Environmental Pollution 284 (2021) 117405

Contents lists available at ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

Agricultural impacts drive longitudinal variations of riverine water quality

of the Aral Sea basin (Amu Darya and Syr Darya Rivers), Central Asia☆
Peifang Leng a, b, c, Qiuying Zhang d, Fadong Li a, b, *, Rashid Kulmatov e, Guoqin Wang a, f,
Yunfeng Qiao a, Jianqi Wang a, Yu Peng a, b, Chao Tian a, Nong Zhu a, Hubert Hirwa a, b,
Sayidjakhon Khasanov g
a
Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences,
Beijing, 100101, China
b
College of Resources and Environment, University of Chinese Academy of Sciences, Beijing, 100190, China
c
Department of Lake Research, Helmholtz Centre for Environmental Research-UFZ, 39114, Magdeburg, Germany
d
Chinese Research Academy of Environmental Sciences, Beijing, 100012, China
e
Department of Biology, National University of Uzbekistan, Tashkent, 100170, Uzbekistan
f
International Ecosystem Management Partnership, United Nations Environment Programme, Beijing, 100101, China
g
Tashkent Institute of Irrigation and Agricultural Mechanization Engineers, Tashkent, 100000, Uzbekistan

A R T I C L E I N F O A B S T R A C T

Keywords: River ecosystems are under increasing stress in the background of global change and ever-growing anthropogenic
Amu Darya impacts in Central Asia. However, available water quality data in this region are insufficient for a reliable
Syr Darya assessment of the current status, which come as no surprise that the limited knowledge of regulating processes
Rivers
for further prediction of solute variations hinders the development of sustainable management strategies. Here,
Salinization
Nutrients
we analyzed a dataset of various water quality variables from two sampling campaigns in 2019 in the catchments
Aral Sea basin of two major rivers in Central Asia—the Amu Darya and Syr Darya Rivers. Our results suggested high spatial
heterogeneity of salinity and major ion components along the longitudinal directions in both river catchments,
pointing to an increasing influence of human activities toward downstream areas. We linked the modeling
outputs from the global nutrient model (IMAGE-GNM) to riverine nutrients to elucidate the effect of different
natural and anthropogenic sources in dictating the longitudinal variations of the riverine nutrient concentrations
(N and P). Diffuse nutrient loadings dominated the export flux into the rivers, whereas leaching and surface
runoff constituted the major fractions for N and P, respectively. Discharge of agricultural irrigation water into the
rivers was the major cause of the increases in nutrients and salinity. Given that the conditions in Central Asia are
highly susceptible to climate change, our findings call for more efforts to establish holistic management of water
quality.

1. Introduction recent decades to less than 10% of the original volume in the driest years
(Wurtsbaugh et al., 2017), owing to intensive and continuous irrigation
Access to water resources is essential for human health (Carpenter water withdrawal from the Amu Darya River (ADR), Syr Darya River
et al., 2009); (UNHCR, 2017). In Central Asia, however, limitations in (SDR), and Zarafshan River in the Aral Sea basin (FAO, 2012; Rzymski
both the quantity and quality of water resources have become the major et al., 2019).
concerns of regional socioeconomic development (Karthe et al., 2017; Surface water quality in Central Asia is impaired by excessive
Unger-Shayesteh et al., 2013). The Aral Sea, once the fourth largest lake salinity, agrochemicals, organic pollutants, and heavy metals due to
in the world, experienced a dramatic shrinking in water volume in agricultural drainage collector discharge (Crosa et al., 2006; Groll et al.,

This paper has been recommended for acceptance by Sarah Harmon.

☆

* Corresponding author. Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Natural Resources Research,
Chinese Academy of Sciences, Beijing, 100101, China.
E-mail address: lifadong@igsnrr.ac.cn (F. Li).

https://doi.org/10.1016/j.envpol.2021.117405
Received 3 February 2021; Received in revised form 13 April 2021; Accepted 16 May 2021
Available online 19 May 2021
0269-7491/© 2021 Elsevier Ltd. All rights reserved.
P. Leng et al. Environmental Pollution 284 (2021) 117405

