentation begins where the

line becomes solid. Dashed vertical line at 2 lg/kg is the

concentration above which the concentrations are above

analytical background concentrations and may be environmen-

tally relevant. Due to uncertainties in the age models, errors in

age dates may be as high as ± 20 years (see text for

explanation)

J Paleolimnol

123

other lakes studied and with the timing of increased

irrigation during the period of the Soviet Union.

Lacustrine sediments and LOI550

The short length (\60 cm) of the lacustrine sediments

in most of the cores except for GAU and SAP suggests

a short depositional history for the lakes. Sharp

increases in LOI550 occur in all lakes above basal

quartz sands, and the percentage of organic matter

present in the muds increase to the surface, except in

GAU (Fig. 3a–d). In most lakes, LOI550 percentages

are highest since 1950 when most irrigation was

occurring. LOI550 percentages in all lakes are below

10 % before 1950 (except for a few samples just above

10 % in GAU), and after 1950many lakes show steady

increases in LOI550, with some lakes reaching 20 %

within 30 years. The steady increase in LOI550 in the

last 30–50 years indicates that the lakes are becoming

more productive and that perennial relatively fresh

water is present in the lakes. The increasing produc-

tivity means that more water and nutrients are

available to the lakes. Stratigraphic increases in

organic carbon content (related to higher lake produc-

tivity) coincided with the peak of Soviet expansion of

irrigation between 1951 and 1991, and it is likely that

this expansion of irrigation and fertilizer use has

allowed for the increased productivity in these lakes.

The three cores from TUY show similar patterns to

the other cores, with much higher LOI550 percentages

at the top of the cores. However, increases in LOI550are not steady in these cores (Fig. 3d) and there is a

period between 1930 and 1970 where LOI550 does not

increase, particularly in the cores taken near the center

A B

C D

Fig. 6 Concentrations of RDDT for all lakes studied plotted by

calendar years, except XIT, which is plotted by depth. a Lakes

located in the western, b central, and c eastern part of Khorezm.

d Concentrations of RDDT for the 3 cores taken in TUY from

the edge and center part of the lake. TUY is located in the eastern

part of Khorezm.Dashed lines show where sand is located in the

bottom of the cores. Lacustrine sedimentation begins where the

line becomes solid. Dashed vertical line at 5 lg/kg is the

concentration above which the concentrations are above

analytical background concentrations and may be environmen-

tally relevant. Due to uncertainties in the age models, errors in

age dates may be as great as ± 20 years (see text for

explanation)

J Paleolimnol

123

of the lake. LOI550 then increases sharply in the last

30 years or so. The cause for the constant LOI550period in TUY is not known, but some of the other

lakes (e.g., SAP, NAY, and ESH) also have this period

of constant LOI550 at about the same time. It may be

related to water availability or longer-term climate

changes, but given how heavily water is manipulated

in this region, it is more likely caused by changes in

water availability to these lakes.

Pollen

For the purposes of the discussion we examine the

KOS and the undated XIT record together as both have

similar recent paleoecological histories. The pollen

content of the basal sediments in both KOS and XIT

records the composition of the vegetation growing in

the area prior to the establishment of the lakes, and

initial plant succession once the water bodies became

permanent. Low percentages of tree pollen (Pinus,

Juglans, Populus, and Salix) in both records suggest

trees were not locally common at the time of lake

formation. The abundance of Chenopodiaceae and

Tamarix in the lower sections of both records suggest

that these lakes initially had arid environments as both

are characteristic of saline soils. The Chenopodiaceae

is relatively floristic rich (4.2 %) in the arid and semi-

arid zones across Uzbekistan while Tamarix, a salt

tolerant species, is common in high saline water

table sites (Gintzburger 2003). The desert soils of

Uzbekistan have evolved under semi-arid and arid

conditions and have been intensively studied for their

agricultural development due to the availability of

irrigation water (Letolle and Mainguet 1993).

Following freshwater inputs from irrigation there is

a dramatic increase in Typha/Sparganium and corre-

sponding decreases in Chenopodiaceae and Tamarix.

This change in vegetation provides evidence for the

transition to freshwater environments reflected by

natural plant succession in and around the lake.

