One river, two streams: chemical and chromium isotopic features of the Neglinka River (Karelia, northwest Russia)

ABSTRACT

The physico-chemical and hydrochemical characteristics of run-off of the Neglinka River Basin (northwest Russia) monitored for a year are different for the upstream and downstream sections. The river known hydrologically as the Neglinka, consists hydrochemically of two different streams: one represented by the upstream part of the basin, and the other one by the downstream. The upstream water is characterized by low mineralization (water hardness 0.08–0.43 mmol L⁻¹) and low δ⁵³Cr values (+0.30 to +0.42‰), whereas the lower part is characterized by high mineralization (water hardness 0.37–3.46 mmol L⁻¹) and high δ⁵³Cr values (+0.92 to +1.73‰). The difference in chemical composition of the upstream and downstream waters could be due to the underground discharge input. Aqueous chromium (Cr) mobilized from weathering profiles may have been reduced from soluble Cr(VI) to insoluble Cr(III) during the riverine transportation. Partial removal of Cr from the water balance resulted in a decrease in Cr concentration and an increase in δ⁵³Cr values.

KEYWORDS:

• hydrochemistry
• run-off
• chromium isotopes
• Neglinka River
• northwest Russia

Editor R. WoodsAssociate editor G. Chiogna

1 Study area

The short (14-km-long) Neglinka River originates in a small forest lake and in its lower part runs through the city of Petrozavodsk (Fig. 1). The average annual water discharge in the estuary of the river is 0.47–0.51 m³ s⁻¹ (Ieshina et al. 1987, Kalinkina et al. 2012, Borodulina 2013). Most of the Neglinka basin (46.1 km²) is located in forested lowland, but the downstream reaches of the Neglinka River run through the hilly environment of the Lake Onega shore. The riverbed stretches along the Petrozavodsk depression in a crystalline basement. The depression is filled with a sedimentary complex of late Proterozoic quartzite-sandstone, and is covered with a thick layer (up to 100 m) of Quaternary glacial sediments (Borodulina 2006, Karpechko 2013). The bedrock and glacial sediments are covered in places with a thin (<1 m) soil layer.

Figure 1. Schematic location map of the sampling sites: 1 is the upstream (university) site, and 2 is the downstream (city) site. Inset: Location of the sampling area (star) in the border area between Russia and Finland.

Geochemical studies conducted so far have shown that the run-off water in the downstream section of the Neglinka River Basin was characterized by concentrations of basic cations (Ca²⁺, Mg²⁺, Na⁺ and K⁺) in the range of 6–70 mg L⁻¹, with a total mineralization level varying from 35 to 400 mg L⁻¹ throughout the year (Borodulina 2013). Biogenic elements and compounds such as P_min, P_tot, and N-NH₄⁺ displayed changes over the long-term observations. Whereas average yearly concentrations of N-NH₄⁺ decreased from 0.18 mg L⁻¹ in 1976 to 0.07 mg L⁻¹ in 2016, concentrations of P_min and P_tot increased for the same period from 59 to 148 μg L⁻¹, and from 76 to 235 μg L⁻¹, respectively (Pirozhkova 1981, Dzjubuk and Kljukina 2015, Sabylina and Efremova 2017). Annual observations showed that electrical conductivity (EC; a proxy to water mineralization) at the river mouth varied from ~200 to ~400 μS cm⁻¹ during most of the year depending on the amount of atmospheric precipitation, but dropped sharply to 35 μS cm⁻¹ with the beginning of the spring snowmelt (Borodulina 2013). The acid-alkaline features of water in the river mouth changed similarly: relatively steady pH values (7.5–7.8) were typical for winter, then a sharp decrease in pH (down to 5.8) was observed in April–May, while significant variations in pH were observed during summer and autumn (6.6–8.2; Borodulina 2013).

Studies conducted on the Neglinka River Basin run-off to date have mostly dealt with changes in concentrations of basic cations and biogenic elements, and with physico-chemical properties of water (e.g. Pirozhkova 1981, Ieshina et al. 1987, Borodulina 2006, 2013). Dzjubuk and Kljukina (2015) reported some limited information on the Neglinka River water composition and physico-chemical properties from three sampling sites (upstream, midstream and downstream) for the period 2013–2014, but water samples were collected three times per year only. Data on concentrations of Zn, Cu and Pb in upstream and downstream Neglinka River Basin run-off samples collected in August 1999 are given in Komulaynen and Morozov (2007). Two most comprehensive works concerning the issues raised in this study are those by Litvinenko and Regerand (2013) and Slukovskii (2014). To the best of our knowledge, no detailed monthly two-site observations of the Neglinka River Basin run-off over an extended period of time were ever conducted.