2015b; Reyer et al., 2017). Growing concerns call for more scientific we conducted basin-scale field sampling campaigns along the ADR and
investigations on the assessment of river water quality to provide critical SDR and measured a set of water physical and chemical parameters. We
guidance for local water management and sustainable development. longitudinally analyzed the salinity and nutrient concentrations and
Despite the increasing availability of water quality data in river eco­ correlated them with various nutrient sources by the Integrated Model to
systems, there is a significant under-representation of inland waters in Assess the Global Environment–Global Nutrient Model (IMAGE-GNM).
Central Asia (Janssen et al., 2019; Lioubimtseva et al., 2005), which We further assessed how river salinity and nutrients were regulated by
frequently hampers the proper evaluation of the current water quality anthropogenic activities. Water quality assessments of these two major
issues on the global scale (Gerlitz et al., 2020; Sehring, 2020). rivers in the Aral Sea basin offer the possibility to elucidate the effects of
The ADR and SDR are the two major riverine inflows to the Aral Sea, catchment properties and human drivers on salinity and nutrient efflu­
serving as important components of the element cycles and key factors ents from rivers in Central Asia. From our findings, we speculated on the
of the development of catchment areas regarding hydropower and implications of water management for riverine ecosystems in these two
agriculture. However, these two rivers suffer from both climate change major rivers and potentially for the whole of Central Asia. Holistic water
and increases in perturbations from human activities in the catchments management would assist in restoring the riverine ecosystems in both
(Groll et al., 2015b; Haag et al., 2019). The majority of the Aral Sea basin river basins to combat the detrimental effects of human pressure and
experiences arid climate conditions; a general deficit of freshwater is climate change.
prevalent in this region. Specifically, the area of the Pamir Alay and
Western Tien Shan glaciers, the origin of the ADR and SDR, shrunk by 2. Material and methods
16.8% and 25% during the period of 1960–2005, respectively (Borisova,
2012). 2.1. Study area and sampling sites
Human impacts on water quantity and quality are apparent.
Regarding agricultural activity, a large amount of water has been Central Asia experiences a strongly continental arid to semiarid
withdrawn for irrigation from the ADR and SDR that subsequently flows climate, with hot cloud-free summers and humid temperate winters in
back into the rivers (Groll et al., 2015a). Water needs for irrigation the south and cold winters in the north (O’Hara, 2000). The average
remain a priority even under minimal flow conditions during the annual precipitation for Central Asia is about 273 mm. The ADR and
growing season, resulting in significant pressures on maintaining the SDR basins lie within 57◦ 36′ 29′′ –79◦ 13′ 01′′ N, 34◦ 28′ 59′′ –51◦ 45′ 47′′ E.
base flow needed for downstream ecosystems. A considerable amount of The ADR has an average annual water flow of 78 km3 (UNEP, 2011). Its
river water is deteriorated in quality, predominantly due to the total length from its headstream, the Panj River, is 2540 km. Its catch­
discharge of drainage water and evaporation in the catchment (Groll ment area is 534,739 km2. The SDR is the longest river in Central Asia
et al., 2015a). Alterations of the hydrology often reinforce such hydro­ (3019 km) and its basin is estimated at 782,617 km2, arising at the
geochemical changes of the rivers. In addition, modern human activities confluence of the Naryn and Kara Darya rivers in Uzbekistan and
demand higher food and energy production, which helps accelerate N draining into the Aral Sea. However, its annual flow is much smaller
and P mobilization throughout the hydrosphere (Bouwman et al., 2009). than that of the ADR, on average 37 km3 (Saidmamatov et al., 2020).
Most economic activities using water and land resources contribute to Melted snow and glaciers feed both rivers, and the maximum discharge
pollution, with negative impacts on the water quality in the middle and is observed in summer (begins in April and peaks in June) and minimum
downstream parts of both ADR and SDR. Water pollution in Central Asia in January and February. The year-round availability of the flow favors
has three potential major causes: agricultural discharge from irrigated the use of the river water for irrigation during the dry summer. Both
areas (Törnqvist et al., 2015), wastewater discharge from industrial rivers are characterized as transboundary rivers that flow through
activities (e.g., mining, construction, manufacturing, petrochemicals, several countries of Central Asia before ending in the northern part
etc.) (Bezborodov et al., 2010), and improper municipal wastewater (SDR) and southern part (ADR) of the Aral Sea. Numerous irrigation
management in expanding urban areas (Morris et al., 2006). These ac­ canals were constructed in the 1960s due to major water diversion and
tivities take place in all countries in Central Asia and significantly affect agricultural irrigation schemes. Furrow irrigation is the predominant
water quality and human health (Bekturganov et al., 2016). Finally, the irrigation method within both basins (Saiko and Zonn, 2000).
combined effects of climate change and land reclamation for agriculture We divided the main ADR into two reaches: the upper reach (up­
and mining in some regions are important drivers of the degradation of stream of Turkmenabat, Turkmenistan), where most of the water flow is
both forests and riparian zones (Batbayar et al., 2019; Karthe et al., generated and most cities are distributed, and the lower reach (down­
2017), which subsequently has a large negative impact on water stream of Turkmenabat), which flows through the desertic and semi-
resources. desertic part of the basin. The main SDR was divided into two rea­
A lack of long-term data hampers further observations on the current ches: the upper reach (upstream of Kyzyl-Orda, Kazakhstan), where
conditions, impeding the creation of strategies for improving regional most of the water flow is generated, and the lower reach (downstream of
water management (Crosa et al., 2006; Unger-Shayesteh et al., 2013; Kyzyl-Orda). The upper reach of the SDR irrigates the most productive
Xenarios et al., 2019). Salinity, as a major factor limiting the distribution cotton-growing region in Central Asia (Rzymski et al., 2019).
of biota, can vary from the tens of milligrams to hundreds of grams per The longitudinal variations in water quality were investigated using
liter in inland waters (Williams, 1987), which has deleterious effects on data from two sampling campaigns organized on April 15–21 and July
