Introduction

Rivers are the main natural systems delivering detrital particles from continents to the oceans. River systems cover a vast majority of the land and the particles they carry reflect geological and ecological situation depending on the geographical and climatic conditions of the river basin. The chemical composition of river sediments depends on several factors, namely, source rock composition, pattern and degree of source rock weathering, sediment sorting and recycling, as well as post-residual changes during diagenesis and metamorphism (Taylor and McLennan 1985; Sawyer 1986; Feng and Kerrich 1990; Nessbit et al. 1996; Singh and Rajamani 2001; Tripathi et al. 2007; Derakhshan-Babaei et al. 2020; Sorokina 2020).

In recent years, scientists have paid a lot of attention to studying chemical weathering since the process plays an essential part in Earth’s climate regulation due to carbon dioxide consumption through geological time. Numerous research works have attempted to understand the role of climatic, tectonic, lithological and hydrological factors in chemical weathering (e.g., Garrels and Mackenzie 1971; Bluth and Kump 1994; Louvat and Allègre 1997; Gaillardet et al. 1999a, b; Kump et al. 2000; West et al. 2005; Gislason et al. 2009; Maher and Chamberlain 2014; Torres et al. 2015; Winnick and Maher 2018). Thus, plentiful analytical data concerning different continents have been amassed and described in a series of unifying works (e.g., Gaillardet et al. 1999a, b; Viers et al. 2009; Li and Yang 2010).

Most of the mentioned studies focused on investigating the chemical composition of river sediments paying more attention to its dependence on the source rock type and/or climate. However, the impact of the factors and their ability to regulate the chemical composition of sediments in various geomorphological/tectonic conditions is ambiguous. The studies revealed contrariety of the results about the influence of tectonics, lithology, weathering and river processes on sediment composition (e.g., Yang et al. 2004; Roddaz et al. 2006; Liu et al. 2007, 2012; Borges et al. 2008; Garzanti et al. 2011; Shao et al. 2012; Chetelat et al. 2013). To better understand the mechanism and scale of the processes taking place in the river draining systems, more geochemical data from different geological and climatic environments are needed.

This work presents the research of the chemical composition of river sediments of the Amur River (Fig. 1), which was performed to distinguish chemical composition dependence on geological and geomorphological conditions. The object of the study is unique due to certain factors. First, the Amur is an extensional river system. Second, different river parts are characterized by different geomorphological conditions, whereas climatic conditions determining the degree of chemical weathering of drained rock are more or less the same.

Fig. 1
figure 1

Local map of the upper and middle Amur River with sapling sites

Full size image

Study area, geographic and geological background

The Amur River (Fig. 1) is one of the largest watercourses in eastern Asia. It mostly flows west–east through north-eastern China, Transbaikal and eastern Russia. It drains an area of 1855 th. km2 of mountainous territory. In Russia, the Amur is divided into three parts according to the water regime (Muranov 1966), geomorphological features of the plain and the channel: “Upper Amur’ (from the confluence of the Shilka and the Argun to the mouth of the Zeya River), “Middle Amur” (from the mouth of the Zeya to the mouth of the Ussury) and “Lower Amur” (from the mouth of the Ussury to the Sea of Okhotsk). The article describes river sediments of the Upper and Middle Amur.

In “Upper Amur” the river mostly flows through the low-mountain terrain and rocks often come very close to the river. In the plot, the river goes between the ranges of the Great Khingan Mountains in the south and the Yankan–Tukuringra ridges in the north mainly drains Jurassic depositional sequences and Early Cretaceous effusive and intrusive rock. Paleozoic depositional sequences and Paleozoic granitoids are less developed (Petruk and Kozlov 2009).

The river bed of the Amur in “Upper Amur” is characterized by distinct depth erosion in the tectonic uplift setting (Sorokin 2013). River alluvium is pebble-gravel and is easily carried away during summer highwaters and spring ice drifting (Sorokin 2013). The Amur valley in the reach is relatively narrow, about 2–4 km, and the tributary valleys are very narrow—1.0–1.5 km, therefore, the rivers have great strength. During high water, the Amur and its tributaries acquire a mountain river's properties capable of carrying pebbles and stones.

