Spatial distribution characteristics and pollution levels of heavy metals in surface water and sediments of the Heihe cascade reservoir system, China

Global river systems have been increasingly fragmented and impacted by dam construction, sand mining and water diversion for water and energy needs (Nilsson et al. 2005; Vukovic et al. 2014; Bing et al. 2016), and one of the most important activities is the construction of dams. It has been reported that more than 45,000 large dams have been constructed throughout the world, of which China ranks first, with 22,265 dams and a percentage of 44.80% (He et al. 2006). These dams have unparalleled benefits worldwide in terms of flood control, irrigation, shipping, water supply, and power generation and adjust the uneven spatiotemporal distribution of water resources (Zhai et al. 2010; Li et al. 2013). Nevertheless, such large artificial lakes can have significant negative effects on a river's hydrodynamic conditions, including significantly intercepting natural water delivery (Wang et al. 2018) and changing the hydrological characteristics (Fremion et al. 2016) and hydrodynamic conditions (Huang et al. 2019). Additionally, human activities cause a large number of heavy metals to enter rivers; these heavy metals undergo adsorption, sedimentation, flocculation and complex formation through adsorption to inorganic substances, which are ultimately deposited in the reservoir area (Bednarova et al. 2013; Brady et al. 2016).

Sediments are the most important source and sink for the accumulation and redistribution of heavy metals (Steiger et al. 2001; Miao et al. 2020; Wang et al. 2020). A study revealed that reservoir sediments usually accumulate at a rate of ≥2 cm/yr (Audry et al. 2004a), causing reservoirs globally to shrink by approximately 0.5% to 1.0% of their storage capacity each year (Kummu et al. 2010). These accumulated sediments physically reduce the water storage capacity, cause ecological and environmental problems, and gradually reduce the comprehensive benefits of the reservoir (Palazon & Navas 2014; Tang et al. 2018). Furthermore, changes in conditions under certain circumstances, such as bioturbation (Zoumis et al. 2001), resuspension caused by dredging (Audry et al. 2004b), flooding, water temperature stratification and water level fluctuations caused by regular impoundment, cause changes in the pH, dissolved oxygen and redox conditions of the river water-sediment interface (Martin & Pedersen 2002; Hahn et al. 2018). Heavy metals in sediments are rereleased into the water, causing ‘secondary pollution’, which further affects ecosystems and human health through the food chain and biological enrichment (Jafarabadi et al. 2017; Bing et al. 2019). Therefore, it is still urgent to evaluate the distribution mechanism, source analysis and risk level of heavy metals in water and sediments in reservoir areas to formulate strategies and methods for controlling heavy metal pollution in rivers.

Water and sediment pollution with heavy metals is a worldwide problem (Zahra et al. 2014), and heavy metals are considered to be one of the most threatening pollutants in aquatic ecosystems because of their high enrichment, high toxicity, persistence, nonbiodegradable nature and high bioaccumulation potential (Yan et al. 2016; Lu & Yu 2018). Generally, heavy metals can enter aquatic environments from natural sources (e.g., weathering of soil and rocks) and anthropogenic sources (e.g., atmospheric deposition, industrial wastewater, mining and mineral processing), which greatly promote the geochemical behavior and accumulation of heavy metals worldwide (Varol 2011; Mazurek et al. 2017; Saleem et al. 2018; Ali et al. 2019).