freshwater habitats and organisms (Israël et al., 2014). Moreover, the 1–8, 2019, respectively. We tried to distribute the sampling sites along
ionic compositions and nutrients of river water provide critical infor­ both river systems as evenly as possible, whereas due to the remoteness
mation on the alkalization and trophic states of the water, which are and difficulties in transportation, some sampling sites were not reach­
valuable for increasing irrigation and fertilizer efficiencies (Liu et al., able. In total, 29 and 17 river water samples were collected in the ADR
2019). Currently available datasets in Central Asia were mostly and SDR basins, respectively, including the mainstreams and the cor­
addressing water salinity (Cañedo-Argüelles et al., 2018; Ma et al., 2019; responding tributaries. The locations of the sample sites are shown in
Qadir et al., 2009), whereas far less data for key nutrients (nitrogen and Fig. 1.
phosphorus) have been involved (Jarsjö et al., 2017; Törnqvist et al.,
2015). We realize that the in-situ nutrient data availability in this region 2.2. Analytical methods
would be highly beneficial for understanding the drivers of the deteri­
oration of river water quality. At each sampling site, a 1.5-L water sample was taken at 10 cm below
To address the limitation on water quality data and estimate the the surface. We collected samples carefully in a clean glass container
potential impacts of human activities on water quality in Central Asia, where the river was shallow to prevent mixing with sediment. pH, water

2
P. Leng et al. Environmental Pollution 284 (2021) 117405

Fig. 1. Geographical locations and sampling sites of the Amu Darya and Syr Darya River catchments in Central Asia (left) and land use cover (right). The land cover
map is derived from the Copernius Global Land Service from 2019, which is available at https://land.copernicus.eu/global/products/lc.

temperature, dissolved oxygen (DO), and electrical conductivity (EC) the area, depth and residence time of water bodies, water flux direction
were measured in situ using multi-parameter probes (Hach H40d, USA). and discharge in rivers.
Samples were prepared within one day for filtration and stabilization. Diffuse N enters the rivers through surface runoff, groundwater, and
Water samples were first filtered using 0.45 μm cellulose acetate fabric leaching, while diffuse P is transported to rivers through surface runoff
filters for subsequential measurements of NO−3 -N, NH+
4 -N, dissolved total and soil erosion. Both N and P retention in each water body are calcu­
phosphorus (DTP), Mg, Ca, Na, and K concentrations. All samples were lated on the basis of estimated water residence time and nutrient uptake
kept cool and dark in high-density polyethylene bottles until analyses. velocity. After retention in waterbodies, water and nutrients are trans­
For nutrient stabilization, we poisoned the water samples with saturated ported to downstream grid cells. More details of the model can be found
HgCl2 to stop microbial activity. The concentrations of the chemical in Beusen et al. (2015). Altogether, the IMAGE-GNM model takes into
components of river water were determined at the Center for Physical account the hydrology, nutrient input into surface water, and in-stream
and Chemical Analysis of the Institute of Geographic Sciences and retention, thus can be used to explore changes in various basin-scale
Natural Resources Research (Beijing, China) in less than 12 days. Cation nutrient processes, and the effect on the functioning of downstream
concentrations, including Ca2+, Mg2+, Na+, and K+, were determined by ecosystems.
inductively coupled plasma–optical emission spectrometry (ICP-OES)
(PerkinElmer, USA). SO2− 4 , Cl , and NO3 were determined using the ion
− −
2.4. Data analysis
chromatography method (Dionex ICS 900; Dionex, USA). HCO−3 and
CO2−3 were analyzed by inverse titration using 0.01 M H2SO4. NH4 was
+
First, all data were quality checked using scatter plots. Before dis­
determined using the automated phenate method. DTP was determined carding, observed outliers and suspicious data were re-checked in the
by ICP-OES (Worsfold et al., 2016). Water salinity was express as the analytical lab. Values below the detection limits were substituted by half
sum of the measured main ions (cations and anions) in a 0.5-L volume of of the detection limit. Before statistical analysis, data were log-
the water sample (Text S1 for quality control, precision and detection transformed to reduce heteroscedasticity and normalize model re­
limits of water chemical parameters). siduals to meet assumptions of normality. All nutrient constituents were
plotted with salinity to determine if changes in concentration were
2.3. Nutrient loading of the IMAGE-GNM conservative (due to mixing) or non-conservative (due to biological or
physical processes). To depict the overall main hydrochemical types of
We utilized the outputs from the IMAGE-GNM to identify the sources river water, we applied Piper diagrams (Piper, 1944) to classify water
of the nutrients in the studied rivers. IMAGE-GNM is a spatial explicit 2D chemical types.
model that uses the grid cell (0.5 × 0.5◦ ) as the basic hydrological unit. Pearson correlation analysis was applied to determine the signifi­
The model simulates the nutrient (N and P) delivery to the surface water cance of the correlations between salinity and nutrient components. We
via diffuse (e.g., agriculture, natural ecosystems, aquaculture) and point tested the correlations between nutrients with different N and P sources
sources (e.g., wastewater) to analyze the annual nutrient dynamics in to determine the major contributions of river nutrient conditions. We
rivers (Beusen et al., 2016). In specific, IMAGE estimates nutrient used the nutrient loading estimates derived from human and natural
loading from various sources in the natural and human systems. The nutrient inputs and outputs in the year of 2010 provided by IMAGE
nutrient loading data is then used as the input to the GNM (Global which has been calibrated and validated (Bouwman et al., 2017; Zhang
Nutrient Model), which simulates nutrient transport and transformation et al., 2017). Here, we assumed the nutrient loading data from 2010
in each grid cell, including nutrient from upstream grid cells, and diffuse were comparable to those from 2019.
and point sources within the grid cell, and the nutrient removal pro­ We further conducted a generalized linear model (GLM) analysis to
cesses. The hydrological data in each cell are provided by the global quantify the relative importance of each nutrient source in dictating
hydrological model (PCR-GLOBWB) (van Beek et al., 2011), including variations in NO−3 and DTP concentrations in river waters. The analysis