In “Middle Amur” the river goes along the southern edge of Amur–Zeya trench at the border with the mountainous branch of Maliy Khingan. Here Amur River has formed a wide (up to 30–50 km) asymmetrical valley. Its left bank is relatively low (up to 30–40 m) consisting of middle and upper quaternary alluvial sediments, whereas its right bank is steeper, relatively low (up to 50–100 m) and mostly consists of older friable sediments of Upper Cretaceous—Lower Cenozoic or effusive cretaceous rock (Petruk and Volkova 2012). Bed width changes from 650 to 1100 m in unbranched part and to 2–3 km in branched parts.

Downward from the Arkhara River’s mouth, the Amur plain narrows and digs into the Maliy Khingan ridges. The Amur here forms a narrow and deep valley with a steep bank, this part of the valley is called “Khingan canyon”. The bed width is 600–100 m. At the valley slopes, series of river terraces and their fragments comprising inequigranular sand, sand-gravel-pebble sediments and clays stand out. Lower and upper plains are prominent. Within the Khingan canyon, the Amur River cleaves different rock assemblages: gneisses, acidic igneous complexes, depositional assemblages, cretaceous andesites, and rhyolites are less often (Petruk and Volkova 2012).

Sampling and methods

River sediments were studied within three localities: (1) from the confluence of the rivers Shilka and Argun to the mouth of the Kalinovka River, “Upper Amur” (UA); (2) from the mouth of the Zeya River to Khingan Canyon, “Amur–Zeya valley” (AZV), and (3) within “Khingan Canyon” (KC). These localities differ in geomorphological and geological features.

Sampling was conducted along the left bank of the Amur at 10–15 km interval. The sampling plots are shown in Fig. 1. Identical parts of the river (shallow with the medium current) were chosen to sample sandy or sand-clayey river sediments. The weight of every sample accounted for about 0.5 kg. The samples were dried until air-dry condition, then fractions less than 1 mm in size were isolated and grounded with a mechanical soil-grinder.

The chemical analysis was done in the Laboratory of Analytic Chemistry, the Geological Institute, the Far East Branch of the Russian Academy of Sciences in Vladivostok using an atomic-emission spectrometry with inductively coupled plasma (ICP-AES) and mass spectrometry with inductively coupled plasma (ICP-MS). The samples were dissolved in a mixture of HF, HNO3, and HClO4 acids (2.5:1:0.5) (Zarubina et al. 2002). The concentrations of the major elements and Rb, Pb, Nb, Ba, Zr, Sr, Y, Co, Cu, V, Ni, Cr were determined by X-ray fluorescence spectrometry at the Institute of Geology and Nature Management, the Far East Branch of the Russian Academy of Sciences on ICAP 6500 Duo (Thermo Electron Corporation, USA) and Pioneer S4 (Bruker, USA) spectrometers. The concentrations of elements Sc, REE, Ta, Th, and U were determined by the ICP-MS method on a quadrupole Agilent 7500 with a mass spectrometer (Agilent Technologies, USA). The relative standard deviation for all elements did not exceed 5–10%.

The statistical analysis was performed with the software RStudio (https://r-project.org). For parametric data analysis of variance and post hoc Tukey’s range test was used; for non-parametric data—the Kruskal–Wallis test (Kruskal and Wallis 1952) and Dunn’s post hoc test with Holm–Bonferroni correction. (Holm 1979).

Results

Chemical composition of river sediments

The river sediment samples plot as wacke and as shale on the log(SiO2/Al2O3) versus log(Fe2O3/K2O) classification diagram (Fig. 2a; Herron 1988), and plot as greywacke and rarely as litharenite on the log(SiO2/Al2O3) versus log(Na2O/K2O) diagram (Fig. 2b; Pettijohn et al. 1987).

Fig. 2
figure 2

Plot of log(Fe2O3/K2O) vs log(SiO2/Al2O3) (Herron 1988) (a), plot log(Na2O/K2O) vs log((SiO2/Al2O3) (Pettijohn et al. 1987) (b) of the analyzed samples

Full size image

At present, element concentrations normalized diagrams (spider-diagrams) are used internationally to determine igneous, metamorphic and sedimentary rock's chemical composition.

The multielement diagram analysis (Fig. 3) shows that compositions of river sediments in different reaches of the Amur River compared to the upper continental crust are depleted of the oxides of the basic elements: Na, Ca, Mg. Si, Al, Ti, Fe and K concentrations are close to UCC. Mn oxide concentration normalized to the UCC shows a positive anomaly in all the parts of the river.