Several studies on heavy metals in water and sediments in reservoirs have focused on Asia, including China (Tang et al. 2018; Bing et al. 2019), Europe, including Israel (Hahn et al. 2018), and South America, including Brazil (Souza & Wasserman 2015). For example, the average concentration of each heavy metal in the water level fluctuation zone (WLFZ) of the Nuozhadu mega-reservoir in the upper reaches of the Mekong River is higher than that in the infralittoral reference zone (IRZ), and the Cu, Mn, and Ni concentrations increase with elevation, while the Cr and Zn concentrations decrease with elevation (Song et al. 2019). Meanwhile, the construction of the Mekong dams resulted in slightly lower heavy metal concentrations in the main stream at high water levels than at low water levels and slightly higher health risks for heavy metals in the main stream than in the tributaries, especially between the predam and reservoir areas, with significant differences in chromium concentrations and no significant differences in the concentrations of other heavy metals (Wang et al. 2012; Zhao et al. 2020). In addition, compared with single dams, the interception effect and ecological impact of step dams are more complex and systematic, leading to significant changes in water levels, backwaters and physicochemical properties in different river sections, while the complexity of the environment in different regions makes the redistribution pathways and rates of heavy metal content in reservoir sediments along the flow direction different (Fang & Deng 2011; Smith et al. 2014; Cardoso-Silva et al. 2017). To date, only a few studies have reported the effect of cascade dam construction on the geochemical characteristics of heavy metals in the water or sediment of an entire river system, while the risks and influencing factors of heavy metal pollution in inland river cascade reservoirs in arid and semiarid areas remain unknown. The inland rivers originating from the Tibetan Plateau in China have relatively large slopes, shallow water depths, and fast flow rates. These rivers are typical ecologically fragile areas and are sensitive to environmental changes, and with the implementation of the ‘Silk Road’ revival initiative, the strategic position of inland river basins in the resource and economic development in western China will become increasingly prominent (Yan et al. 2019). However, under the background of global warming and water shortages, the ecological degradation resulting from heavy metal pollution in rivers caused by cascade dams will become more severe (Kummu & Varis 2007). As such, accurate analysis of the distribution mechanism and potential risks of heavy metals in water and sediments in cascade reservoir areas is of great significance to ensure the health of reservoirs and rivers in inland river basins.

The HRB is located in northwestern China, within the range of 96°42′–102°04′ E, 37°45′–42°40′N (Figure 1). The core drainage area is approximately 143,000 km² with a length of 821 km; it originates in the Qilian Mountains, flows through the Hexi Corridor of Gansu Province and enters the western part of the Inner Mongolia Plateau (Qin et al. 2010), forming a globally unique ‘ice river oasis desert lake’ horizontal eco-landscape (Wu et al. 2017). The elevation range of the upper reaches of the Heihe River is 2,600–4,850 m a.s.l., and the main tributaries include the Yeniugou River and Babao River. The upper reach area has a mean annual temperature of −3 to 4 °C, the annual average precipitation exceeds 350 mm, and the vegetation is very sparse, mainly alpine meadow. The elevation of the middle reach area decreases from 2,600 m to 1,350 m a.s.l., the annual average precipitation decreases from 250 mm to < 100 mm (Li et al. 2019), the annual average temperature is approximately 3–7 °C, and the average potential evaporation rate is approximately 1,400 mm, which is a typical irrigated agricultural belt. The planning area of coal, copper, lead and zinc in the upper reaches of Qilian County and Tianjun County is one of the 22 key planning areas for mineral resource exploration delineated on the Tibetan Plateau; the area contains heavy metal elements such as iron, chromium, lead, zinc and copper, and nonferrous metals are also found in the middle reaches of the Liyuan River. The water energy reserves in the upper reaches of the Heihe River are 106.88 × 10⁴ kW, with an exploitable capacity of 52.8 × 10⁴ kW and an annual electricity output of 38.48 × 10⁸ kW/h. To meet the needs of flood control and power generation, there are a series of cascade reservoirs and water diversion projects (e.g., Dipanzi, Huangzangsi, Baoping, Sandaowan, Erlongshan, Dagushan, Xiaogushan, Longshou II, Longshou I, Longqu, Maweihu, Houtouhu, Shihuiguan, Bailang, Dahexia, Chulong and Yinggezui).