The information published to date on water parameters in the Neglinka River has allowed us to hypothesize that involvement of underground discharge might have influenced the chemical and physical properties of run-off water significantly (cf. Borodulina 2013, Borodulina et al. 2015). Since mineralization of groundwater in the region is usually several times higher than that of the surface water (Ieshina et al. 1987, Borodulina 2006), a significant increase in water mineralization for the Neglinka River downstream section may point to groundwater input. To test this hypothesis, we studied the chemical composition of run-off water samples added by a limited number of Cr isotope data by monitoring the water properties for a period of almost a year.

2 Sampling and analyses

Run-off samples were collected at two sites on the Neglinka River stream, separated by some 7 km. The first site is located in the upstream section outside the city limits, and the second site is located approx. 50 m upstream of the river mouth within the limits of the city of Petrozavodsk (Fig. 1). These two sampling sites were chosen because a comparison of the observations would allow us to make suggestions about the contribution of different solutes to the water balance along the river. One additional sample was collected from the Bezymjanny tributary, which enters the Neglinka River in the middle of the city (Fig. 1).

2.1 Non-isotopic analyses

Chemical analyses of run-off water were performed in the chemical laboratory of the Northern Water Problems Institute (Petrozavodsk, Russia). Samples prepared for analysis of dissolved metals were filtered through a 0.45-μm membrane filter and then acidified with ultrapure concentrated HNO₃ to pH <2. Total iron (Fe_tot) was analysed from unfiltered samples. Before the analysis, the well-mixed sample was digested in the microwave for 30 min in the presence of concentrated HNO₃ and H₂O₂ to obtain a totally transparent solution. Concentrations of Cr, Cu, Fe and Mn were measured with the use of an AA-6800 Shimadzu atomic adsorption spectrophotometer (https://www.shimadzu.com/). While concentrations of Fe and Mn were measured with the use of the atomic absorption method (flame atomization), concentrations of Cu and Cr were measured using a deuterium-corrected electro-thermal atomization method (e.g. Price 1979, Lozovik and Efremenko 2017). Concentrations of suspended iron were calculated by subtraction of dissolved iron (Fe_dis) from concentrations of Fe_tot.

Concentrations of major ions (Na⁺, K⁺, Ca²⁺, Mg²⁺, SO₄^2- and Cl⁻) were measured in unacidified subsamples filtered through the 0.45-μm membrane filter. Sodium, K, Ca and Mg concentrations were measured on an atomic-adsorption spectrophotometer (AA 6200 Shimadzu). While Na and K were measured in the atomic emission mode, Ca and Mg were measured in the atomic adsorption mode (e.g. Price 1979, Lozovik and Efremenko 2017). Photometric methods with application of Sulfonazo III Na salt, and Cl⁻ with Hg thiocyanate and Fe(III) nitrate were used to determine SO₄^2- (Lozovik and Efremenko 2017).

Unfiltered and unacidified samples were used to determine EC and pH of water along with concentrations of biogenic elements and compounds (P_min, P_tot, N-NH₄⁺). Electrical conductivity was measured using an Agilent 3200C conductometer (https://www.agilent.com). A pH analyser (PHM 410) was used to determine pH by potentiometric method. Biogenic components were identified by photometric methods (Murphy and Riley 1962, Boeva 2009, Lozovik and Efremenko 2017). All photometric analyses were conducted on a Portlab 501 spectrophotometer. Dissolved organic carbon (DOC) concentrations were determined using an experimental set-up of a standard UV/peroxidosulphate method (Zobkov and Zobkova 2015).

2.2 Isotopic analyses

Isotope measurements were performed in laboratories of the Czech Geological Survey (Prague, the Czech Republic). Immediately after sampling, water samples were adjusted to pH ≥ 8 by ultrapure NH₄OH in order to maintain Cr(VI) stability (cf. Cadkova and Chrastny 2015), and then filtered through the 0.45-μm membrane filters. Elution chromatography, modified from Chrastny et al. (2011) and Farkas et al. (2013), was applied for Cr(VI) purification. Volumes of sample containing ~1 μg Cr were loaded on columns filled with 2 mL of the anion exchange resin AG1-X8 (100–200 mesh, Bio-Rad, USA). Cations were eluted by washing the resin with 0.1M HCl. Chromium was eluted from the resin by washing with 6M HCl. Collected Cr samples were evaporated and re-diluted in 2% HNO₃ to be analysed. The NIST® (National Institute of Standards and Technology) SRM (standard reference material) 979 chromium isotopic standard was processed through the columns along with samples to ensure lack of fractionation during the elution procedure. The procedural blank was negligible (15–20 ng Cr) as compared to the amount of Cr in samples.

