**Impact of Permafrost Thaw and Climate Warming on Riverine Export Fluxes of Carbon, Nutrients and Metals in Western Siberia**

**Oleg S. Pokrovsky 1,2,3,\*, Rinat M. Manasypov <sup>1</sup> , Sergey G. Kopysov 4, Ivan V. Krickov 1, Liudmila S. Shirokova 2,3, Sergey V. Loiko 1, Artem G. Lim 1, Larisa G. Kolesnichenko 1, Sergey N. Vorobyev <sup>1</sup> and Sergey N. Kirpotin <sup>1</sup>**


Received: 30 April 2020; Accepted: 22 June 2020; Published: 24 June 2020

**Abstract:** The assessment of riverine fluxes of carbon, nutrients, and metals in surface waters of permafrost-affected regions is crucially important for constraining adequate models of ecosystem functioning under various climate change scenarios. In this regard, the largest permafrost peatland territory on the Earth, the Western Siberian Lowland (WSL) presents a unique opportunity of studying possible future changes in biogeochemical cycles because it lies within a south–north gradient of climate, vegetation, and permafrost that ranges from the permafrost-free boreal to the Arctic tundra with continuous permafrost at otherwise similar relief and bedrocks. By applying a "substituting space for time" scenario, the WSL south-north gradient may serve as a model for future changes due to permafrost boundary shift and climate warming. Here we measured export fluxes (yields) of dissolved organic carbon (DOC), major cations, macro- and micro- nutrients, and trace elements in 32 rivers, draining the WSL across a latitudinal transect from the permafrost-free to the continuous permafrost zone. We aimed at quantifying the impact of climate warming (water temperature rise and permafrost boundary shift) on DOC, nutrient and metal in rivers using a "substituting space for time" approach. We demonstrate that, contrary to common expectations, the climate warming and permafrost thaw in the WSL will likely decrease the riverine export of organic C and many elements. Based on the latitudinal pattern of riverine export, in the case of a northward shift in the permafrost zones, the DOC, P, N, Si, Fe, divalent heavy metals, trivalent and tetravalent hydrolysates are likely to decrease the yields by a factor of 2–5. The DIC, Ca, SO4, Sr, Ba, Mo, and U are likely to increase their yields by a factor of 2–3. Moreover, B, Li, K, Rb, Cs, N-NO3, Mg, Zn, As, Sb, Rb, and Cs may be weakly affected by the permafrost boundary migration (change of yield by a factor of 1.5 to 2.0). We conclude that modeling of C and element cycle in the Arctic and subarctic should be region-specific and that neglecting huge areas of permafrost peatlands might produce sizeable bias in our predictions of climate change impact.

**Keywords:** river flux; weathering; organic matter; permafrost; trace element; river

#### **1. Introduction**

Arctic warming is anticipated to result in massive carbon (C) mobilization from permafrost peat to atmosphere, rivers and lakes, thereby potentially worsening global warming via greenhouse gases (GHG) emissions [1,2]. Permafrost peatlands cover roughly 2.8 million km<sup>2</sup> or 14% of permafrost<sup>a</sup>ffected areas, mostly in Northern Eurasia (Bolshezemelskaya Tundra, 0.25 <sup>×</sup> <sup>10</sup><sup>6</sup> km2; western Siberia, 1.05 <sup>×</sup> 106 km2; Northern Siberia and Eastern Siberia lowlands, 0.5 <sup>×</sup> 106 km2) and contain a huge amount of highly vulnerable carbon in soil and surface waters [3]. Except for several regional studies of peatland lakes and small streams in Canada [4–6] and northern Sweden [7], the control factors, timing and reality of C and related elements release from soil and sediments are largely unknown, in part because element export by rivers across the permafrost peatlands is still poorly quantified.