Thirty-one sampling sites were selected for field surveys in the Heihe River in the summer of 2019 and 2020 (July–August). Six sampling points (T1, T2, T4, T5, T12 and T13) in the tributaries and eight sampling points (M1, M2, M10, M11, M14 ∼ M17) in the main stream were used as reference points for the natural river sections to reveal the contents of heavy metals in the river under natural conditions. Five samples (T3, T6, T8, T10, T11) were located at the reservoir center of the tributary. Seven samples (M3 ∼ M7, M12 and M13) were located at the reservoir center of the main stream. Three samples (T7, T9, T14) were located at the reservoir tail of the tributary. Two samples (M8 and M9) were located at the reservoir tail of the main stream. At each location, surface sediments were collected at a depth of 0–5 cm at the left, middle and right sides of the cross-section. The samples were evenly mixed in the field, packed in clean polyethylene bags, marked, and stored at 4 °C for transport to the laboratory. Then, the samples were freeze-dried, impurities were removed, and the samples were ground with a mortar and passed through a 100 mesh nylon sieve. A total of 4 g of each sample was digested with the HNO₃-HF-HClO₄ method. The heavy metal elements copper (Cu), nickel (Ni), lead (Pb), zinc (Zn), cadmium (Cd), arsenic (As) and manganese (Mn) were determined by inductively coupled mass spectrometry (ICP‒MS). The relative standard deviations of replicates were within ± 5% of the mean.

The specific method of surface water sample collection was in accordance with the Standard Methods for the Examination of Water and Wastewater. The collected surface water samples were stored at 4 °C, and determination was completed within 1 week. The concentrations of Mn, Ni, Pb, Zn, As, Cr, Cd, and Cu were measured using ICP‒MS (5300DV, USA).

Analysis of variance (ANOVA) (Fisher test, p < 0.05) was used to distinguish the spatial differences in heavy metal concentrations in surface water and sediments. Principal component analysis (PCA) is commonly used to screen multiple variables in a complex system (Bai et al. 2016), and in this study, it was applied to identify the sources of heavy metals in the sediments based on the relationships among variables with different geochemical properties. The suitability of the data for PCA was determined using Kaiser‒Meyer‒Olkin (KMO) and Bartlett's sphericity tests. Factor analysis (FA) was applied to identify the sources of heavy metals, combined with multiple linear regression (MLR) to quantify the contribution of each source of heavy metals, which was named FA-MLR for supplying quantitative information about the contributions of each source type (Zhou et al. 2007). Experimental data processing and statistical analysis were carried out by using OriginPro (version 9.0, USA), IBM SPSS Statistics (version 24.0, USA) and ArcGIS (version 10.6, USA).

As shown in Table 2, the average concentrations of noncarcinogenic heavy metal elements followed the order Mn (146.50 ⁠) > Zn (25.2 ⁠) > Cu (6.70 ⁠) > Pb (4.50 ⁠), and those of carcinogenic heavy metals followed the order Ni (9.50 ⁠) > As (2.50 ⁠) > Cr (2.00 ⁠) > Cd (1.10 ⁠). The average concentrations of the eight heavy metals were much lower than the criteria for drinking water in China (GB5749-2006), except for Mn in the surface water of the Heihe River. Moreover, the mean concentrations of these heavy metals were also much lower than the drinking water standards of the World Health Organization (WHO) and the third-class surface water environmental quality standards (GB 3838-2002).

The average concentrations of As, Cd, Cr, Cu, Mn, Ni, Pb and Zn in surface sediments were 13.57, 0.23, 103.17, 38.73, 791.06, 49.65, 24.17 and 94.60 ⁠, respectively. The average concentrations of all heavy metals were far greater than the environmental background values of Gansu Province, and the average concentrations of As, Cd, Cr, Cu, Mn, Ni, Pb and Zn were 1.80, 2.64, 2.10, 2.07, 1.41, 1.74, 1.68, and 1.41 times the soil background values in Gansu Province, respectively. However, different results were obtained relative to the soil background values of the Tibetan Plateau; for example, the average concentrations of As, Cr, Ni and Pb were lower than those of the Tibetan Plateau.