The Cr isotope measurements were performed on a multi-collector inductively coupled plasma-mass specrometer (MC-ICP-MS) Neptune (Thermo) operating in a high-resolution mode (M/∆M_(95%) = 10 000). The applied instrumental set-up is described in Farkas et al. (2013). Because Cr has four stable isotopes, it was possible to utilize a ⁵⁰Cr-⁵⁴Cr double spike for mass bias and isobaric interference corrections. The analytical solution was introduced into the plasma using an ARIDUS II (Cetac, USA) desolvating nebulizer system and an ASX-112 FR (Cetac, USA) autosampler. All Cr isotopes and elements with isobaric overlaps were measured simultaneously. Polyatomic interferences were avoided by making measurements on the Cr peak shoulders. A NIST® SRM 979 standard doped with a double spike was analysed after every four samples. A blank measurement was made between each sample and/or NIST® SRM 979 standard.

Chromium isotope compositions are expressed in the per mil (‰) notation relative to the Cr standard, as follows:

(1) $\begin{array}{rl}& \text{[}{\delta }^{53}\mathrm{C}\mathrm{r}({\mathrm{\%}}_0)=\left[{(}^{53}\mathrm{C}\mathrm{r}{/}^{52}\mathrm{C}{\mathrm{r}}_{\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}})/{(}^{53}\mathrm{C}\mathrm{r}{/}^{52}\mathrm{C}{\mathrm{r}}_{\mathrm{N}\mathrm{I}\mathrm{S}\mathrm{T}979})-1\right]\\ & \times 1000\end{array}$(1)

The δ⁵³Cr value of the NIST 979 doped with the double spike was monitored over the course of the measurements, and the average value obtained was 0.02 ± 0.01‰ (2σ) (n = 10). The isotope composition of all samples was corrected to this value. The average δ⁵³Cr value for the NIST 979 standard processed through the chromatographic columns was 0.03 ± 0.01‰ (2σ) (n = 8). The average reproducibility of δ⁵³Cr measurements for samples was ±0.05‰ (2σ) (n = 12).

3 Results

The physico-chemical, chemical and isotopic characteristics of water from the two studied sampling sites were found to be significantly different (cf. Dzjubuk and Kljukina 2015). Concentrations of major ions were about an order of magnitude higher in samples from the downstream section of the Neglinka River than from its upstream section (Table 1). Concentrations of most measured metals were also higher in downstream samples (Table 1; Fig. 2). While, in upstream samples, the behaviour of dissolved iron (Fe_dis) was somewhat erratic, with concentrations varying from 390 to 1280 μg L⁻¹, downstream samples displayed a steady decrease in Fe_dis concentrations (except for November 2017) from 1050 to 280 μg L⁻¹ over the course of observation. In all cases, total iron (Fe_tot) concentrations remained mostly higher in downstream samples, with a large peak (>11 mg L⁻¹) in April 2018. For the upstream site, most Fe is represented by its dissolved variety. Such a difference in Fe behaviour was due to high amounts of suspended iron (Fe_sus) as compared to the downstream samples. Manganese concentrations varied from 18 to 66 μg L⁻¹ for upstream samples and from 48 to 184 μg L⁻¹ for downstream samples, depending on the season (a drop in concentrations was noted for downstream samples in April 2018). Concentrations of Cu varied from 0.6 μg L⁻¹ for the upper part of the river to 10 μg L⁻¹ for the lower part. Chromium concentrations display strong seasonal changes for both upstream and downstream samples (Fig. 2(c)). For most of the year, concentrations of Cr (unlike other metals) are higher in the upper part of the stream (1.0–1.5 μg L⁻¹) than those measured in the lower part of the stream (0.5–1.0 μg L⁻¹). However, starting from April 2018, the picture changed and concentrations became nearly the same for water from the both parts of the river: 0.7–1.4 μg L⁻¹ of Cr in the upstream section and 0.7–1.6 μg L⁻¹ in the downstream.

Table 1. Characteristics of run-off water samples from the Neglinka catchment. Sample locations, US: upstream; DS: downstream. B: Bezymjanny. EC: electrical conductivity; Hard.: hardness; Turb.: turbidity; SD: standard deviation.