One of the largest peatlands in the world is the Western Siberia Lowland (WSL) which exhibits a disproportionally high contribution to C storage and exchange fluxes [8] and presents a prominent exception to well-studied aquatic and soil ecosystems in mountainous territories of Northern America and Scandinavia. Specific features of the WSL are: (1) developed on flat, low runoff terrain with long water residence time in both lentic and lotic systems; (2) acts as important C stock in the form of organic-rich histosols (peat soils); (3) emits substantial amount of CO2 and CH4 to the atmosphere from the surface of inland waters; and (4) exhibits high dissolved organic carbon (DOC) concentration and low pH in waters in contact with peat soils. These factors determine rather unique and still poorly known aquatic communities in dystrophic to mesotrophic peatland waters. Due to the lack of nutrients, shallow photic layer and high sensitivity to water warming, aquatic ecosystems of frozen peatlands are highly vulnerable to the permafrost thaw and can respond unexpectedly to ongoing climate warming in terms in their C, nutrient, and metal storage and export fluxes as well biodiversity and organisms adaptation strategies.

Western Siberia is a key region for biogeochemical studies in the Arctic [2] as this region includes the continuous-discontinuous permafrost transition and experiences substantial thermokarst. As a result, this high-priority region is most susceptible to thaw-induced change in solute transport by rivers and its export to the ocean and atmosphere, notably via activation of deep underground flow [9] and exposure of large volumes of previously frozen peat and mineral soils [10–12]. Raised by incontestable Arctic amplification of overall climate change, the fate of carbon, nutrient and metal in Arctic rivers is at the forefront of field and modeling studies [13–20]. In particular, the Arctic Great Rivers Observatory (GRO) program assessed concentrations and export fluxes of 6 largest Arctic Rivers including the Ob River, draining sizeable part of the WSL [21,22]. However, widespread loss of hydrological monitoring in the beginning of 21st century was especially pronounced for small and medium-size rivers of the permafrost-affected part of the WSL [23]. At the same time, the Ob River is not suitable for modeling of possible changes in western Siberia because (1) it is strongly influenced by its largest permafrost-free tributary, the Irtysh River; and (2) it cannot be used for assessing the fluxes of northern (permafrost) zones as it integrates vast territory of variable permafrost coverage (20% in average) and landscape parameters. For these reasons, the study of small rivers at the WSL territory is more suitable for assessing both mechanisms of flux formation and its possible future changes.

Currently, a dominant paradigm is that riverine fluxes of C and inorganic nutrients are increasing in virtually all permafrost-affected rivers [14,24,25]. In particular, the increase in suspended versus dissolved transport of elements may be due to abrasion of lake shores and riverbanks. At the same time, enhanced groundwater input will lead to an increase in the transport of truly dissolved forms. Recently, following the pioneering work of Frey et al. [26–28], the concentrations of dissolved, particulate and colloidal carbon, nutrient and metals in WSL rivers, lakes and soil waters have been studied over a sizable latitudinal gradient [29–38]. These results allowed a first-order assessment of the consequences of climate warming and permafrost thaw on river water concentrations of dissolved and particulate forms of elements from western Siberia to the Arctic Ocean and C emission to the atmosphere. For this, a "substituting space for time" approach was employed, which postulates, in a broad context, that spatial phenomena which are observed today can be used to describe past and future events [39].

Thus, a concentration pattern of major and trace elements (TE) in WSL river suspended matter implies that, upon a progressive shift of the permafrost boundary northward, there will be a sizeable decrease in concentrations of alkalis, alkaline-earths, divalent heavy metals, and trivalent and tetravalent hydrolysates in northern rivers in currently discontinuous and continuous permafrost zones [31]. This decrease may be by a factor of 2–5 from the position of the minimum elemental concentration in sporadic to discontinuous permafrost zones relative to the continuous permafrost zone. Concerning the dissolved element concentration and potential transport in the WSL rivers, the following results were achieved implying the substituting space for time scenario. From the permafrost thaw perspective, the increase in depth of the active layer and connectivity of a river with underground water reservoirs may decrease the colloidal fraction (1 kDa–0.45 μm) of OC, Fe, Al and number of divalent metals as well as low-soluble trivalent and tetravalent hydrolysates in the sporadic and isolated permafrost zone as it becomes permafrost-free [30]. The forestation of wetlands and lake drainage may slightly diminish colloidal transport of DOC and metals at the expense of low molecular weight forms. Overall, given the significant role of seasonal and forestation effects on colloidal forms of OC and TE in WSL rivers, major changes in the speciation of riverine C and nutrients in the WSL may occur due to changes in vegetation rather than in temperature and precipitation [30]. A northward permafrost boundary shift with increase in air and water temperature may decrease or maintain, rather than increase of major nutrient (K, P, N, Si) and DOC concentrations in rivers draining through continuous permafrost zone [37].