3
P. Leng et al. Environmental Pollution 284 (2021) 117405

was performed using the function ‘calc.relimp’ in the package ‘relaimpo’ Considering higher annual discharge, the annual salinity loads to the
of R (Grömping, 2006). Aral Sea are likely much higher from the SDR.
Moreover, we graphically presented the results of our sampling The major ion concentrations (Ca2+, Mg2+, Na+, K+, SO2− 4 , Cl ,
−
2−
campaigns using QGIS 3.12 software (QGIS Development Team, 2019). HCO3 , and CO3 ) were highly spatially variable with a significant lon­
−

For nutrient source identification, we extracted the nutrient budgets gitudinal increase in both rivers. The median major anion concentra­
from the grid cells in which sampling sites were located using the plugin tions followed the order SO2− 4 > Cl > HCO3 . The major cations were
− −

“point sampling tool”. To identify the effects of agricultural land and Na+, Ca2+, and Mg2+. It could be attributed to the fact that solonchaks
population distribution on nutrient loadings, we acquired population are the main geological feature of the riparian soils (saline soil domi­
density data from the WorldPop website (https://www.worldpop.org/). nated by neutral soluble salts: Cl, Na, SO4, Ca, and Mg) (Crosa et al.,
The land cover data were derived from the Copernius Global Land Service 2006). Cation and anion proportions did not differ substantially between
from 2019. We calculated the agricultural land area for each grid cell; the mainstreams and tributaries in both rivers, except anion composi­
then, we checked the Spearman correlations with nutrient loadings in tions varied greatly in the SDR, which could be ascribed to different
the model. All statistical analysis was performed using R ver. 3.6.2 (R hydrogeological conditions within the basin.
Core Team, 2019). Raw data were generated at the Institute of Our results highlighted the differences of longitudinal variations
Geographic Sciences and Natural Resource Research. The data that between nutrients and salinity. Samples with high NO−3 concentrations
support the findings of this study are available from the corresponding were consistently located in regional farmlands and urban areas. NO−3
author, upon reasonable request. peaked (5864 μg L− 1) in the tributaries of the upper reach of the SDR. In
the upper reach of the ADR, NO−3 concentrations were relatively low and
3. Results and discussion less variable, despite higher NO−3 concentrations in the tributaries and
the lower reach. NO−3 concentrations in the SDR were higher in the
3.1. Overall chemical status of rivers upper reach (2790–5801 μg L− 1) compared to the lower reach (56–467
μg L− 1).
The analytical data measured from two sampling campaigns are DTP peaked (2076 μg L− 1) in the tributaries of the upper reach of the
summarized in Table S1. The observed parameters varied considerably ADR. It was consistently lower in the lower reach of the mainstream of
in the ADR and SDR. Regarding salinity, there was no significant dif­ ADR (4–142 μg L− 1) with higher DTP at tributary sites, suggesting large
ference between the ADR and SDR (p = 0.46, Wilcoxon signed-rank nutrient inputs from the tributaries. We noted that the sites with high
test), with salinity ranging from 152 to 3838 mg L− 1 in the ADR and DTP concentrations were mostly surrounded by large areas of irrigated
175–5990 mg L− 1 in the SDR. In both rivers, salinity increased towards agricultural land, such as Jizzakh Province, Uzbekistan, where irrigated
downstream and reached a maximum near the Aral Sea. Salinity agriculture is widely distributed (Kulmatov et al., 2020). As DTP data in
increased acutely in the downstream part of the SDR (Fig. 2), in contrast the upper reach of the SDR could not be directly used, we did not assess
to the gentle increase through the river continuum of the ADR. the spatial patterns of DTP in the SDR.

Fig. 2. Spatial distribution of salinity (a), nitrate (b), dissolved total phosphorus (DTP) (c), sodium absorption ratio (SAR value) (d), and major ionic concentrations
(anions in (e) and cations in (f)) of the sampling sites in the Amu Darya and Syr Darya Rivers.

4
P. Leng et al. Environmental Pollution 284 (2021) 117405

3.2. Hydrochemical characteristics shifts in ionic proportions and concentrations under strong mixing ef­
fects. Thus, high variability of the major ion components in both rivers
The major hydrochemical type of water samples was Ca–Mg–SO4–Cl, was likely due to the discharge of irrigation drainage water. Extreme
which accounted for 85% of the water samples, followed by the values observed in tributaries suggested that the tributaries were
Ca–Mg–HCO3 type (Figure S1). Most of the river waters were charac­ densely affected by the weathering of rock and soil in the drainage basin.
terized by significantly more alkaline earth elements (Ca2+ + Mg2+) Combined with the evidence of the longitudinal increase in salinity in
than alkali elements (Na+ + K+). Strong acids (SO2− 4 + Cl ) exceeded
−
both rivers, we speculated that limited rainfall and low drainage density
weak acids (HCO−3 ). The tributary samples of the ADR consistently had a of the catchment promoted river salinization.
greater proportion of Cl− and lower proportion of SO2− 4 than the To assess whether the river water is suitable for irrigation, we
mainstream samples (Figure S1). In the SDR, tributaries had a greater applied the sodium absorption ratio (SAR) to evaluate the degree of
proportion of SO2−4 and lower proportion of Cl than the mainstream.
−
alkalization of irrigation water (Text S2 for detailed methods). Our re­
Considering the intrusion of municipal sewage and industrial waste­ sults showed that 24% of the samples were not suitable for irrigation,
water, it is plausible the river waters would shift to have greater pro­ which were collected from the downstream and the tributaries. The
portions of SO2−4 and Cl (Li et al., 2010), which was supported by our
−
variations of SAR were consistent with salinity (Fig. 2). 22% of the water
data with higher proportions of SO2− 4 and Cl distributed in tributaries.
−
samples, which were derived from the upstream parts, were excellent for
High variations of hydrochemical types reflected the heterogeneous irrigation. Due to widely distributed irrigated lands (Fig. 1), little nat­
distribution of the agricultural land and urban areas along the rivers ural water recharge, and strong evapotranspiration alongside both rivers
(Figure S1). The soil texture in both basins consists of sands, clays, and (Shibuo et al., 2007), the accumulation effect of salinity is greatly
silt loam (Chen et al., 2017; Funakawa and Kosaki, 2007; Qadir et al., related to agricultural irrigation. Unraveling the contribution of irriga­
2009). Due to well water permeability and poor water and fertilizer tion lands to river water quality will require more investigations of
retention performance of sandy clay, irrigation practices discharge wa­ nutrient loadings within the basins.
ters containing salt and fertilizer back into the river channels, leading to