Sampling	Sample	pH	EC	Hard.	Na	K	Mg	Ca	Cl	SO4	N-NH4	Turb.	Pmin	Ptot	Mn	Fedis	Fetot	Cu	Cr	δ^53Cr	SD (2σ)
date	location		(μS cm^−1)	(mmol L^−1)	(mg L^−1-)	(mg L^−1)	(mg L^−1)	(mg L^−1)	(mg L^−1)	(mg L^−1)	(mg L^−1)	(mg L^−1)	(μg L^−1)	(μg L^−1)	(μg L^−1)	(μg L^−1)	(μg L^−1)	(μg L^−1)	(μg L^−1)	(‰)	(‰)
06.09.2017	US	4.30	35.4	n.d.	n.d.	n.d.	n.d.	n.d.
06.09.2017	DS	7.30	296	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	1.0	1.1	0.420	0.031
04.10.2017	US	4.60	30.8	0.15	1.46	0.32	0.79	1.69	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	n.d.	3.0	1.0	1.596	0.058
04.10.2017	DS	6.79	206	1.50	12.4	2.37	6.85	18.8	1.27	1.03	0.04	0.58	15	38	49	980	980	0.9	1.2	n.d.
07.11.2017	US	5.41	30.5	0.22	2.04	0.60	1.12	2.47	13.3	13.7	1.0	1.74	187	251	138	1050	1240	2.2	0.9	n.d.
07.11.2017	DS	7.36	216	1.49	13.6	2.53	7.00	18.4	1.70	0.90	0,03	2.90	19	44	45	790	1010	2.9	1.5	0.402	0.038
05.12.2017	US	5.32	28.5	0.19	1.84	0.42	1.03	2.18	14.0	12.0	1.9	21.5	56	82	113	790	1830	3.6	1.0	1.730	0.065
05.12.2017	DS	7.62	244	1.50	14.0	2.23	7.20	18.0	1.50	1.50	0.06	0.58	21	38	27	849	891	0.9	1.0	n.d.
10.01.2018	US	5.23	32.5	0.21	1.67	0.48	1.11	2.45	15.8	15.0	0.37	2.90	87	119	156	1038	1350	2.3	0.7	n.d.
10.01.2018	DS	7.26	220	1.26	12.9	2.10	6.10	15.2	1.38	1.79	0.06	0.58	14	35	47	860	880	1.5	1.0	n.d.
06.02.2018	US	5.76	30.1	0.22	1.87	0.49	1.28	2.36	15.7	11.9	0.26	1.16	52	90	125	900	1060	2.1	0.7	n.d.
06.02.2018	DS	7.62	329	2.36	15.9	2.75	12.1	27.5	1.70	1.14	0.11	0.58	28	49	44	1040	1040	1.0	1,1	n.d.
01.03.2018	US	6.31	41.7	0.30	2.57	0.59	1.77	3.06	25.0	19.0	0.33	<0.58	74	98	167	810	980	1.8	0,5	n.d.
01.03.2018	DS	7.62	380	2.82	20.0	3.60	14.7	32.2	1.84	1.16	0.17	1.74	39	62	66	1280	1340	0.6	1,0	0.295	0.033
09.04.2018	US	6.02	41.7	0.27	2.97	0.97	1.53	2.78	29.5	22.5	0.34	0.58	69	92	184	740	900	1.7	0,5	0.922	0.051
09.04.2018	DS	7.49	244	1.16	19.0	3.13	4.83	15.1	2.98	3.08	0.08	<0.58	24	64	52	890	1180	1.0	0,7	n.d.
07.05.2018	US	4.59	24.8	0.08	1.01	0.37	0.41	0.87	27.5	12.6	0.86	74.0	316	544	180	720	11600	10	0,8	n.d.
07.05.2018	DS	6.83	53,4	0.37	3.20	0.75	1.75	4.50	0.62	1.81	0.01	0.58	5	26	22	390	390	0.9	0,7	n.d.
05.06.2018	US	6.44	31.3	0.23	2.80	0.49	1.32	2.38	4.86	6.05	0.05	10.4	24	76	52	420	950	2.3	0,7	n.d.
05.06.2018	DS	7.91	402	2.84	21.0	3.78	14.1	33.7	1.39	1.55	0.03	<0.58	28	61	40	680	910	1.4	1,4	n.d.
05.06.2018	B	7.52	405	3.18	16.9	3.34	15.4	38.4	29.8	23.8	0.02	11.0	34	109	85	335	1260	3.3	1,6	n.d.
04.07.2018	US	6.87	54.4	0.43	3.51	0.84	2.58	4.30	22.9	21.1	0.04	5.20	21	61	346	605	1280	1.4	0,9	n.d.
04.07.2018	DS	7.88	583	3.46	47.0	4.82	16.9	41.4	1.78	0.95	0.12	<0.58	54	107	18

Note: The AA-6800 has a minimum detection limit of 0.10 mg/L for Cr, 0.03 mg/L for Cu, 0.02 mg/L for Mn and 0.1 mg/L for Fe in the flame method, and 0.5 µg/L for Cr and 0.5 µg/L for Cu in the electro-thermal atomization method. Chromium (wavelength 357.9 nm), Cu (324.8 nm), Mn (279.5 nm) and Fe (248.3 nm) specific hollow cathode lamps were used for analyses.