All these predictions described above were made based solely on evolution of concentrations of soluble and particulate C, metals and nutrients in WSL rivers, without taking into account the hydrological flux. The latter could not be addressed until now due to lack of reliable information on river discharge over different seasons. The present work aims to fill this gap by quantifying the export fluxes of ~30 WSL rivers of various size, combined with new hydrological modeling of region and season-specific river runoff of the WSL territory. The chosen rivers encompass a large gradient of climate, vegetation and permafrost distribution and thus enables a substituting space for time scenario to provide a tentative prediction of possible changes in riverine fluxes for short- and long-term prospectives. The present work is built on seasonally resolved sampling performed in 2015 and 2016, and incorporates thorough hydrological modeling to calculate the seasonal discharge. Building upon our previous studies of the WSL rivers [29–31,33,34,37], here we assess for the first time, element export fluxes (yields) across a large permafrost-affected territory and large number of rivers. The obtained elementary yields are essential for further modeling of biogeochemical cycles in the permafrost regions.

#### **2. Study Site and Methods**

#### *2.1. Rivers of Western Siberian Lowland, their Sampling and Analyses*

The 32 rivers of the Ob, Pur and Taz watersheds in the WSL were sampled (Figure 1). Detailed climatic, lithological and physio-geographic characteristics for the WSL are presented elsewhere [29,33,34]. The dominant lithology is Quaternary deposits (silt, clay, sand) overlaid by peat. The climate gradient of sampled rivers presents a decrease in mean annual air temperature (MAAT) from −0.5 ◦C in the south (Tomsk region) to −9.5 ◦C in the north (Arctic coast). Annual precipitation is fairly constant ranging between 550 mm in the south and 600 mm at the lower reaches of the Taz River. The river runoff ranges from 190 <sup>±</sup> 30 mm y−<sup>1</sup> in the south to 300 <sup>±</sup> 20 mm y−<sup>1</sup> in the north [40]. The distribution of the permafrost reflects the south-north MAAT gradient and changes from isolated and sporadic in the south to discontinuous and continuous in the north. In 2015, we sampled rivers in spring (18 May–25 June) and summer (25 July–21 August). In 2016, sampling was performed in spring (17 May–15 June 2016), summer (1–29 August 2016), and autumn (24 September–13 October 2016). We followed the change of seasons during our sampling campaign and moved from the south to the north in spring and from the north to the south in autumn thus collecting the river water at approximately the same period of the discharge. Note that more frequent sampling would be desirable to accurately evaluate the annual export flux, but rather harsh environment and logistical issues

constrained sampling. In contrast, route sampling is a common way to assess chemical weathering in extreme environments [41,42], and it is accepted that single sampling during high flow season provides the best agreement with time-series estimates [43].

**Figure 1.** The WSL river sampling points. The numbers on the map correspond to the following rivers: (1) Ob' (Pobeda), (2) Maliy Tatosh, (3) Chaya, (4) Vyalovka, (5) Ob' (Aleksandrovskoye), (6) Vakh, (7) Agan, (8) Tromyegan. (9) Vach-Yagun, (10) Vachinguriyagun, (11) Pintyr'yagun, (12) Kamgayakha, (13) Khatytayakha, (14) Pyakupur, (15) Lymbyd'yakha, (16) Apoku-Yakha, (17) Etu-Yakha, (18) Seryareyakha, (19) Purpe, (20) Aivasedapur, (21) Tydylyakha, (22) Yamsovey, (23) Pur, (24) Ngarka Khadyta-Yakha, (25) M. Kheyakha, (26) Nuny-Yakha, (27) Taz. The upper left insert is from Google Maps®.