One-way ANOVA results showed that the concentrations of As, Cd, Cr, Cu, Mn, Ni and Zn in the surface water of the natural section of the main stream and tributaries were essentially not different (P > 0.05) (Figure 2(b)). There were significant differences in the concentrations of Cd, Cr and Pb (P < 0.05) in the surface water of the main stream and tributary reservoirs (Figure 2(c)) and in Cd, Cu, Mn, Ni and Zn (P < 0.05) at the reservoir tail (Figure 2(d)). The contributions of heavy metals were different between the main stream and tributaries. Among them, in the reservoir center, Cd and Pb had high values in the main stream, and the values of Cr were almost the same in the tributaries and main stream, indicating that the main stream contributed more than the tributaries to Cd and Pb and that the tributaries and main stream contributed equally to Cr. In the reservoir tail, Cd had high values in the tributaries, and Cu, Mn, Ni and Zn had high values in the main stream, indicating that the tributaries contributed more to Cd, while the main stream contributed to Cu, Mn, Ni and Zn.

One-way ANOVA results showed that there were significant differences in the concentrations of Cd, Cr, Ni and Zn (P < 0.05) in the surface sediments of the main and tributary reservoirs of the natural river section (Figure 3(b)). There were significant differences in the concentrations of Cr and Ni (P < 0.05) in the surface sediments of the main and tributary reservoirs (Figure 3(c)) and in Cd (P < 0.05) at the reservoir tail (Figure 3(d)). The results show that the contributions of the main stream and tributaries to heavy metals were different. In the natural reaches, Cr, Ni and Zn had high values in the tributaries, and the values of Cd were almost the same in the tributaries and main stream, indicating that tributaries contributed more to Cr, Ni and Zn, while the tributaries and main stream had the same contribution to Cd. In the reservoir center, Cr and Ni had high values in the main stream, indicating that the main stream contributed more to Cr and Ni. In the reservoir tail, the values of Cd were almost the same in the tributaries and main stream, and the tributaries and main stream had the same contribution to Cd.

The samples used for PCA all passed the Kaiser‒Meyer‒Olkin (KMO > 0.78) and Bartlett (p < 0.001) tests. In this study, three principal components (PCs) with eigenvalues greater than 1 explained 86.27% and 87.46% of the total variance in surface water data in the tributaries and main stream, respectively (Table 3). In the tributaries, PC1 explained 53.65% of the total variance and was dominated by Cu, Mn, Ni, Pb and Zn, which probably had the same or similar complex sources. PC2 accounted for 23.48% of the total variance and mainly comprised As. PC3 accounted for 22.40% of the total variance and was dominated by Cd and Cr. In the main stream, the percentage contribution of PC1 to the total variance was 41.34%, and PC1 mainly comprised Cr, Cu and Mn, indicating the same or similar sources. PC2 accounted for 28.28% of the total variance and was dominated by Ni, Pb and Zn. PC3 accounted for 17.85% of the total variance and was dominated by As and Cd.

The samples used for PCA all passed the KMO (KMO > 0.78) and Bartlett (p < 0.001) tests. In this study, three PCs with eigenvalues greater than 1 explained 93.12% and 81.41% of the total variance in surface sediment data in the tributaries and main stream, respectively (Table 4). For the tributaries, PC1 explained 41.27% of the total variance and was dominated by As, Cr, Cu and Ni, indicating the same source of contamination or similar geochemical behavior. PC2 accounted for 33.10% of the total variance and mainly comprised Cd, Pb and Zn. PC3 accounted for 22.057% of the total variance and had strong negative loadings (>0.90) for Mn. In light of the results in Section 3.1, Mn concentrations were lower in the surface water than in the sediments and did not show significant spatial differences; thus, Mn was mainly from natural sources. In contrast, the rest of the elements (e.g., Cr, Cu, Zn, Ni and As) were likely to come from anthropogenic sources. In the main stream, the percentage contribution of PC1 to the total variance was 28.14%, and PC1 mainly comprised Cr and Ni. PC2 accounted for 27.82% of the total variance and was dominated by Cd, Pb and Zn. PC3 accounted for 25.46% of the total variance and was dominated by As, Cu and Mn. Combined with the ANOVA results in Sections 3.1 and 3.2, it is highly likely that these heavy metals in the main stream came from human activities.