The AA-6200 has a minimum detection limit of 0.1 mg/L for Na and 0.1 mg/L for K in airacetylene flame, and 1.0 mg/L for Na and 0.5 mg/L for K in airpropane flame. A minimum detection limit is 0.2 mg/L for Ca and 0.04 mg/L for Mg in airacetylene flame. Sodium (589 nm) and K (766.5 nm) were measured in the atomic emission régime, while Ca (422.7 nm) and Mg (285.2 nm) were measured in atomic adsorption mode with the use of specific hollow cathode lamps.

n.d.: no data available.

This study confirms observations made by Borodulina (2013) about changes in water hardness along the stream: the water hardness of upstream samples varied from 0.08 to 0.43 mmol L⁻¹, whereas that of downstream samples varied from 0.37 to 3.46 mmol L⁻¹, depending on the season (Table 1; Fig. 3(c)). Biogenic elements and compounds (P_min, P_tot, N-NH₄⁺) were also much more abundant in downstream samples, with concentrations peaking in April 2018 (Table 1). Physico-chemical water parameters were also different for samples collected from the two sites (Table 1; Fig. 3). Electrical conductivity of upstream samples varied from 24.8 to 54.4 μS cm⁻¹, whereas it was almost an order of magnitude higher in downstream water (206–583 μS cm⁻¹ over the year dropping to 53.4 μS cm⁻¹ for samples collected in early May 2018; cf. Borodulina 2013). Upstream water displayed a wide range of pH values, varying over the period of observation from acidic to neutral (pH = 4.3–6.9), whereas the acidity was roughly neutral (pH = 6.8–7.9) in downstream samples.

Figure 3. Variation in (a) pH, (b) electrical conductivity, and (c) water hardness values over the period of observation. For the conductivity and water hardness, values for the city samples are divided into 10 for the sake of better presentation.

Pilot analyses of Cr isotopic systematics for the Neglinka River showed that δ⁵³Cr values are different for the two studied sites (up to ~1.3‰). Downstream water samples displayed much higher δ⁵³Cr values (from +0.92 to +1.73‰) than upstream water samples (from +0.30 to +0.42‰), i.e. the downstream water was much richer in the heavier isotope ⁵³Cr (Table 1). The range of upstream δ⁵³Cr values is comparable to that reported by Qin and Wang (2017) for values from elsewhere (mostly around +0.3 ‰), whereas downstream water displayed very high δ⁵³Cr values (cf. Farkas et al. 2013).

4 Discussion

One of the observed features of the Neglinka water was a sharp increase in concentration of biogenic elements/compounds, Cu and Fe for downstream samples collected in early April 2018 (Table 1). During that time, drainage of organic, dust and technogenic pollutants from the city snow cover and soil occurred (it is known that snow cover in the city of Petrozavodsk has elevated concentrations of Cu, Mn, Pb, Zn and other trace metals) (Krutskikh and Kritchevtsova 2011). This resulted, first, in a severe increase of particulate load in the river water within the city limits. While, over most of the year, the water turbidity had varied between 0.6 and 1.0 mg L⁻¹, it increased sharply in early April 2018 to 74 mg L⁻¹, and then slowly decreased in May and June 2018 to 10–11 mg L⁻¹. Second, a strong increase in concentrations of biogenic elements was observed during the same period for downstream water samples (P_min: 316 μg L⁻¹, P_tot: 544 μg L⁻¹ and N-NH₄⁺: 0.86 mg L⁻¹). That clearly suggests strong pollution of the Neglinka River with biogenic solutes. Third, typical for the Neglinka upstream samples, low pH values and a low amount of suspended particles, accompanied by high concentrations of organic compounds, are overall favourable for the existence of iron in dissolved form (Linnik and Nabivanets 1986).