Fig. 3. Total nitrogen and phosphorus loads in the Amu and Syr Darya Catchments from the IMAGE-GNM model (unit: kg grid− 1 yr− 1, each grid 0.5 ◦ × 0.5 ◦ ) with
spatial distribution of agricultural area in each grid cell (unit: km2, calculated from land use data) and average population density (numbers of persons in each grid
cell, derived from WorldPop website, https://www.worldpop.org/). The total nitrogen loads could be divided into point and diffusive nitrogen, which includes the
leaching process with the surface runoff and soil erosion. The total phosphorus loads could be divided into point and diffusive phosphorus, which includes the surface
runoff and soil erosion. The point loads are colored in blue; the diffuse loads are showed in green; and population density and agricultural area are in red. (For
interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

5
P. Leng et al. Environmental Pollution 284 (2021) 117405

3.3. Sources of nutrients in rivers importance of each nutrient source in governing the spatial variations of
riverine NO−3 concentrations (Table 2). The results confirmed that point
3.3.1. IMAGE-GNM model results sources (57.7%) and leaching processes (42.2%) were the predominant
We applied the robust model results for further estimation of the drivers of riverine NO−3 variations in the ADR and SDR, respectively.
linking between nutrient loading and riverine concentrations. A detailed Considering the consistency of the distributions between croplands
description of IMAGE-GNM model with additional validations was and grid cells with high N leaching, polluted drainage effluents from
provided in Beusen et al. (2015). The model results are in good agree­ irrigated areas likely leached into the drainage systems and entered the
ment with long-term nutrient concentration measured in several rivers river channels, resulting in the diffuse source strongly correlating with
around the globe without calibration (Beusen et al., 2016; Liu et al., river NO−3 . Given the evidence that N leaching contributed more than
2018). The contributions of point and diffuse sources to nutrient loads the other two subsections (surface runoff and soil erosion), we suggest
were relatively comparable between the ADR and SDR basins (Fig. 3). that leaching, which is associated with the hydrologic connectivity of
For both N and P, diffuse sources were higher than point sources. Diffuse groundwater and wetland soils to deliver nutrients via subsurface
N and P were primarily determined by leaching and surface runoff drainage, determined the N budgets in rivers. By contrast, when N enters
among all transport pathways, respectively. Diffuse NO−3 load signifi­ rivers with the help of more irrigation water and increased fertilizer use,
cantly correlated to the cropland area (Table S2, p < 0.001, R2 = 0.38), N fertilizer discharge directly increases soil N losses from both surface
indicating a remarkable contribution of agricultural input in the model. runoff and subsurface leaching (Liu et al., 2019). Combined with the
Point sources of N and P significantly correlated to the population land-use data in both catchments, we found that irrigated croplands,
density, with high coefficients of determination (R2 = 0.74 for both). which had a significant positive relationship with NO−3 (Table 1), may
The population distribution and diffuse sources of N and P were also control river N through leaching processes with changing hydrology and
strongly related (Table S2). Compared to the different P loads in the two contamination.
basins, N loads were more highly correlated to croplands/population, The uncorrelation of NO−3 with the diffuse source is likely due to the
implying that N sources were more susceptible to the spatial variability sampling periods and the locations of irrigation agriculture systems. We
in agriculture and population. NO−3 loadings varied in different sections sampled the ADR basin in April when croplands intensively required
of the rivers, with highest NO−3 loads observed in the upper reaches and irrigation. While during the field sampling, the river flows were too low
the lowest loads in the lower reaches of both basins (Fig. 3). to support irrigation and were subsequently unable to hydrologically
connect to riparian land delivering solutes from shallow soil layers
3.3.2. Nitrate (Rasilo et al., 2017). Drainage effluent quality greatly depends on local
We established Pearson correlations between NO−3 concentrations hydrologic characteristics. The weaker correlation between agricultural
and different N sources from the IMAGE-GNM results for both rivers to areas and NO−3 concentrations in the ADR also proves that agricultural
find out the drivers controlling the longitudinal variations of NO−3 land was less related to river NO−3 during the sampling period (Table 1).
concentration in each river. Point-source N was dominated by NO−3 in Furthermore, less irrigated agriculture land in the ADR catchment
the ADR with a rather weak response (R2 = 0.17, p < 0.05; Table 1), geographically connected to the mainstream, resulting in a lack of sig­
whereas diffuse N significantly contributed to river NO−3 only in the SDR nificant relationship with diffuse-source N. Thus, we surmise that dis­
(R2 = 0.78, p < 0.001; Table 1). Our findings indicated that N point and charges from water treatment plants and direct sewage inputs,
diffuse sources had different influences on NO−3 concentrations in the substituting the diffuse sources, governed riverine NO−3 in the ADR
two rivers. We hypothesized that diffuse N sources would dominate river during the sampling period. Comparing lower NO−3 in ADR dominated
NO−3 inputs; however, point sources, such as sewage and wastewater by point sources to higher one in SDR dominated by diffuse sources, we
discharge from urban and rural populations, were the important driver infer larger impacts of diffuse sources on riverine NO−3 . Among all
of river NO−3 in the ADR. The strong correlations between the diffuse diffuse N sources, controlling leaching processes from agricultural ac­
subsections (leaching, surface runoff, and soil erosion of N) and river tivities should be the most effective management practice on N removal
NO−3 in the SDR indicate that these subsections combinedly contribute to from river waters.
river NO−3 . GLM was further performed to quantify the relative
3.3.3. Dissolved total phosphorus
The different responses of NO−3 (p < 0.001) and DTP (p > 0.05)
Table 1 concentrations to salinity suggested different sources and trans­
Simple linear regression among variables in the Amu Darya and Syr Darya rivers.
formation pathways of nutrient species. This suggested that in-stream
All data have been log-transformed, statistical significance “***” represents p <
removal processes or soil sorption did not dominate the variations in
0.001; “**” represents p < 0.01; “*” represents p < 0.05. N (Point) and P (Point):
the point N and P sources from the result of IMAGE-GNM model in both
nutrient concentrations (Lintern et al., 2018; Varanka et al., 2015). Since
catchments; N (Diff) and P (Diff): the diffuse N and P sources, respectively; N if that had been the case, the nutrient variations in rivers would be
(Runoff) and P (Runoff): the surface runoff loads of N and P, respectively; N positively correlated. Samples with high DTP were always bound to
(Erosion) and P (Erosion): the soil erosion loads of N and P, respectively; N urban locations. Similarly, Pratt and Chang (2012) found a positive
(Leaching): the N leaching load; Cropland: the area of cropland in the grid cells. correlation between DTP concentrations at 14 river sites and the
Amu Darya River Syr Darya River