Myriophyllum is present at both sites but at different

abundances in the early part of both lake histories. The

low levels of Myriophyllum at KOS suggest that the

lake filled in quickly when irrigation water became

available. At XIT Myriophyllum persists for longer in

the basal sediments suggesting the lake existed as a

shallow pond initially before the lake became deeper

and more permanent. In general, the basal sediments

contain lower pollen concentrations possibly due to

the high sand content in these sections of the cores.

While rice pollen is encountered at all levels, the

first appearance of weedy pollen and other plant crops

(cotton and potatoes) occur near the top of the cores

and become more abundant, likely reflecting intensi-

fication of agricultural activities after freshwater

inputs from irrigation to the area. The transition to

pollen types dominated by freshwater plants in the

sediment is concomitant with the greater carbon

content of the upper sediments, and higher pollen

concentrations overall at both sites. The combination

of these factors demonstrates that the lakes are

becoming fresher and more productive over time.

The lack of pollen variation in ESH may be

explained by its position on the border of the irrigated

area in Khorezm (Fig. 1). There is little irrigated land

around the lake, and the lake receives only irrigation

drainage water and has no outlet (Crootof et al. 2015).

The lake water is currently hyposaline to hypersaline

(8 g/L up to 50 g/L) depending on the season and

water supply to the lake. During drought conditions,

evapotranspiration is high and salinity increases

substantially compared to the other lakes studied here

(Crootof et al. 2015). Increased aridity is detrimental

to the preservation of pollen, which may be another

factor in the lack of pollen variation observed in ESH.

Thus, variations in pollen types are more variable and

less definitive than the other lakes.

Preliminary data from DOV and ATA both show

that they have similar paleoecological histories to

KOS and XIT, as they are initially dominated by plants

adapted to arid environments and eventually replaced

by plants adapted to more freshwater conditions. SHK

is similar to the ESH site as there is little variation in

the down-core pollen record.

Pesticide occurrence

The pesticidesDDT andHCHwere used extensively in

Uzbekistan in general and in Khorezm specifically

from the 1950s through 1991 (when Uzbekistan gained

its independence). In the late 1970s, the total amount of

pesticides used in Central Asia generally was between

30 and 35 kg/ha, almost 30 times higher than in the rest

of the former Soviet Union during this time (UNESCO

2000). Therefore, high concentrations of these pesti-

cides might be expected in soil and lake sediments.

J Paleolimnol

123

The low concentrations of pesticides found in the

lakes studied is surprising given the history of

pesticide use in the Aral Sea Basin during the Soviet

era (see for example Galiulin and Bashkin 1996;

Tornqvist et al. 2010). However, pesticides in

Khorezm may not travel far. The region is arid with

little rainfall and runoff into the lakes. The landscape

surrounding the lakes has low relief so what runoff

does occur does not likely mobilize the agricultural

soils that are far from the lakes. Furthermore, the water

that fills the lakes is predominately Amu Darya water

(Shanafield et al. 2010; Scott et al. 2011). This water is

stored in a large upstream reservoir above the

agricultural fields and is likely low in pesticides.

Low concentrations (maximum of 0.14 lg/L for

RDDT and maximum of 0.05 lg/L for RHCH) weredetected in water samples collected during 2006–2008

(Crootof et al. 2015), indicating that the recent

pesticide concentrations in the lake sediments are

likely to be low compared to lakes worldwide. While

pesticide concentrations in all lakes are low at the lake

sediment surface (maximum concentration of 20 lg/kg for both RDDT and RHCH), we observe increasingconcentrations of RHCH at the surface of DOV, KOS,

SAP, SHK, and SHY (Fig. 5a–c). The reason for this is

not known, but pesticide storage areas that contain

soils with high concentrations of pesticides still exist

in the region and could be mobilized (Crootof 2011). It

is also possible that the source can be agricultural

lands with high concentrations of these persistent

pesticides in some areas of Khorezm.