Fig. 3
figure 3

Multielement normalized diagram for the river sediments of the Amur River (Taylor and McLennan 1985)

Full size image

Concentrations of most trace elements, namely Cs, Pb, Nb, Ba, Zr, Sr, Zn, Cr, correspond to the UCC (Fig. 3). Ta, Sc concentrations are much lower than in the UCC. Co, Cu, V, Ni concentrations vary depending on the part of the river. So, these trace elements are in deficiency in “Upper Amur” and “Khingan Canyon” (Fig. 3). In “Amur–Zeya valley”, the element concentrations are close to the UCC. U, Y and Zn range widely in concentrations.

Concerning rare earth elements (REE), it is worth mentioning that the total number of REEs (ƩREE) in river sediments of the Amur vary from 68 to 197 ppm (Supplementary Table S1). The least ƩREE value is common for river sediments of “Khingan Canyon” (mean value 101 ppm). At “Upper Amur” and “Amur–Zeya valley”, ƩREE values in the river sediments are much higher (mean values 143 and 154 ppm, respectively).

REE chondrite-normalized patterns for river sediments in different Amur River parts are visually quite similar to each other (Fig. 4).

Fig. 4
figure 4

Chondrite normalized REE for the river sediments of the Amur River (McDonough and Sun 1995)

Full size image

At the same time there are differences in the distribution of light, medium and heavy lanthanides which is reflected in the ratio (La/Yb)n, (La/Sm)n, (Gd/Yb)n, and Eu/Eu*. So, (La/Yb)n ratio values which show the degree of fractionation of light and heavy REEs range widely from 9.6 to 15.6 (Supplementary Table S1, Fig. 5). Maximum value (14.1 on average) of the pattern is characteristic of river sediments in “Amur–Zeya valley”. In “Upper Amur” and “Khingan Canyon” they are much lower and amount to 12.8 and 12.1, respectively (Fig. 5). (La/Sm)n and (Gd/Yb)n ratio values range narrowly 3.6–4.5 and 1.8–2.6, respectively, and there are no distinct trends (Supplementary Table S1). Finally, all sample compositions have negative europium anomaly, and their (Eu/Eu*) ratio value ranges from 0.60 to 0.95 (Supplementary Table S1, Fig. 5). However, there is a trend to increasing the value downwards the Amur River.

Fig. 5
figure 5

The spatial variation of the ratio ƩREE, (La/Yb)n, Eu/Eu* of the river sediments of Amur river

Full size image

Recycling of river sediments and parent rock type

It was noted that the Amur denudates contain various rock complexes in “Study area, geographic and geological background”. Furthermore, taking account of the fact that the valley was formed, at least, in the early Cenozoic period, the recycling of the material inevitably occurs throughout the valley.

In a mineralogical aspect, the latter leads to concentrating heavy accessory minerals in the sediments and, consequently, concentrating such elements as Zr, Th, Sc (McLennan 1993, 2001; McLennan et al. 1995). However, this typically abundant in felsic rocks, whereas Sc is primarily hosted in mafic rocks. Therefore, the Th/Sc ratio does not change during sedimentary recycling and mineral transformation and may be used to detect differences in source materials' original chemical composition (McLennan 1993). Chemical element Zr is present mainly in zircon, a highly stable mineral concentrated in recycled sediments. Thus, the Zr/Sc ratio is increased during sedimentary recycling due to zircon enrichment. So, the Zr/Sc and Th/Sc ratios are clear and essential indicators of sedimentary recycling and sediment composition changes (McLennan 1993, 2001, McLennan et al. 1995).

River sediments of the Amur in all studied plots are characterized by high Zr/Sc values (over 10) (Supplementary Table S1, Fig. 6), which indicates a zircon addition, and accordingly, significant sedimentary recycling.

Fig. 6
figure 6

Zr/Sc versus Th/Sc diagram for Amur River’s sediments (McLennan 1993)

Full size image

Chemical weathering of the source area

The chemical composition of clastic sedimentary rocks can provide information about weathering in the source region. To characterize the weathering processes in the erosion area indices of weathering are used, they are presented in Table1S. First, the chemical indicators will be described, the values of which increase with weathering (Fig. 7).