Coefficient p R 2
Coefficient p R2 Table 2
GLM analysis of the relationships between nutrient concentrations (NO−3 and
NO−3~ Salinity − 0.52 *** 0.43 − 0.54 *** 0.82
NO−3~ N (Point) 0.11 * 0.17 0.11 0.59 0.02 DTP) and corresponding explanatory variables from IMAGE-GNM model in the
NO−3~ N (Diff) 0.03 0.80 0.00 0.54 *** 0.78 Amu Darya (n = 28) and Syr Darya (n = 16) rivers.
NO−3~ N (Leaching) − 0.01 0.84 0.00 0.24 *** 0.68 Variables NO−3 DTP
NO−3~ N (Runoff) − 0.02 0.76 0.01 0.33 ** 0.49
NO−3~ N (Erosion) 0.08 0.59 0.02 0.54 ** 0.55 Amu Syr Amu Syr
NO−3~ Cropland 0.02 * 0.20 0.03 *** 0.57 SE RI SE RI SE RI SE RI
DTP ~ Salinity 0.06 0.40 0.03 − 0.66 0.30 0.13
DTP ~ P (Point) 0.15 0.36 0.06 − 0.04 0.59 0.04 Point 0.07 57.7 0.11 5.0 0.14 6.9 0.04 6.7
DTP ~ P (Diff) − 0.44 0.18 0.12 0.16 0.09 0.31 Leaching 0.07 12.2 0.10 42.2 – – – –
DTP ~ P (Runoff) 0.34 ** 0.38 0.15 ** 0.74 Surface runoff 0.13 8.0 0.14 23.8 0.19 60.4 0.06 66.4
DTP ~ P (Erosion) 0.70 * 0.27 0.20 * 0.40 Soil erosion 0.30 22.1 0.21 28.9 0.47 32.8 0.10 26.9
DTP ~ Cropland 0.02 * 0.32 0.04 * 0.44
SE, standard error; RI, relative importance (%) explained by the variable.

6
P. Leng et al. Environmental Pollution 284 (2021) 117405

proportion of the catchment urbanized in a study of American rivers. application of N fertilizer increased greatly with low fertilization effi­
Urban and agricultural areas generate both particulate and dissolved ciency (Kienzler et al., 2011). Large water diversions and water losses
nutrients through wastewater discharge (Suárez and Puertas, 2005), from rivers, which are often associated with irrigation, can considerably
manure and fertilizer application (Law et al., 2004), livestock waste modify water, nutrients, and element cycles. The evapotranspiration
(Arheimer and Lidén, 2000), and atmospheric deposition. return flow into the atmosphere was nearly 100% of the applied irri­
With respect to sources of DTP in both rivers, we observed that gation water in the Aral Sea basin irrigation (Shibuo et al., 2007).
diffusive-source P, as the primary contributor, enhanced the DTP con­ Strategies for reducing N leaching should be developed towards fertil­
centrations in both rivers. Among all subsections of diffuse sources of P, izer efficiency improvement, irrigation management, and technological
soil erosion considerably contributed to DTP in both rivers (ADR: R2 = interventions to support optimized management practices (Min et al.,
0.27, p = 0.03; SDR: R2 = 0.40, p = 0.05; Table 2). The results from GLM 2012). With respect to mitigating P losses by surface runoff, constructing
showed that the dominant process accounting for DTP variations in both more riparian boundaries to increase flow distances to streams and using
rivers was surface runoff (Table 2; 60.4 and 66.4% in the ADR and SDR, residue cover on the soil surface through conservation practices could be
respectively), whereas the point sources were the minor contributor. effective (Smith et al., 2016). It is also vital for the ADR to reduce point
P surface runoff occurs in P-saturated soils with long histories of sources of inputs. To this end, adequate treatment of wastewater and
heavy over-fertilization. P loss via surface runoff can cause a consider­ sewage would considerably improve the water quality, especially in
able depletion of soil P (Beusen et al., 2016). Considering land-use data, terms of DTP.
higher DTP was associated with croplands, showing that P transport
appeared to be linked to agricultural erosion processes. DTP was likely 3.4. Merits and implications
to be affected by particulate P discharges under high flows due to the
different processes by which these sources are retained, transformed, Our study reveals that the mechanisms behind the longitudinal
and assimilated within river channels (Withers and Jarvie, 2008). There patterns of riverine nutrients were related to the point and diffusive
is a high probability of particulate P penetrating drainage tiles, even­ nutrient sources throughout the basin, on the basis of the modeling
tually entering river channels. Thus, subsurface and in-stream processes approach integrating the effect of societal and economic development
likely explained the longitudinal variability in P in both rivers. Nutrient on the nutrient loading to the aquatic environment. Our study is the first
losses from agriculture and natural areas imply the contribution of soil to establish the direct causal relationship between multiple nutrient
nutrients to the surface water and eventually to the Aral Sea (Beusen sources and in-field measured fluvial nutrient concentrations. This
et al., 2016). Our findings highlight the need for more efficient measures approach distinguishes from previous studies evaluating the empirical
controlling P from surface runoff than other sources to reduce risk for contribution of natural and anthropogenic sources to nutrients in rivers
eutrophication in the downstream waterbodies. based on indices such as human population, land use cover and gross
We further checked whether there was an effect of upstream loads on domestic product (GDP) in the catchments (Ahearn et al., 2005; Bu et al.,
local nutrient variations. We did not detect any significant correlation 2016). The advances of IMAGE model were identified and served as a
between observed nutrient concentrations and upstream loading from robust basis to develop proper methodologies towards in-depth assess­
the grid cells from the model (p ranged between 0.06 and 0.88). One ment and management of river water quality, especially in the dryland
possible explanation could be the relatively short spiraling length of the agricultural catchments where excessive nutrients are difficult to
nutrients (from hundreds of meters to ~10 km for N and P in rivers manage to maintain water quality (Bostanmaneshrad et al., 2018; Kelley
(Iwata et al., 2020; Newbold et al., 1983; Newbold et al., 1981)). In et al., 2017). More specifically, it provides a good insight on quantifying
addition, our ability to disentangle the relative contribution from up­ the response of the water quality status to changes in point and
stream sections and local loading was also restricted by the spatial res­ non-point (diffusive) sources in both rivers.
olution of the model (0.5◦ on a global scale), which was much lower than In-situ data collection through a long-term continuous monitoring
the nutrient spiraling length. Thus, we conclude that in comparison to program with high spatiotemporal resolution is critical to clarify the
local nutrient loads, the effect of upstream loads in our study is of minor regulatory mechanisms of salinity and nutrients across different tem­
importance to the variations of nutrient concentrations. poral scales. Such information is also helpful to make new policies and
management strategies, considering altered climate regimes (e.g.,
3.3.4. Hydrologic controls on nutrients melting glaciers, changes in river runoff regimes, seasonal water short­
Salinity had a weakly negative correlation with discharge from the ages, and altered vegetation patterns) (Reyer et al., 2017; Seim et al.,
model in the ADR (p = 0.06), and no significant correlation in the SDR 2017). In the context of fulfillment of crop yield stability, expansion of
(p = 0.75). The lack of significant correlations between discharge and integrated agriculture might not only enhance the acceleration of
salinity may be associated with the model discharge data we used, salinity but also nutrients in riverine systems in the future, especially in
without consideration of temporal variations. Moreover, salinity in drylands where the impact of water scarcity accentuates threat (Jones
rivers may not change when agricultural return waters carrying a large and van Vliet, 2018; Kulmatov et al., 2015a; Vörösmarty et al., 2010).
number of minerals drain into the river channels (Crosa et al., 2006). Törnqvist et al. (2015) showed that observed decreases in riverine dis­
We found a significant negative correlation between NO−3 concen­ solved inorganic N in ADR were primarily due to increased recirculation
tration and salinity (ADR: p < 0.001, R2 = 0.43; SDR: p < 0.001, R2 = of irrigation water, which extended the flow path lengths and enhanced
0.82; Table 1). It could be due to a simple mixing between high-NO−3 N attenuation. Moreover, due to the special status of the ADR and SDR as
irrigation water and high-salinity river water. NO−3 discharge into international transboundary rivers, international trade, relations, and
streams was dominated by rapid irrigation flow pathways accompanied regulation can significantly affect the water utilization patterns. Thus,
by continuous river water use and evaporation along the rivers (e.g., information sharing and coordinated planning are essential in guiding
overland flow, tile drains) (Tesoriero et al., 2009). Of note is the national policies and implementation toward self-sufficiency or inter­
leaching process carries not only nutrients but also salts into drainage national cooperation.
channels, eventually into rivers (Kulmatov et al., 2015b).
The ADR and SDR are heavily burdened with high salinity and 4. Conclusions
excessive nutrient inputs. Among various sources of N and P, more ef­
forts should be made on reducing or removing diffuse sources, especially This study provides direct evidence of large-scale spatial distribution
for N from leaching and P from surface runoff. Due to the expansion of of salinity and nutrients in the Amu Darya and Syr Darya Rivers. Few
agricultural areas and the requirement of food yield increment per area studies have explicitly presented longitudinal variations of water quality
related to inadequate N fertilizer application management, the total at a large spatial scale in these two major rivers of Central Asia, and none