The highest concentrations ofRDDT are not seen at

the surface of the lake sediments. Only four lakes

(SHK, ESH, NAY, and TUY) have RDDT concentra-

tions in the lake sediments greater than the background

value of 5 lg/kg (Fig. 5a–d). The three cores from

TUY all have the RDDT peak at approximately the

same time period, between about 1950 and 1970

(Fig. 5d). The two cores from the center of the lake

both peak around 1970. The older age of the RDDTpeak for the core at the edge of the lake, may be due to

uncertainty in the age model for this core, or because

pesticide applications reached the edge of the lake

before it made it to the center of the lake. DDT use in

Uzbekistan was prohibited in 1983, so the peak in

concentrations in RDDT concentrations in TUY

during the 1970s was likely due to its increased use

before the ban. The low concentrations of RDDT in

the shallowest core depths in TUY and all the cores is

likely due to its non-use after 1983, or to the possibility

that other less-expensive pesticides have been used

instead since 1983.

Concentrations of both RDDT and RHCH are

greater than background concentrations in pre-lacus-

trine sands in the core taken from the center of TUY

(TUY-2), and concentrations of RDDT are higher in

core TUY-1 taken on the edge of TUY (Figs. 5d, 6d).

It is not clear why concentrations in the sand would be

as high, or higher than, the lacustrine sediments, but

movement through anoxic groundwater is possible

over a long period of time. Scott et al. (2011) showed

that although groundwater contributions to this lake

are small, there is some movement of groundwater to

the lake.

Only SHK shows peaks in concentrations of both

RDDT and RHCH (Figs. 5a, 6a). The highest con-

centrations are between 1950 and 1990, which is

during peak use of these pesticides (Li et al. 2004;

Elpiner 1995). Although the concentrations of RDDTare low overall, the peak concentration of 45 lg/kg is

similar to other areas of the world (Rosen and Van

Metre 2010). In addition, 20 years of degradation

since peak pesticide use in 1990 may be a significant

way to reduce pesticide contamination because these

lakes are relatively carbon-rich, which may cause

higher than usual degradation rates.

Origin of the lakes

The need for large amounts of Amu Darya river water

in an arid environment led to salinization of the soil

and the need for flushing salts from the soil every

spring (Ibrakhimov et al. 2011). The construction of

irrigation and drainage canals to distribute Amu Darya

river water likely has led to the formation or

enhancement of most of the lakes in the region, which

supports our results that indicate most of the lakes are

less than 100 years before present. Evidence for this

comes from the short length of the lacustrine sedi-

ments and age models for the lakes studied. In

addition, comparing water level measurements from

the 1950s (before the most intensive irrigation began)

to the mid-2000s from six wells in Khorezm shows

that groundwater levels have risen by more than 10 m

(Scott et al. 2011). The water table is now near the

surface (\2 m) (Ibrakhimov et al. 2011). Given that

the groundwater table was more than 10 m lower in

the 1950s than it is now, it is unlikely that many natural

J Paleolimnol

123

lakes would have been able to exist in the arid climate

without a connection to the groundwater or a perma-

nent surface water source. There are no natural surface

water connections to the lakes, only constructed

irrigation and drainage canals, so natural lakes would

have had to rely on groundwater to exist year round.

Scott et al. (2011) showed that groundwater discharge

to TUY and KHO is not sufficient to keep the lakes

perennial, and this is likely the case for most of the

other lakes due to the very flat land surface and lack of

vertical head to supply sufficient groundwater to the

lakes to outstrip evaporation. Although the age of

some of the lakes may predate the large-scale irriga-

tion development by the Soviet Union that led to

declining Aral Sea water levels beginning in 1960

(Micklin 1988), these lakes may have still formed

from human-induced water diversions, just at an

earlier date than is hypothesized here. The Amu Darya

flooded seasonally before dams regulated its flows,

which may have also provided seasonal water to these

lakes in earlier times (Tanton and Heaven 1999).

Conclusions

Our analysis of the first sediment cores taken from

lakes in Khorezm indicates thin accumulation of

lacustrine sediments, and relatively consistent 137Cs

and 210Pb age dating, which indicate that lakes are

generally less than 150 years old and some are less

than 100 years old. A few lakes may be older, but

limitations in the age dating techniques makes

assigning older constrained ages uncertain.

Loss on ignition and pollen data indicate that when

perennial lakes began accumulating sediment, they

were low in carbon and sediments were composed

mostly of sand. As more water became available

through irrigation, lakes became fresher, deposited

fine-grained sediments, and became permanent. The

highest organic content of the cores is always at or near

the top of the cores, within sediment that was

deposited since 1950 when water use was greatest.