After identifying the possible sources of heavy metals using factor analysis, FA-MLR was used to quantify the contribution of each source (Tong 2012). As shown in Table 5, the correlation coefficients between the estimated and measured values of FA-MLR all reached highly significant levels, indicating that FA-MLR can well identify pollution sources and quantify their contributions. In the tributaries, 60.14% of As originated from mining activities, with industrial pollution contributing 17.55% and parent material and rock weathering contributing 12.15%. The contributions of Cd, Cr and Ni on the three axes were not very different, and the main sources of Cu were mining activities and industrial activities, which contributed 70.82% and 12.52%, respectively. A total of 85.84% of Mn originated from rock weathering; 71.82% of Pb came from industrial contamination; and industrial contamination provided 60.29% of Zn, while mining activities and rock weathering contributed 20.76% and 12.98%, respectively. In the main stream, 59.30% of As came from agricultural activities, and mining activities contributed 28.24%; the main sources of Cd were industrial and agricultural activities, which contributed 65.08% and 21.05%, respectively; 53.43% and 57.71% of Cr and Ni, respectively, originated from mining activities; 67.55% and 49.54% of Cu and Mn, respectively, came from agricultural activities; and 78.63% and 85.85% of Pb and Zn, respectively, originated from industrial activities. However, mining, industrial activities, agriculture and rock weathering cannot fully explain all of the sources of heavy metals, indicating the existence of other factors that also contribute to the distribution and accumulation of heavy metals, although the contribution of other factors to heavy metals was below 16% (0.52%–15.39%).

The contribution of heavy metals in the surface water of the Heihe River basin is different between the main stream and the tributaries, which is mainly due to the confluence of two upstream tributaries into the main stream at Huangzangsi coupled with the confluence of the Liyuan River tributaries into the main stream in the middle reaches and the interference of the cascade dams and human productive activities, which makes the contribution of heavy metals in the main stream greater than that in the tributaries. However, the ANOVA showed that the operation of cascade dams has a significant effect on the spatial distribution of heavy metals in surface water, and it can be seen that heavy metals in surface water are mainly affected by tributaries as well as industrial and agricultural production. In addition, according to Bao (2018), the heavy metal concentrations in the surface water of the Heihe River were much lower than those in this study, especially those for Cu, As and Cd, which might be related to the sampling time and the location of the sample point selection.