7
P. Leng et al. Environmental Pollution 284 (2021) 117405

have explored the combination between water quality survey and Appendix A. Supplementary data
nutrient modeling in the Amu Darya and Syr Darya Rivers (ADR and
SDR) based on direct in-situ data collection. Here we show that the Supplementary data to this article can be found online at https://doi.
longitudinal increase in salinity in both rivers was indicative of the loss org/10.1016/j.envpol.2021.117405.
of nutrients from irrigated croplands, which was ascribed to limited
rainfall and the low drainage density of the catchment in the region, References
resulting in shifted hydrochemical compositions and irrigation unsuit­
ability of surface water. Using direct causal relationship between Ahearn, D.S., Sheibley, R.W., Dahlgren, R.A., Anderson, M., Johnson, J., Tate, K.W.,
2005. Land use and land cover influence on water quality in the last free-flowing
nutrient loading estimates from IMAGE model and observed riverine river draining the western Sierra Nevada, California. J. Hydrol. 313, 234–247.
nutrient concentrations, we are able to conclude that variations of nu­ Arheimer, B., Lidén, R., 2000. Nitrogen and phosphorus concentrations from agricultural
trients were predominantly driven by diffuse sources of nutrients, catchments—influence of spatial and temporal variables. J. Hydrol. 227, 140–159.
Batbayar, G., Pfeiffer, M., Kappas, M., Karthe, D., 2019. Development and application of
spatially related to irrigated croplands and urban areas. Among this, GIS-based assessment of land-use impacts on water quality: a case study of the
leaching displayed a strong correlation with river NO−3 concentration in Kharaa River Basin. Ambio 48, 1154–1168.
the SDR. Undeniably, the hydrologic connectivity of the drainage basin Bekturganov, Z., Tussupova, K., Berndtsson, R., Sharapatova, N., Aryngazin, K.,
Zhanasova, M., 2016. Water related health problems in Central Asia—a review.
strongly influences the river water quality. Water 8, 219.
Our findings provide a snapshot of river water quality for the ADR Beusen, A.H.W., Bouwman, A.F., Van Beek, L.P.H., Mogollón, J.M., Middelburg, J.J.,
and SDR in Central Asia. Given the contributions of nutrient budgets to 2016. Global riverine N and P transport to ocean increased during the 20th century
despite increased retention along the aquatic continuum. Biogeosciences 13,
riverine concentrations, controlling N pollution in rivers calls for more
2441–2451.
efforts on alleviating N leaching through wise agriculture fertilization Beusen, A.H.W., Van Beek, L.P.H., Bouwman, A.F., Mogollón, J.M., Middelburg, J.J.,
and irrigation practices. Strategies towards reducing P from surface 2015. Coupling global models for hydrology and nutrient loading to simulate
runoff, such as conservation tillage and expansion of riparian bound­ nitrogen and phosphorus retention in surface water–description of IMAGE–GNM and
analysis of performance. Geosci. Model Dev. (GMD) 8, 4045–4067.
aries, are recommended as effective measures for mitigating riverine P. Bezborodov, G., Shadmanov, D., Mirhashimov, R., Yuldashev, T., Qureshi, A.S.,
The issue in this study is probably universal in most dryland catchments Noble, A., et al., 2010. Mulching and water quality effects on soil salinity and
where agricultural activities heavily rely on surface water irrigation. sodicity dynamics and cotton productivity in Central Asia. Agric. Ecosyst. Environ.
138, 95–102.
Further studies could concentrate on the quantification of the relative Borisova, E.A., 2012. Features of the water crisis in central Asia. History and Modern
importance of different sources as a prerequisite for ecological restora­ Times 1.
tion efforts. Therefore, it provides a good tool to evaluate the effects Bostanmaneshrad, F., Partani, S., Noori, R., Nachtnebel, H.-P., Berndtsson, R.,
Adamowski, J.F., 2018. Relationship between water quality and macro-scale
when developing plans for water quality improvements in the dryland parameters (land use, erosion, geology, and population density) in the Siminehrood
catchments. Further detailed investigations of salinity and nutrient River Basin. Sci. Total Environ. 639, 1588–1600.
temporal dynamics with updated nutrient budgets are required to obtain Bouwman, A.F., Beusen, A.H.W., Billen, G., 2009. Human alteration of the global
nitrogen and phosphorus soil balances for the period 1970–2050. Global
a more holistic understanding of river contamination in these arid river Biogeochem. Cycles 23.
systems, especially within the context of the high sensitivity of this re­ Bouwman, A.F., Beusen, A.H.W., Lassaletta, L., van Apeldoorn, D.F., van Grinsven, H.J.
gion to climate change. M., Zhang, J., et al., 2017. Lessons from temporal and spatial patterns in global use
of N and P fertilizer on cropland. Sci. Rep. 7, 40366.
Bu, H., Liu, W., Song, X., Zhang, Q., 2016. Quantitative impacts of population on river
Credit author statement water quality in the Jinshui River basin of the South Qinling Mts., China.
Environmental Earth Sciences 75, 292.
Peifang Leng: Conceptualization, Methodology, Software, Writing – Cañedo-Argüelles, M., Kefford, B., Schäfer, R., 2018. Salt in freshwaters: causes, effects
and prospects - introduction to the theme issue. Phil. Trans. Roy. Soc. Lond. B Biol.
original draft, Writing – review & editing; Qiuying Zhang: Resources, Sci. 374, 20180002.
Supervision, Writing – review & editing; Fadong Li: Conceptualization, Carpenter, S.R., Mooney, H.A., Agard, J., Capistrano, D., DeFries, R.S., Díaz, S., et al.,
Supervision, Writing – review & editing, Funding acquisition; Rashid 2009. Science for managing ecosystem services: beyond the Millennium Ecosystem
Assessment. Proc. Natl. Acad. Sci. Unit. States Am. 106, 1305–1312.
Kulmatov: Writing – review & editing; Guoqin Wang: Data curation, Chen, X., Abudulla, S., Jilili, A., 2017. Present Situation and Utilization of Soil in
Resources; Yunfeng Qiao: Formal analysis, Data curation; Jianqi Wang: Kazakhstan. China Environment Press, Beijing, China.
Visualization, Investigation, Formal analysis; Yu Peng: Visualization, Crosa, G., Froebrich, J., Nikolayenko, V., Stefani, F., Galli, P., Calamari, D., 2006. Spatial
and seasonal variations in the water quality of the Amu Darya river (central Asia).
Investigation, Formal analysis; Chao Tian: Formal analysis, Data cura­ Water Res. 40, 2237–2245.
tion; Nong Zhu: Formal analysis, Investigation; Hubert Hirwa: Writing – FAO (Food and Agriculture Organization), 2012. AQUASTAT Transboundary River Basin
review & editing; Sayidjakhon Khasanov: Writing – review & editing. Overview –Aral Sea. Italy, Rome.
Funakawa, S., Kosaki, T., 2007. Potential risk of soil salinization in different regions of
Central Asia with special reference to salt reserves in deep layers of soils. Soil Sci.
Declaration of competing interest Plant Nutr. 53, 634–649.
Gerlitz, L., Vorogushyn, S., Gafurov, A., 2020. Climate informed seasonal forecast of
water availability in Central Asia: state-of-the-art and decision making context.
The authors declare that they have no known competing financial
Water Security 10, 100061.
interests or personal relationships that could have appeared to influence Groll, M., Opp, C., Kulmatov, R., Ikramova, M., Normatov, I., 2015a. Water quality,
the work reported in this paper. potential conflicts and solutions-an upstream-downstream analysis of the
transnational Zarafshan River (Tajikistan, Uzbekistan). Environmental Earth
Sciences 73, 743–763.
Acknowledgement Groll, M., Opp, C., Kulmatov, R., Sun, Z., Normatov, I., Bernardi, A., et al., 2015b.
Managing Central Asia’s transboundary rivers: case studies of the Zarafshan
This study was supported by the Strategic Priority Research Program (Tajikistan/Uzbekistan) and Tarim (Kyrgyzstan/China) rivers. River basin
management VIII. WIT transactions on ecology 196, 149–160.
of Chinese Academy of Sciences (No. XDA20040302), the National Grömping, U., 2006. Relative importance for linear regression in R: the package
Natural Science Foundation of China (No. U1803244 and No. relaimpo. J. Stat. Software 17, 1–27.
41761144053), Shihezi Science and Technology Project (No. 2019ZH13 Haag, I., Jones, P.D., Samimi, C., 2019. Central asia’s changing climate: how temperature
and precipitation have changed across time, space, and altitude. Climate 7, 123.
and 2021BC003), and. P. Leng is supported by the CAS- DAAD Joint Israël, N.M.D., VanLandeghem, M.M., Denny, S., Ingle, J., Patiño, R., 2014. Golden alga
Fellowship Programme for Doctoral Students. Thanks to colleagues at presence and abundance are inversely related to salinity in a high-salinity river
the Helmholtz Center for Environmental Research-UFZ for valuable ecosystem, Pecos River, USA. Harmful Algae 39, 81–91.
Iwata, T., Ikeda, Y., Uzawa, R., Ikarashi, Y.J.L., 2020. Lagrangian tracking measurements
discussions. The datasets used and analyzed in current study are avail­ revealed the temporal dynamics of nitrogen and phosphorus spiralling in a large
able from the corresponding author on reasonable request. Japanese river. Limnology 21, 1–13.