Based on detailed pollen data from at least two sites

(KOS and XIT) the lakes most likely started out as

ephemeral saline lakes that became fresh as Amu

Darya irrigation water was transported to the lakes

through irrigation canals, groundwater, and occasional

runoff. Analysis of RHCH and RDDT data indicate

that while some lakes show peaks in either RHCH or

RDDT concentrations between 1960 and 1991 when

the region was part of the Soviet Union, remaining

concentrations of these pesticides are low compared to

published literature, and some lakes continue to have

increasing concentrations of RHCH. It is uncertain

why RHCH should still be increasing in some of the

lakes, as its use has been greatly curtailed since 1991.

Although some small natural lakes may exist in

Khorezm, they are not the dominant lake type in this

region. The young ages of the lakes indicate that they

were formed mostly from imported irrigation water

from the Amu Darya. As has been shown by water

budget calculations (Scott et al. 2011), without annual

input of irrigation water, it is unlikely that the lakes

will remain perennial or fresh. Therefore, changes in

irrigation practices and water management in the

region will play an important role in the future role of

these lakes in the Khorezm landscape.

Acknowledgments The authors would like to thank Benjamin

Trustman and Scott Mensing for preliminary pollen separations,

and Flavio Anselmetti for his help in getting cesium and lead

dating done at ETH, Switzerland. Marie Champagne helped to

plot the pollen diagrams. Thisworkwould not have been possible

without support and contributions from John Lamers and

Zentrum fur Entwicklungforschung (ZEF), Dilorom Fayzieva,

Marhabo Bekchonova, Diana Shermetova, Nodirbek

Mullaboev, Margaret Shanafield, and Hayot Ibrakhimov. The

project was funded by a NATO Science for Peace Grant No.

982159 and a grant from the National Science Foundation EAR

0838239. The US Geological Survey has approved this

manuscript for publication. All interpretations reflect the

scientific judgment of the authors and do not necessarily reflect

the views of their organizations. The use of product names if for

identification and does not imply endorsement of the US

Government or any of the authors respective organizations.

References

Appleby PG (2001) Chronostratigraphic techniques in recent

sediments. In: Last WM, Smol JP (eds) Tracking envi-

ronmental change using lake sediments, vol 1., Basin

Analysis, coring, and chronological techniquesKluwer

Academic Publishers, Dordrecht, pp 171–203

Bauer RE, Edlund E, Orvis K (1991) CALPALYN 2.0. U.C.

Berkeley. Department of Geography Palynology Labora-

tory, Berkeley

Beasy KM, Ellison JC (2013) Comparison of three methods for

the quantification of sediment organic carbon in salt mar-

shes of the Rubicon Estuary, Tasmania, Australia. Int J Biol

5:1–13

Boroffka NGO (2010) Archaeology and its relevance to climate

and water level changes: a review. In: Kostianoy AG,

Kosarev AN (eds) The Aral Sea Environment. The

J Paleolimnol

123

Handbook of Environmental Chemistry. Springer. doi:10.

1007/698_2009_1

Chaturvedi M, Datta K, Nair PKK (1998) Pollen morphology of

Oryza (Poaceae). Grana 37:79–86

Crootof A (2011) Assessing water resources in Khorezm,

Uzbekistan for the development of aquaculture. M.S.

Thesis, University of Nevada, Reno, p 106

Crootof A, Mullabaev N, Saito L, Atwell L, Rosen MR, Bek-

chonovaM, Ginatullina E, Scott J, Chandra S, Nishonov B,

Lamers JPA, Fayzieva D (2015) Hydroecological condi-

tion and potential for aquaculture in lakes of the arid region

of Khorezm, Uzbekistan. J Arid Environ 117:37–46.

doi:10.1016/j.jaridenv.2015.02.012

Elpiner LI (1995) Medical–ecological problems of the eastern

Aral-land. GeoJournal 35:43–46

Faegri J, Iversen K (1989) Textbook of pollen analysis, 4th edn.