For the downstream water samples, however, the city water drainage into the Neglinka River changes the physico-chemical properties as well as the amount of suspended particles. As a result, redistribution of metallic forms occurred, and the proportion of suspended Fe significantly increased. Since the snow cover of Petrozavodsk city has high concentrations of various metals (Krutskikh and Kritchevtsova 2011), a very sharp increase in concentrations of Cu and Fe_tot noted in April 2018 in the downstream Neglinka River (10 μg L⁻¹ of Cu and >11 mg L⁻¹ of Fe_tot) could have resulted from the city drainage system being overwhelmed by polluted snowmelt products. That is consistent with the observations of Slukovskii (2014, 2015) and Slukovskii and Svetov (2016), who conducted a study of the Neglinka River sediments, and concluded that the sediments from the urban part of the river are moderately to heavily polluted with metals.

Run-off from the upstream section of the Neglinka River (the rural part), on the other hand, did not display an increase in turbidity in April 2018 (0.6–2.9 mg L⁻¹ over the course of observations). Since no urban pollution occurred here, no sharp changes in concentrations of organic and particulate load took place in April 2018. In early May 2018, sharp drops in both concentrations of all measured ions and values of physico-chemical parameters were observed for both upstream and downstream water samples (Table 1; Figs. 2 and 3). The reason for this is quite obvious: extensive snowmelt occurred in the region; therefore, all river water was strongly diluted with the products of the snowmelt (cf. Borodulina 2013).

4.1 Non-isotopic study

From the analytical results obtained, it appears that two different water streams are present within the single river system studied. It is interesting that in July 2011, the hardness of the water (another proxy to its mineralization) was measured along the Neglinka and Lososinka rivers (two major streams running through Petrozavodsk; Fig. 1) (Borodulina 2013). While the water hardness in the Lososinka River remained quite steady over the whole length of the stream (0.82–0.95 mmol L⁻¹), that in the Neglinka River increased by an order of magnitude in the downstream direction (from 0.28 to 3.76 mmol L⁻¹). If the sources of mineralization were atmospheric precipitation and/or industrial pollution, then the change in mineralization level would be similar for the two closely located streams. However, a significant increase in water hardness for the Neglinka River vs steady hardness for the Lososinka River suggests an influence of underground discharge for the Neglinka River. The presence of underground springs along the downstream reaches of the Neglinka River and along the left-side tributary, the Bezymjanny (Borodulina 2013), and the fact that mineralization of groundwater is usually significantly higher than mineralization of the surface water (Ieshina et al. 1987, Borodulina 2006) are consistent with such a suggestion.

In addition to monitoring the composition of the Neglinka River water, we conducted several analyses of chemical and physico-chemical properties of water from the Bezymjanny tributary (Fig. 1). The tributary (~750 m long) originates in a small spring characterized by EC of 220–255 μS cm⁻¹ and pH of 6.9–7.3, i.e. values similar to those measured for the Neglinka downstream section. The average annual concentration of Cr in the spring water was 0.5 μg L⁻¹, and that of Cu was 0.8 μg L⁻¹ (Borodulina 2006). Concentrations of Cr and Cu in the Bezymjanny water samples remained about twice lower than in the Neglinka samples. In early June 2018, Cr was 0.9 μg L⁻¹ for the Bezymjanny tributary and 1.6 μg L⁻¹ for the downstream Neglinka River, while Cu was 1.4 and 3.3 μg L⁻¹, respectively. On the other hand, concentrations of the other elements and compounds in the Bezymjanny tributary were very close to those measured in the downstream Neglinkla samples (Table 1). The physico-chemical properties of water from the Bezymjanny tributary were also very close to those in the Neglinka downstream section: pH values for the Bezymjanny tributary varied from 6.9 to 7.5, and EC varied from 199 to 405 μS cm⁻¹ (Borodulina 2013, Dzjubuk and Kljukina 2015; this study; cf. Table 1 for the Neglinkla River). These values are very close to those in the Bezymjanny spring source. Such a distribution of the parameters for different parts of the Neglinka River is consistent with the suggestion that, although the Neglinka River is a single system in hydrological terms, hydrochemically it is represented by a combination of two independent systems: (i) its lower part (along with the Bezymjanny tributary), and (ii) its upstream part (from the source to approximately the place where the Bezymjanny tributary enters).