The contributions of the main stream and tributaries to heavy metals in surface sediments are also different, and the operation of cascade dams has a significant effect on the spatial distribution of heavy metals in surface sediments. This may be caused by the weakening of hydrodynamic action and differing adsorption of fine particles in different reaches of the cascade reservoir. Coupled with the wider scope of this investigation, the distribution density of tailings and differences in human activities make the pollution sources of heavy metals more complex. Combined with the results of principal component analysis, the highest levels of Cr, Cu and Ni were found in the surface sediments of the tributaries. Murat et al. (2019) showed that Cr and Ni accumulate in sediments due to the widespread existence of mafic and ultramafic lithologies. The coal, Cu, Pb and Zn planning areas in the Tibetan Plateau, including Qilian County, Tianjun County and the Liyuan River, are concentrated mineral development areas; although most of the mines closed after 2000, the remaining tailings and leachate have led to the accumulation of Cr, Cu and Ni in the sediment (Bu 2016), while the weathering of rocks also contributes to the accumulation of heavy metals. In addition, the tributaries mainly have plateau and mountainous terrain, mostly characterized by terraced pasture areas, and livestock pollutants may be the cause of high As levels. There are high levels of Cr and Ni in the surface sediments of the main stream. Generally, the contents of Cr and Ni in the soil were relatively low, while certain small mines and smelting activities were distributed along the main stream. Additionally, in light of the results in Section 3.1, Cr and Ni showed significant spatial differences (ANOVA, p < 0.05), with high CF values. Thus, these elements were assumed to originate from mineral mining activities. Cd, Pb and Zn are similarly distributed in the tributaries and main stream. Zhang et al. (2015) reported that Pb was mainly derived from water transportation facilities. However, considering the sparseness of the area and the limited vehicular movement, the likelihood of Pb coming from vehicle exhaust and agricultural pollution is very low (Wei & Yang 2010). Likewise, Cd is widely used in nonferrous metal processing and electroplating (Wang et al. 2006; Zhang et al. 2015). Thus, it is highly likely that Pb and Cd come mainly from heavy wastewater generated by industrial pollution (Wang et al. 2016). The distribution and transport of Zn in sediments depends on the local environment; Zn is difficult to leach from soils because it is readily adsorbed by organic matter (Zahra et al. 2014), Zn has high CF values, and Zn has been reported to be derived from steel smelting and coal combustion (Guo et al. 2011; Ma et al. 2014; Zhao et al. 2020). With high levels of Mn in tributary surface sediments, the mean values all exceeded the soil environmental background values, suggesting that Mn may be derived from natural sources (e.g., parent material and rock weathering) and some slag (Zhao et al. 2014). Regarding the high As, Cu and Mn contents in the main stream surface sediments, considering that the terrain in the main stream is dominated by corridor plains with well-developed oasis agriculture, we determined that Cu, As and Mn mainly originate from agricultural activities in the region. Cu is considered a major landmark pollutant in agricultural production (Wang et al. 2021). Moreover, the application of chemical fertilizers, pesticides and insecticides leads to the accumulation of As and Mn (Bing et al. 2016).

In addition, FA-MLR demonstrated that there were both similarities and differences in heavy metals in the tributaries and main stream of the Heihe River. The main pollutants are primarily derived from industrial activities, mining activities, agricultural activities and natural effects in that order, and there are significant differences in the contribution ratios of individual heavy metals in the tributaries and main stream, which may be related to the geographical distribution. The tributaries are characterized by relatively limited economic and social development, and the exploitation of natural resources, especially mining activities, intensifies the accumulation of heavy metals. In contrast, the main stream is densely populated, and industrial and agricultural wastewater and domestic sewage lead to the large enrichment of heavy metals, while cascade dams intensify the exchange and transport of heavy metals in surface water and sediments, leading to a redistribution of source contribution ratios (Tam & Wong 2000; Xu et al. 2014).

The overall contamination level of heavy metals in surface water is low, while the high pollution at point T13 may be due to the extensive mining of minerals such as Cu, Mn, and Pb and construction materials in the vicinity of the Liyuan River. In addition, the area is located in the Qilian Mountain metallogenic belt, and a large number of mining activities have historically stripped the surface soil and bedrock, accelerating the weathering of the rocks, which is also an important cause of the accumulation of heavy metals in surface water. Several sampling sites in natural streams in the main stream (M16 and M17) and the tail of the reservoir (M8) were less polluted, while the rest were unpolluted. However, compared with tributaries, the pollution levels of heavy metals in the natural reaches, reservoir center and tail of the main stream all showed varying degrees of increase compared to the tributaries, which may be related to the convergence of the tributaries into the main stream or to human activities. Bao (2018) showed that the level of heavy metal pollution in the surface water of the Heihe River was generally unpolluted and showed a gradual increase from upstream to downstream, which is comparable to our findings. Overall, there was no significant human-made contamination or heavy metal accumulation in the water of the reservoir.