Hafner Press, New York

Fayzieva DKh, Atabekov NS, Usmanov IA, Azizov AA, Ste-

blyanko SN, Aslov S, Khamzina AS (2004) Hydrosphere

and health of population in the Aral Sea Basin. In: Fayzieva

D (ed) Environmental health in central Asia: The present

and the future. WIT Press, Southampton, pp 51–100

Galiulin RV, Bashkin VN (1996) Organochlorinated com-

pounds (PCBs and insecticides) in irrigated agrolandscapes

of Russia and Uzbekistan. Water Air Soil Pollut

89:247–266

Gintzburger G (2003) Rangelands of the arid and semi-arid

zones in Uzbekistan. ICARDA, Montpellier, p 426

Heiri O, Lotter AF, Lemcke G (2001) Loss-on-ignition as a

method for estimating organic and carbonate content in

sediments: reproducibility and comparability of results.

J Paleolimnol 25:101–110

Hill DW,McCarty PL (1967) Anaerobic degradation of selected

chlorinated hydrocarbon pesticides. J Water Pollut Control

Fed 39:1259–1277

IbrakhimovM,Martius C, Lamers JPA, Tischbein B (2011) The

dynamics of groundwater table and salinity over 17 years

in Khorezm. Agric Water Manag 101:52–61

Jensen S, Mazhitova Z, Zetterstrom R (1997) Environmental

pollution and child health in the Aral Sea region in Kaza-

khstan. Sci Total Environ 206:187–193

Kaiser B (2005) The volume of water reservoirs in Khorezm,

Uzbekistan. Internal report. ZEF/UNESCO Khorezm

Project, Urgench, Uzbekistan and University of Applied

Sciences, Nordostniedersachsen, Germany

Kapp RO, Davis OK, King JE (2000) Ronald O. Kapp’s pollen

and spores, 2nd edn. American Association of Stratigraphic

Palynologists, College Station

Karimov B, Lieth H, Kurambaeva M, Matsapeva I (2005) The

problems of fishermen in the southern Aral Sea region.

Mitig Adapt Strateg Glob Chang 10(1):87–103

Klaminder J, Appleby P, Crook P, Renberg I (2012) Post-de-

position diffusion of 137Cs in lake sediment: implications

for radiocaesium dating. Sedimentology 59:2259–2267

Korosi JB, Cheng W, Blais JM (2015) Organic pollutants in

sediment core archives. In: Blais JM, Rosen MR, Smol JP

(eds) Environmental contaminants: using natural archives

to track sources and long-term trends of pollution.

Springer, Dordrecht, pp 161–185

Letolle R, Chesterikoff A (1999) Salinity of surface waters in

the Aral Sea region. Int J Salt Lake Res 8:293–306

Letolle R, Mainguet M (1993) Aral. Springer, Paris, p 357

Li YF, Zhulidov AV, Robarts RD, Korotova LG (2004) Hex-

achlorocyclohexane use in the former Soviet Union. Arch

Environ Contam Toxicol 48:10–15

Manz M, Wenzel KD, Dietze U, Schuurmann G (2001) Per-

sistent organic pollutants in agricultural soils of central

Germany. Sci Total Environ 277:187–198

Martius C, Froebrich J, Nuppenau EA (2009) Water resource

management for improving environmental security and

rural livelihoods in the irrigated Amu Darya lowlands. In:

Brauch HG (ed) Facing global environmental change:

environmental, human, energy, food, health and water

security concepts. Springer, Berlin, Heidelberg, pp 749–761

McAndrews JH, Berti AA, Norris G (1973) Key to the quater-

nary pollen and spores of the Great Lakes Region. Royal

Ontario Museum Life Sciences Miscellaneous Publication,

Royal Ontario Museum, Toronto, p 61

Micklin PP (1988) Dessication of the Aral Sea: a water manage-

ment disaster in the Soviet Union. Science 241:1170–1176

Oberkircher L, Shanafield M, Ismailova B, Saito L (2011)

Ecosystem and social construction—an interdisciplinary

case study of the Shurkul Lake landscape in Khorezm,

Uzbekistan. Ecol Soc 16(4):20. doi:10.5751/ES-04511-

160420

Pollanen R, Valkama I, Toivonen H (1997) Transport of

radioactive particles from the Chernobyl accident. Atmos

Environ 31:3575–3590

RD 52.24.71-88 (1988) Manual on organochlorine pesticides

and their metabolites determination in bottom sediments.