We noted that, for most of observation period, concentrations of Cr were approx. 10–30% higher in water collected from the upstream section than in that collected from the downstream section, including the Bezymjanny tributary. This suggests that the addition of Cr to the downstream water was insignificant (even from the underground discharge), i.e. the main source of Cr was located in the upper part of the river. In particular, a very low Cr concentration in the spring source of the Bezymjanny tributary is consistent with this suggestion. Deviations from the main-observed Cr concentration pattern occurred in April and May 2018, when Cr concentrations decreased in upstream run-off samples and increased in downstream samples (Fig. 2(c)). In June 2018, both the Neglinka sampling sites displayed a sharp increase in concentrations of Cr (1.4–1.6 μg L⁻¹). Such an increase for the downstream section might have been due to limited incorporation of industrial Cr-bearing pollutants (cf. Komulaynen and Morozov 2007), whereas the Cr concentration in the upstream section likely just came back to the natural seasonal level. Although overall Cu concentrations are higher in the urban part of the river than in the rural part, the consistent behaviour of Cu concentrations in both parts (Fig. 2(d)) suggests that no significant industrial Cu input occurred. The only obvious urban Cu input was noted in April 2018, when its concentration in the downstream water reached 10 μg L⁻¹ (cf. Komulaynen and Morozov 2007).

Figure 2. Variation in (a) Fe, (b) Mn, (c) Cr and (d) Cu concentrations in the Neglinka River over the period of observation.

4.2 Isotopic study

In order to better understand the behaviour and origin of Cr in the Neglinka River, we measured Cr isotopic composition in six samples from both the upstream and downstream sections of the river. Overall, Cr exists in water-insoluble [Cr(III)] and water-soluble [Cr(VI)] oxidation states (e.g. Qin and Wang 2017). Reactions of Cr(III) oxidation to Cr(VI) and/or Cr(VI) back-reduction to Cr(III) result in Cr fractionation (D´Arcy et al. 2016 and references therein). During oxidation of soil/bedrock Cr(III), the heavier ⁵³Cr isotope would be released preferentially into a soluble Cr(VI) form. This Cr(VI), enriched in the heavier ⁵³Cr isotope, can be mobilized to water and become a subject of preferential reduction of the lighter ⁵²Cr isotope. Therefore, further enrichment of the remaining aqueous Cr(VI) in the heavier isotope (increase of a δ⁵³Cr value) would occur as a result of back-reduction. Although δ⁵³Cr values for river run-off from elsewhere average mostly around +0.3‰ (Qin and Wang 2017), a number of recent studies have reported δ⁵³Cr values of streams and rivers ranging between – 0.17 and +1.68‰. The highest values could result from fractionation due to Cr oxidation followed by back-reduction (D´Arcy et al. 2016 and references therein).

A significant difference in δ⁵³Cr values between run-off samples from the upstream and downstream sections of the Neglinka River (+0.30 to +0.42‰ and +0.92 to +1.73‰, respectively) implies at least two major scenarios of aqueous Cr generation and evolution: (i) two independent sources with different Cr isotopic characteristics supplied Cr for the two different parts of the river; and (ii) a sole source supplied Cr, but aqueous Cr(VI) reduction occurred during the riverine transport. The first scenario would easily explain the difference in δ⁵³Cr values for the two parts of the river by incorporation of materials with different Cr isotopic characteristics. Industrially polluted groundwaters might have been the one of the Cr sources for the first scenario. Izbicky et al. (2012) and Novak et al. (2014, 2017a) showed that polluted groundwater in industrial areas might have produced δ⁵³Cr values of dissolved Cr(VI) varying from +1.6 to +3.9‰, which are significantly higher than technological solutions themselves. Unlike in the native waters, the positive δ⁵³Cr shift in industrially contaminated water is believed to be solely due to Cr(VI) reduction (Novak et al. 2017b). Waters related to bodies of ultramafic rocks could be the second possible Cr source in the first scenario. Farkas et al. (2013) demonstrated that serpentinite catchment streams may display δ⁵³Cr values of up to +3.9‰. Therefore, addition of serpentinite-related groundwater would increase δ⁵³Cr values of the river water. However, no serpentinite bodies exist, either in the vicinity of the city of Petrozavodsk or in the Neglinka catchment. Therefore, incorporation of Cr from serpentinite-related groundwater seems to be unlikely in the case considered. An additional objection to the first scenario is that both industrially polluted groundwater and groundwater from serpentinite catchments have much higher concentrations of Cr than the concentrations measured for any water sample from the Neglinka River (Novak et al. 2014, 2017a, 2017b). Therefore, incorporation of either industrially polluted groundwater or serpentinite-related groundwater would increase the concentration of Cr, which is not the case for the Neglinka downstream water.