Fig. 7
figure 7

The spatial variation of the CIA, CIW, PIA, V, WIP, STI, R indices for Amur River sediments, the black line indicates a trend, the grey field—standard error (diagram on the left). The variation of the mean index values and confidence intervals of the river sites (diagram on the right)

Full size image

Chemical Index of Alteration, CIA = 100 × (Al2O3/(Al2O3 + CaO + Na2O + K2O)) (Nessbit and Young 1982), Chemical Index of Weathering, CIW = 100 × (Al2O3/(Al2O3 + CaO + Na2O)) (Harnois 1988), and Plagioclase Index of Alteration, PIA = 100 × ((Al2O3–K2O)/(Al2O3 + CaO + Na2O–K2O)) (Fedo et al. 1995) are very close in nature. They are based on a ratio of mobile and immobile elements (Al is supposed to be immobile in these indices) and are used to assess the degree feldspar changes into the clay. As a degree of chemical weathering increases, alkali and alkaline earth elements are lost, and CIA, CIW and PIA values increase from 50 for unweathered rocks to values close to 100 for kaolinite.

CIA, CIW, PIA values in river sediments of the Amur range 50–63, 60–71 and 51–67, respectively Supplementary Table S1). That indicates that in the erosion area, the rocks slightly changed by chemical weathering prevail. It is also worth mentioning that there is a distinct trend to increase of the indices down the Amur River at “Upper Amur” and “Amur–Zeya valley”, which indicates that more weathered rock has been added (Fig. 7). Farther down the river, at “Khingan Canyon”, trend changes into the opposite, and simultaneously, the CIA, CIW and PIA diagrams get the “saw”-shape due to a wide range of values (Fig. 7).

Vogt’s Residual Index, V = (Al2O3 + K2O)/(MgO + CaO + Na2O), (Vogt 1927; Roaldset 1972) also belongs to the chemical indices whose values increase during rock weathering. The index is calculated as a ratio of oxides of immobile cations (Al) to mobile cations (Mg, Ca, Na) in case if potassium content remains constant during weathering (Haskin et al. 1968). Index V reflects the maturity of sedimentary deposits, and the higher the index value is, the more intense the weathering processes are (Price, and Velbel 2003; Haskin et al. 1968).

V Index values in the river sediments of the Amur vary within the interval from 1.5 to 1.8 (Supplementary Table S1), which indicates insignificant rock transformation by chemical weathering in the erosion area. Besides, V Index value increases (along with CIA, CIW, PIA indices) down the Amur River from “Upper Amur” to “Amur–Zeya Valley” (Fig. 7), whereas in “Khingan Canyon” it decreases.

Further, chemical indices are described whose values should go down during rock weathering.

Silica–Titania Index, STI = (100) × [(SiO2/TiO2)/((SiO2/TiO2) + (SiO2/Al2O3) + (Al2O3/SiO2))] (De Jayawurdena and Izawa 1994) unlike the indices mentioned above are calculated not only with silica and aluminium but also with titanium. That makes the index sensitive to the content of heavy minerals in river sediments. In recent rock, STI value reaches 90 and more, whereas it drops to almost zero in more weathered rock. STI values of the Amur River sediments lie in the interval from 82 to 87 (Supplementary Table S1), which means that the rocks slightly transformed by chemical weathering prevail in the erosion area. From “Upper Amur” to “Amur–Zeya valley”, STI values gradually decrease, but in “Khingan Canyon”, on the contrary, they increase (Fig. 7).

Ruxton ratio, R = SiO2/Al2O3 (Ruxton 1968) relates silica loss to total-element loss and considers alumina immobile during weathering. In recent sediments, R values reach 10, whereas in weathered rock they drop to almost zero. As seen from Supplementary Table S1, R values in the Amur River’s sediments vary from 6 to 10, which indicates slightly weathered rock. From “Upper Amur” to “Amur–Zeya valley” R values gradually decrease, whereas, in “Khingan Canyon” on the contrary, they increase (Fig. 7), so the trend of R-value changes is identical to the one for STI in the river sediments of the Amur River.