Rostov-on-Don. p 38 (in Russian)Rosen MR, Crootof A, Reidy, L, Saito L, Nishonov B, Scott JA

(2016) Datasets for determining the origin of shallow lakes

in the Khorezm Province, Uzbekistan, and the history of

pesticide use around these lakes. US Geological Survey

Data Release. https://dx.doi.org/10.5066/F7319T07

Rosen MR, Van Metre PC (2010) Assessment of multiple

sources of anthropogenic and natural chemical inputs to a

morphologically complex basin, Lake Mead, USA.

Palaeogeogr Palaeoclimatol Palaeoecol 294:30–43

Santisteban JI, Mediavilla R, Lopez-Pamo E, Dabrio CJ, Ruiz

Zapata MB, Jose Gil-Garcıa M, Castano S, Martınez-Al-

faro PE (2004) Loss on ignition: A qualitative or quanti-

tative method for organic matter and carbonate mineral

content in sediments? J Paleolimnol 32:287–299

Scott J, Rosen MR, Saito L, Decker D (2011) The influence of

irrigation water on the hydrology and lake water balance of

two small arid-climate lakes in Khorezm, Uzbekistan.

J Hydrol 410:114–125

Severskiy I, Chervanov I, Ponomarenko Y, Novikova NM,

Miagkov SV, Rautalahti E, Daler D (2005) Aral Sea—

GIWA regional assessment 24. University of Kalmar on

behalf of United Nations Environment Programme, Kalmar

Shanafield M, Rosen MR, Saito L, Chandra S, Lamers J, Nis-

honov B (2010) Identification of nitrogen sources to four

small lakes in the agricultural region of Khorezm,

Uzbekistan. Biogeochem J 101:357–368

Sharom MS, Miles JRW, Harris RC, McEwen FL (1980) Persis-

tence of 12 insecticides in water. Water Res 14:1089–1093

Smol JP (2008) Pollution of lakes and rivers: a paleoenviron-

mental perspective, 2nd edn. Blackwell Publishing,

Oxford, p 383

J Paleolimnol

123

Stockmarr J (1971) Tablets with spores used in absolute pollen

analysis. Pollen Spores 13:615–621

Tanton TW, Heaven S (1999) Worsening of the Aral Basin

crisis: Can there be a solution? J Water Resour Plan Manag

125:363–368

The Governmental Portal of the Republic of Uzbekistan (2016)

Uzbekistan Territory. https://www.gov.uz/en/pages/territory

Tolstov SP (2005) Following the tracks of ancient Khorezmian

civilization. United Nations Educational, Scientific and

Cultural Organization. Originally published in Russian in

1948 by the Publishing House of the Academy of Sciences

of the USSR, Moscow-Leningrad, p 313

Tornqvist R, Jarsjo J, Karimov B (2010) Health risks from large-

scale water pollution: trends in Central Asia. Environ

Internat 37:435–442

[UNESCO] United Nations Educational, Scientific, and Cultural

Organization—Division of Water Resources (2000) Water

related vision for the Aral Sea basin for the year 2025. http://

unesdoc.unesco.org/images/0012/001262/126259mo.pdf

Veldwisch GJA (2008) Cotton, rice and water. The transfor-

mation of agrarian relations, irrigation technology and

water distribution in Khorezm, Uzbekistan. Ph.D. Disser-

tation, University of Bonn, Bonn, p 220

Wehrheim P, Schoeller-Schletter A, Martius C (eds) (2008)

Continuity and change: land and water use reforms in rural

Uzbekistan: Socio-economic and legal analyses from the

region Khorezm. Halle/Saale, Germany: Leibniz-Institut

fur Agrarentwicklung in Mittel- und Osteuropa (IAMO).

http://www.iamo.de/dok/sr_vol43.pdf

Wolfe LN, Zepp RG, Paris DR, Baughman GL, Hollis RC

(1977) Methoxychlor and DDT degradation in water: rates

and products. Environ Sci Technol 11:1077–1081

Zweig RD, Morton JD, Stewart M (1999) Source water quality

for aquaculture: a guide for assessment. The World Bank,

Washington, D.C. https://www.extension.org/mediawiki/

files/e/ed/Source_Water_Quality_for_Aquaculture_A_

Guide_for_Assessment.pdf

J Paleolimnol

123