Cd in surface sediments is at a heavy pollution and high risk level, and the level of heavy metal pollution in the main stream and tributaries varies greatly, especially in the cascade reservoirs, where it tends to accumulate gradually. Mine effluents and the economic activities associated with mining generate household and industrial wastes that normally find their way into rivers. It is our view that since the Heihe River has a rich history of mining, these wastes will continue posing a threat to the aquatic ecosystem (J. Li et al. 2020; Hasimuna et al. 2021). The highest ecological risk was found in the natural reaches of the tributaries, which may be related to these sites being closer to the mines; activities such as the stripping of topsoil and bedrock by mining activities accelerates the weathering of bedrock and wind rowing. Second, this may also be related to the highest toxic response factor of Cd. In addition, the distribution coefficient of Cd in the solid‒liquid phase was relatively low, which indicated that Cd was easily released from the sediment, resulting in a large accumulation of Cd (Shi et al. 2017; Hasimuna et al. 2021). The main stream has a length of over 280 km and flows through many industrial cities and oasis agricultural areas, and its large scale and diverse pollution sources cause a high ecological risk at some sampling sites (Bing et al. 2016; Wang et al. 2021). The Xiaogushan reservoir center of the main stream had a moderate ecological risk, which may be related to the fact that there were many fine-grained sediments in the reservoir center near the dam. Fine-grained sediments tend to have a high specific surface area, and surface adsorption and ion interactions can cause significant enrichment of heavy metals (Wang et al. 2012). Furthermore, when the dissolved oxygen content in the reservoir center was low, deposited heavy metal elements were released into the surface water in large quantities, causing secondary pollution, which should be given great attention.

This study identified and quantified heavy metal contamination in both surface water and sediments from the natural reaches, center and tail of the reservoir in the Heihe River. The main conclusions are as follows:

• (1)

  The average concentrations of heavy metals in surface water in the Heihe River were much lower than the surface water environmental quality standards established by China and the WHO, while the average concentrations of heavy metals in surface sediments exceeded the background values in Gansu Province. The one-way ANOVA results indicated that the concentrations of Cd and Pb in the reservoir center and Cu, Mn, Pb, Ni and Zn concentrations in the reservoir tail in the main stream were significantly higher than those in the tributaries. The surface sediment concentrations of Cd, Pb, Cr, Zn and Mn were significantly different, and the Cr, Ni and Zn concentrations in the natural reaches in the tributaries were significantly higher than those in the main stream.

• (2)

  The pollution assessment results revealed no heavy metal contamination in the surface water. Cd was the priority pollutant of concern in surface sediments due to its fairly heavy contamination and ecological risk in the natural reaches of the tributaries and reservoir center. Furthermore, contamination by As, Cu, Cr, Pb and Zn in the natural reaches in the tributaries and by Cr, Ni and Pb in the reservoir center in the main stream was observed. In general, heavy metal contamination and ecological risks were low in the Heihe River.

• (3)

  Source identification showed that for the tributaries, Cr, Cu and Ni mainly came from both mining activities and rock weathering; Pb, Cd and Zn came from industrial activities; and Mn came from natural sources. For the main stream, Cr and Ni mainly came from mining activities; Pb, Cd and Zn came from industrial activities; and As, Cu and Mn came from agricultural activities. Furthermore, this study provides a good analysis of the potential sources of heavy metals and the contribution of each source. These results systematically describe the influence of cascade dams on the distribution pattern and cumulative effect of heavy metals, identify and quantify the potential sources of heavy metals and the contribution of each source, provide a theoretical reference for the segmental zoning treatment of heavy metal pollution in the inland river basin, and suggest the exclusion and remediation treatment of tailings in the Heihe River basin.

The authors would like to acknowledge the anonymous reviewers, associate editor and editor for their insightful comments and suggestions.

This work was supported by the National Natural Science Foundation of China (Grants No. 52169015), the Natural Science Foundation of Gansu Province, China (Grant No. 21JR7RA227), Qilian Mountains Eco-environment Research Center in Gansu Province (QLS202006), and the Industrial Support Plan Project of Gansu Educational Committee (2021CYZC-33).

All relevant data are included in the paper or its Supplementary Information.