Weathering Index of Parker, WIP = 100 × (2Na2O/0.35 + MgO/0.9 + 2K2O/0.25 + CaO/0.7) (Parker 1970) is based on the proportions of alkali and alkaline earth metals (sodium, potassium, magnesium and calcium) present. These elements are the most mobile of the major elements, and there is no need to assume that sesquioxide concentration remains approximately constant during weathering. The index usage experience has shown that in recent sediments, the WIP value reaches 100, but in weathered rock, it drops to almost zero. As for the results from the river sediments of the Amur River, WIP values vary from 46 to 66 (Supplementary Table S1; Fig. 7), which means that in the erosion area slightly or moderately weathered rock is dominant. Due to the wide variation of the values, the trends obtained with the statistical analysis do not look conclusive (Fig. 7).

Statistical analysis results

Above have been described trends of chemical weathering indices basing on visual analysis of the diagrams. To formalize the trends, the methods of mathematical statistics were used. Figure 7, apart from diagrams showing the indices’ values and trends (on the left), presents the mean values of the indices and confidence intervals of index values for the river parts (on the right).

To show the difference between the river reaches with mathematical statistics methods, it is necessary to determine the comparison method, which depends on the data distribution. Consequently, all data have been tested for normality of data distribution with Shapiro–Wilk’s test (Shapiro and Francia 1972).

Indexes CIA, PIA, R and V values have normal (parametric) distribution, whereas CIW, STI and WIP values are non-parametric. To determine differences of index values in different plots of the Amur River, analysis of variance was performed for parametric indices, and then, in case of discrepancies were found, pair-wise comparison, i.e. Tukey’s post hoc method was applied. As for non-parametric data, statistically significant index differences were distinguished with the Kruskal–Wallis test (Kruskal and Wallis 1952), and further pair-wise comparison by Dunn’s method with Holm–Bonferroni’s correction. (Holm 1979).

Analysis of variance identified statistically significant differences between CIA values (F(2) = 17.38; p < 4e−06) in different plots of the Amur River. Next, the statistically credible difference between the localities was determined with Tukey’s test: “Upper Amur” and “Amur–Zeya valley” (p = 0.004332), “Amur–Zeya valley” and “Khingan Canyon” (p = 0.000003). The differences of PIA and R between the localities, according to the analysis of variance, were (F(2) = 16.56; p < 6.22e−06) and F(2) = 27.58; p < 3.43e−08), respectively. The p values by Tukey’s post hoc test do not exceed 0.05 as well. For V index, the analysis of variance did not distinguish any significant differences between the localities (F(2) = 0.01, p = 0.99).

For non-parametric indices, CIW and STI, Kruskal–Wallis test was applied (H(2) = 17.241; p = 0.0001803 and H(2) = 15.763; p = 0.0003776), which allowed us to distinguish statistically significant differences between the studied localities of the Amur River. Next, the pair-wise comparison by Dunn’s test with Holm–Bonferroni correction was performed. Statistically, significant differences between the localities were determined: “Upper Amur” and “Amur–Zeya valley” (p = 0.0067(CIW), p = 0.0036 (STI)), “Amur–Zeya valley” and “Khingan Canyon” ((p = 0.0001 (CIW), p = 0.0004 (STI). For non-parametric WIP index Kruskal–Wallis test was applied, which did not show any significant differences between the localities (H(2) = 0.3964, p = 0.82).

Therefore, the statistical analysis with indices CIA, CIW, PIA, STI and R indicated statistically significant differences between the studied localities of the Amur River.

Discussion

The results demonstrated that the chemical compositions of the river sediments of the Amur River are more or less variable. That is governed by the factors which are, on the one hand, different in nature, on the other hand, interrelated: (1) the composition of host rocks complexes and degree of chemical weathering; (2) geomorphological conditions including transportation.

In “Upper Amur” the river flows between mountain ridges. Presumably, this is the condition under which the rock slightly transformed by chemical weathering comes into the river, which is reflected in low CIA, CIW, PIA, V values, and on the contrary, high STI and R values (Fig. 7). As soon as the river gets to “Amur–Zeya valley”, the situation changes: CIA, CIW, PIA, V values go up, but STI and R values go down. Its cannot be asserted that the amount of the material delivered from the areas where the chemical weathering processes are well-developed increases in the locality. However, it is obvious that the share of unweathered material, which is dominant in “Upper Amur”, reduces. Farther on down the river, in “Khingan Canyon”, the river flows in mountainous terrain again, and slightly weathered material prevails again, which is seen from the decrease of CIA, CIW, PIA, V values, but the increase of STI and R values (Fig. 7).