The second scenario suggests that the overall decrease in Cr concentrations could be due to partial spontaneous reduction of aqeous Cr(VI) to Cr(III) during riverine transport in the Neglinka River. Insoluble Cr(III) precipitated from water would decrease Cr concentration in the run-off and increase its δ⁵³Cr value. Data for the Neglinka River waters are more consistent with this second scenario. A wide range of δ⁵³Cr values (from +0.30‰ to +1.73‰) appears to be due to variable redox conditions existing in the catchment. Some previous works have shown that, while, during the oxidation, δ⁵³Cr values shift to higher numbers by just a few tenths of the per mil value, the reduction shifts the δ⁵³Cr of the residual reactant to higher values, by 2.2–5.4‰ (Ellis et al. 2002, Zink et al. 2010, Kitchen et al. 2012, Basu et al. 2014). Although there are no data on the rate of such reduction in the environment yet, a work of Cadkova and Chrastny (2015) showed that aqueous Cr(VI) is unstable over even short time periods. In experimental solutions containing various reducing agents, δ⁵³Cr values increased by 1.0–1.8‰ over a period of <72 h. Such change in δ⁵³Cr values was accompanied by a reduction of 27–38% of Cr(VI).

In the case considered, Cr is assumed to be leached from local weathering profiles and delivered to the source forest lake. Although we do not have exact information on Cr isotopic composition of the Neglinka catchment bedrock as yet, bedrock is usually slightly depleted in ⁵³Cr, displaying δ⁵³Cr values of around 0 to – 0.1‰ (cf. Zhu et al. 2018). Since, during oxidation of Cr(III) to water-soluble Cr(VI) δ⁵³Cr, values of the resulting reagent [Cr(VI)] shift to higher values only slightly, some reduction of Cr(VI) mobilized to the Neglinka waters might have occurred during the riverine transport from the Cr source to the upstream sampling site (δ⁵³Cr values of upstream samples are +0.30 to +0.42‰). The accumulation of decomposed vegetation in the forested part of the Neglinka catchment resulted in an increase in DOC concentrations in local run-off. Run-off from the forested part of the catchment has about twice the DOC concentrations (29–31 mg L⁻¹) of the downstream section of the river (5–20 mg L⁻¹). Overall, samples with the lowest δ⁵³Cr values (upstream) have higher Cr and DOC concentrations, and lower pH values as compared to the downstream water (cf. D’Arcy et al. 2016). The observed change in δ⁵³Cr values from on average +0.4‰ for the upstream to on average +1.5‰ for the downstream seems to be consistent with the reduction of Cr(VI) to Cr(III) in the aqueous environment. Therefore, the second scenario explaining the origin and evolution of aqueous Cr in the Neglinka River seems to be more realistic.

In spite of the obvious breakthrough in data collection for the Neglinka River waters, the newly obtained data are still limited for comprehensively answering the question about the influence of underground discharge and urban pollution on the chemical balance of the Neglinka River waters. In order to fulfil this gap in our knowledge, it would be necessary to extend the existing dataset with new observations, including trace metals, and traditional and selected non-traditional stable isotopes. Nevertheless, we were able to make some conclusions on the basis of the collected non-isotopic and isotopic data.

5 Conclusions

We suggest that the river hydrologically known as Neglinka, consists hydrochemically of two different streams: one represented by the upper section of the river, and the other by the lower section. The upstream section was characterized by overall low water mineralization and low values of physico-chemical parameters (pH, electrical conductivity), whereas the downstream section displayed higher water mineralization and higher physico-chemical parameters. Second, the overall strong increase in water mineralization in the downstream section of the river could be mostly due to the input of underground discharge (mineralization of groundwater in the region is usually several times higher than that of the surface water). And third, we can make a preliminary suggestion that Cr(VI) leached from a source in the uppermost part of the Neglinka River Basin may have been partly reduced to insoluble Cr(III) form, starting after mobilization to water and then further during the riverine transport. Precipitation of Cr(III) from water during riverine transport resulted in a decrease in Cr concentration and an increase of δ⁵³Cr values of run-off. Overall low concentrations of Cr in the water samples studied were likely due to low concentrations of Cr in the catchment bedrock. The Neglinka River waters contained Cr(VI) as the main-dissolved Cr species, with an isotopic composition enriched in the heavier ⁵³Cr isotope. Reduction during riverine transport resulted in the observed difference in δ⁵³Cr values between upstream and downstream samples (+0.30 to +0.42‰ for the upstream vs +0.92 to +1.73‰ for the downstream). Detailed study of groundwater released to the surface in springs along the Neglinka stream may help to estimate more precisely the influence of underground discharge on the elemental and isotopic balance of the Neglinka River.

Acknowledgments

Two anonymous reviewers are thanked for critical reading the earlier version of the manuscript, and for their invaluable comments and advice.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by the Czech Geological Survey [Internal Project CGS 331600 to AVA]; Russian Science Foundation [Grant No 18-17-00176 to DAS and GSB].