8
P. Leng et al. Environmental Pollution 284 (2021) 117405

Janssen, A.B.G., Janse, J.H., Beusen, A.H.W., Chang, M., Harrison, J.A., Huttunen, I., Qadir, M., Noble, A.D., Qureshi, A.S., Gupta, R.K., Yuldashev, T., Karimov, A., 2009. Salt-
et al., 2019. How to model algal blooms in any lake on earth. Current Opinion in induced land and water degradation in the Aral Sea basin: a challenge to sustainable
Environmental Sustainability 36, 1–10. agriculture in Central Asia. Nat. Resour. Forum 33, 134–149.
Jarsjö, J., Törnqvist, R., Su, Y., 2017. Climate-driven change of nitrogen QGIS Development Team, 2019. QGIS Geographic Information System. Open Source
retention–attenuation near irrigated fields: multi-model projections for Central Asia. Geospatial Foundation Project.
Environmental Earth Sciences 76, 117. R Core Team. R, 2019. A Language and Environment for Statistical Computing. R
Jones, E., van Vliet, M.T.H., 2018. Drought impacts on river salinity in the southern US: Foundation for Statistical Computing, Vienna, Austria.
implications for water scarcity. Sci. Total Environ. 644, 844–853. Rasilo, T., Hutchins, R.H.S., Ruiz-González, C., del Giorgio, P.A., 2017. Transport and
Karthe, D., Abdullaev, I., Boldgiv, B., Borchardt, D., Chalov, S., Jarsjö, J., et al., 2017. transformation of soil-derived CO2, CH4 and DOC sustain CO2 supersaturation in
Water in Central Asia: an integrated assessment for science-based management. small boreal streams. Sci. Total Environ. 579, 902–912.
Environmental Earth Sciences 76, 690. Reyer, C.P.O., Otto, I.M., Adams, S., Albrecht, T., Baarsch, F., Cartsburg, M., et al., 2017.
Kelley, C.J., Keller, C.K., Brooks, E.S., Smith, J.L., Huyck Orr, C., Evans, R.D., 2017. Climate change impacts in Central Asia and their implications for development. Reg.
Water and nitrogen movement through a semiarid dryland agricultural catchment: Environ. Change 17, 1639–1650.
seasonal and decadal trends. Hydrol. Process. 31, 1889–1899. Rzymski, P., Klimaszyk, P., Niedzielski, P., Marszelewski, W., Borowiak, D., Nowiński, K.,
Kienzler, K.M., Djanibekov, N., Lamers, J.P., 2011. An agronomic, economic and et al., 2019. Pollution with trace elements and rare-earth metals in the lower course
behavioral analysis of N application to cotton and wheat in post-Soviet Uzbekistan. of Syr Darya river and small Aral Sea, Kazakhstan. Chemosphere 234, 81–88.
Agric. Syst. 104, 411–418. Saidmamatov, O., Rudenko, I., Pfister, S., Koziel, J., 2020. Water–energy–food nexus
Kulmatov, R., Mirzaev, J., Abuduwaili, J., Karimov, B., 2020. Challenges for the framework for promoting regional integration in central Asia. Water 12, 1896.
sustainable use of water and land resources under a changing climate and increasing Saiko, T.A., Zonn, I.S., 2000. Irrigation expansion and dynamics of desertification in the
salinization in the Jizzakh irrigation zone of Uzbekistan. Journal of Arid Land 12, Circum-Aral region of Central Asia. Appl. Geogr. 20, 349–367.
90–103. Sehring, J., 2020. Unequal distribution: academic knowledge production on water
Kulmatov, R., Rasulov, A., Kulmatova, D., Rozilhodjaev, B., Groll, M., 2015a. The governance in Central Asia. Water Security 9, 100057.
modern problems of sustainable use and management of irrigated lands on the Seim, A., Schultz, J.A., Leland, C., Davi, N., Byambasuren, O., Liang, E., et al., 2017.
example of the bukhara region (Uzbekistan). J. Water Resour. Protect. 7, 956–971. Synoptic-scale circulation patterns during summer derived from tree rings in mid-
Kulmatov, R., Rasulov, A., Kulmatova, D., Rozilhodjaev, B., Groll, M., 2015b. The latitude Asia. Clim. Dynam. 49, 1917–1931.
modern problems of sustainable use and management of irrigated lands on the Shibuo, Y., Jarsjö, J., Destouni, G., 2007. Hydrological responses to climate change and
example of the bukhara region (Uzbekistan). Coll. Compos. Commun. 7, 956–971. irrigation in the Aral Sea drainage basin. Geophys. Res. Lett. 34.
Law, N., Band, L., Grove, M., 2004. Nitrogen input from residential lawn care practices in Smith, D.R., Harmel, R.D., Williams, M., Haney, R., King, K.W., 2016. Managing acute
suburban watersheds in Baltimore county, MD. J. Environ. Plann. Manag. 47, phosphorus loss with fertilizer source and placement: proof of concept. Agricultural
737–755. & Environmental Letters 1, 150015.
Li, X.-D., Liu, C.-Q., Harue, M., Li, S.-L., Liu, X.-L., 2010. The use of environmental Suárez, J., Puertas, J., 2005. Determination of COD, BOD, and suspended solids loads
isotopic (C, Sr, S) and hydrochemical tracers to characterize anthropogenic effects on during combined sewer overflow (CSO) events in some combined catchments in
karst groundwater quality: a case study of the Shuicheng Basin, SW China. Appl. Spain. Ecol. Eng. 24, 199–217.
Geochem. 25, 1924–1936. Tesoriero, A.J., Duff, J.H., Wolock, D.M., Spahr, N.E., Almendinger, J.E., 2009.
Lintern, A., Webb, J., Ryu, D., Liu, S., Bende-Michl, U., Waters, D., et al., 2018. Key Identifying pathways and processes affecting nitrate and orthophosphate inputs to
factors influencing differences in stream water quality across space. Wiley streams in agricultural watersheds. J. Environ. Qual. 38, 1892–1900.
Interdisciplinary Reviews: Water 5, e1260. Törnqvist, R., Jarsjö, J., Thorslund, J., Rao, P.S.C., Basu, N.B., Destouni, G., 2015.
Lioubimtseva, E., Cole, R., Adams, J.M., Kapustin, G., 2005. Impacts of climate and land- Mechanisms of basin-scale nitrogen load reductions under intensified irrigated
cover changes in arid lands of Central Asia. J. Arid Environ. 62, 285–308. agriculture. PloS One 10, e0120015.
Liu, S., Xie, Z., Zeng, Y., Liu, B., Li, R., Wang, Y., et al., 2019. Effects of anthropogenic UNHCR (UN, 2017. High Commissioner for Refugees). The Sustainable Development
nitrogen discharge on dissolved inorganic nitrogen transport in global rivers. Global Goals and Addressing Statelessness.
Change Biol. 25, 1493–1513. UNEP (UN Environment Programme), 2011. Environment and Security in the Amu Darya
Liu, X., Beusen, A.H.W., Van Beek, L.P.H., Mogollón, J.M., Ran, X., Bouwman, A.F., 2018. Basin. United Nations Environment Programme, Nairobi, Kenya.
Exploring spatiotemporal changes of the yangtze river (changjiang) nitrogen and Unger-Shayesteh, K., Vorogushyn, S., Farinotti, D., Gafurov, A., Duethmann, D.,
phosphorus sources, retention and export to the east China Sea and yellow Sea. Mandychev, A., et al., 2013. What do we know about past changes in the water cycle
Water Res. 142, 246–255. of Central Asian headwaters? A review. Global Planet. Change 110, 4–25.
Ma, L., Abuduwaili, J., Li, Y., Uulu, S.A., Mu, S., 2019. Hydrochemical characteristics and van Beek, L.P.H., Wada, Y., Bierkens, M.F.P., 2011. Global Monthly Water Stress: 1.
water quality assessment for the upper reaches of Syr Darya river in Aral Sea basin, Water Balance and Water Availability, vol. 47. Water Resources Research.
central Asia. Water 11, 1893. Varanka, S., Hjort, J., Luoto, M., 2015. Geomorphological factors predict water quality in
Min, J., Zhang, H., Shi, W., 2012. Optimizing nitrogen input to reduce nitrate leaching boreal rivers. Earth Surf. Process. Landforms 40, 1989–1999.
loss in greenhouse vegetable production. Agric. Water Manag. 111, 53–59. Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D., Prusevich, A., Green, P.,
Morris, B.L., Darling, W.G., Gooddy, D.C., Litvak, R.G., Neumann, I., Nemaltseva, E.J., et al., 2010. Global threats to human water security and river biodiversity. Nature
et al., 2006. Assessing the extent of induced leakage to an urban aquifer using 467, 555–561.
environmental tracers: an example from Bishkek, capital of Kyrgyzstan, Central Asia. Williams, W.D., 1987. Salinization of rivers and streams. An important environmental
Hydrogeol. J. 14, 225–243. hazard. Ambio 16, 180–185.
Newbold, J., Elwood, J., O’neill, R., Sheldon, A.L., 1983. Phosphorus dynamics in a Withers, P.J.A., Jarvie, H.P., 2008. Delivery and cycling of phosphorus in rivers: a
woodland stream ecosystem: a study of nutrient spiralling. Ecology 64, 1249–1265. review. Sci. Total Environ. 400, 379–395.
Newbold, J.D., Elwood, J.W., O’Neill, R.V., Winkle, W.V., 1981. Measuring nutrient Worsfold, P., McKelvie, I., Monbet, P., 2016. Determination of phosphorus in natural
spiralling in streams. Can. J. Fish. Aquat. Sci. 38, 860–863. waters: a historical review. Anal. Chim. Acta 918, 8–20.
O’Hara, S.L., 2000. Lessons from the past: water management in Central Asia. Water Pol. Wurtsbaugh, W.A., Miller, C., Null, S.E., DeRose, R.J., Wilcock, P., Hahnenberger, M.,
2, 365–384. et al., 2017. Decline of the world’s saline lakes. Nat. Geosci. 10, 816–821.
Piper, A.M., 1944. A graphic procedure in the geochemical interpretation of water- Xenarios, S., Schmidt-Vogt, D., Qadir, M., Janusz-Pawletta, B., Abdullaev, I., 2019. The
analyses. Eos. Trans. Am. Geophys. Union 25, 914–928. Aral Sea Basin: Water for Sustainable Development in Central Asia. Routledge.
Pratt, B., Chang, H., 2012. Effects of land cover, topography, and built structure on Zhang, J., Beusen, A.H.W., Van Apeldoorn, D.F., Mogollón, J.M., Yu, C., Bouwman, A.F.,
seasonal water quality at multiple spatial scales. J. Hazard Mater. 209–210, 48–58. 2017. Spatiotemporal dynamics of soil phosphorus and crop uptake in global
cropland during the 20th century. Biogeosciences 14, 2055–2068.