Another worth-mentioning aspect is the relation of chemical weathering indices and rare earth elements behaviour.

Recent studies suggest that REEs are usually immobile and poorly fractionated during weathering, transportation, diagenesis and metamorphism of low and medium level, so they are considered to be a reliable indicator of source rock (e.g., Chaudhuri and Cullers 1979; Goldstein and Jacobsen 1988; McLennan 1989; Elderfield et al. 1990; Dupré et al. 1996; Gaillardet et al. 1997; Douglas et al. 1999; Viers et al. 2000; Tosiani et al. 2004). Most river sediments have similar, homogeneous REE-patterns with enrichment of light REEs (LREEs) and depletion of heavy REEs (HREEs), with weak Ce- and/or Eu-anomalies or without them (Goldstein and Jacobsen 1988; Sholkovitz 1988).

In Table S1, correlation coefficients between ƩREE, (La/Yb)n, (Eu/Eu*) in the river sediments of the Amur River are given on one side, and the chemical weathering indices—on the other. The statistical analysis results suggest that ƩREE has a positive correlation with PIA, CIA, CIW indices, and a negative correlation with the indices STI and R. Eu/Eu* values positively correlate with WIP index.

It may seem that REE behaviour is regulated by weathering processes. However, it is inconsistent with two factors. Firstly, the explanation is suitable for hot and humid climate causing intense weathering (Su et al. 2017). The Amur basin is situated in the temperate climate zone, which has not changed during Holocene age (Yu et al. 2017). Secondly, the explanation implies a positive correlation between (Eu/Eu*) and indices calculated with aluminium (namely CIA, CIW, PIA, V, STI and R), which is not our case. Therefore, other explanations of relations between REEs behaviour and weathering indices in the river sediments of the Amur River should be found.

In our opinion, the most likely explanation of the positive correlation of ƩREE with CIA, CIW, PIA values, and at the same time, negative correlation with STI and R values is the character of sorting and transporting processes in the river sediments. In particular, the river changes from mountainous to plain when it shifts from “Upper Amur” to “Amur–Zeya valley”; consequently, the river flow rate slows down. In river sediments, the portion of clay-size fraction, characterized by selective flotation with light lanthanides, increases (Su et al. 2017). Thus, ƩREE in the river sediments at the shift “Upper Amur” to “Amur–Zeya valley” increases as well as CIA, CIW, PIA values while STI and R values go down, as it was mentioned in part 2.2. At the shift from “Amur–Zeya valley” to “Khingan Canyon” the river flow rate rises, river sediments lose a large part of the clay-size fraction, and the process described above goes contrariwise.

Conclusion

According to the major and trace element geochemistry data of the bulk composition of recent sediments from the upper and middle Amur River, the following conclusions arose:

  1. 1.

    Concentrations of most major and trace elements (e.g., Si, Al, Ti, Fe, K, Cs, Pb, Nb, Ba, Zr, Sr, Zn, Cr) in the river sediments of the Amur correspond to those in the UCC. However, there are significant differences in the bulk composition of recent sediments in different localities of the Amur River.

  2. 2.

    The lowest ƩREE values are common in river sediments of “Khingan Canyon” (mean value 101 ppm). In “Upper Amur” and “Amur–Zeya valley” ƩREE values are much higher (mean values 143 and 154 ppm, respectively). Maximum (La/Yb)n values are common in the river sediments of “Amur–Zeya valley”. In “Upper Amur” and “Khingan Canyon” they are much lower. (Eu/Eu*) ratio value varies from 0.60 to 0.95 and tends to rise down the Amur River.

  3. 3.

    From “Upper Amur” to “Amur–Zeya valley” CIA, CIW, PIA and V values go up, whereas STI and R values go down. Farther down the river to “Khingan Canyon” opposite process occurs. It is due to the changes in the river reaches from mountainous to plain and contrariwise.

  4. 4.

    The statistical analysis results show a positive correlation of ƩREE with PIA, CIA, CIW indices, and a negative correlation with STI and R indices. Eu/Eu* ratio values have a positive correlation with WIP index. This pattern could be explained by characteristics of sorting and transporting river sediments but not by the influence of chemical weathering.