The two sampled years contrasted in mean monthly and annual temperature and precipitation. The summer 2015 was warm (+0.3 ◦C above normal for July), while summer 2016 was cool (−3.1◦ below normal for July). The summer 2015 was wet (180% of normal precipitation for July), and the summer 2016 was less wet (but still 130% of normal precipitation for July). The winter precipitations of 2014–2015 and 2015–2016 were rather similar and sizably higher (ca. 140%) than the normal winter values. The normal values here are defined as between 1970 to 2000 based on the Roshydromet archives [44].

The surface water was collected in a polypropylene 1-L container. Samples were filtered through 0.45 μm cellulose acetate filters using a Nalgene 250 mL filter unit combined with a vacuum pump. All filtrations were run on site, in a protected environment, within 2 h of river water collection. Immediately after filtration, samples for DOC, DIC, major and trace elements were stored in the refrigerator during 1–2 months prior to the analyses, while the samples for nutrients were kept frozen. Dissolved (<0.45 μm) concentrations of nutrients, major and trace elements in WSL river waters were analyzed as described elsewhere [33,34,37,45]. Major anion concentrations (Cl<sup>−</sup> and SO4 <sup>2</sup>−) were measured by ion chromatography (HPLC, Dionex ICS 2000) with an uncertainty of 2%. DOC and DIC were analyzed using a Shimadzu TOC 6000 with an uncertainty of 3–5% [46]. For all major and most trace elements, analyzed by ICP MS, the concentrations in the blanks were below analytical detection limits (≤0.1–1 ng/L for Cd, Ba, Y, Zr, Nb, rare earth elements (REEs), Hf, Pb, Th, U; 1 ng/L for Ga, Ge, Rb, Sr, Sb; ~10 ng/L for Ti, V, total P (Ptot), Cr, Mn, Fe, Co, Ni, Cu, Zn, As). The international certified reference material SLRS-5 (Riverine Water Reference Material for Trace Metals) was used to validate the analysis. Further details of analytical uncertainties and detection limits for TE are provided elsewhere [30,34,45].

#### *2.2. Hydrological Parameters, River Discharge Modeling, and Element Export Flux Calculations*

The runoff was longitudinally modelled using average altitude of the watershed for gauged rivers [47]. For small and medium rivers draining palsa and polygonal bogs of the permafrost zone, we used empirical equations accounting for hydrological characteristics of these watersheds [48]. In this work, only open water period (May to October) was considered. The contribution of winter time export in large rivers typically does not exceed 10% of mean annual runoff [49]; moreover, small (<1000–10,000 km2 watershed) WSL rivers freeze solid in winter [33]. We interpolated discharge by using watershed area change along the river mainstem. The river hydrographs were modelled using HBV Light program package (https://www.geo.uzh.ch/en/units/h2k/Services/HBV-Model.html) as described in refs [50–52]. For the gauged rivers, we modelled daily runoff based using mean water temperature and precipitation from adjacent meteostations. The data for reproducing the water discharge at key sites were taken from hydrological yearbooks and automatic information system of State monitoring of water bodies [53]. The daily air temperature and precipitation were taken from Russian Hydrometeocenter (URL: http://aisori.meteo.ru/ClimateR; [44]). The calibration of model parameters demonstrated that the HBV model adequately reproduces the seasonal dynamics of runoff (quality criteria of 0.75 to 0.90). The uncertainties of modelled discharge at ungauged rivers stem from several factors. First, there is uncertainties in characterizing watershed boundaries for the very flat, but poorly resolved, WSL territory. For small rivers, this uncertainty can amount to 30% due to lack of precise topographic information. For large rivers, it can be up to 7% due to difficulties in determination of exact position of the watershed divide on flat bog-lake landscapes. The second cause of uncertainty is a discrepancy between the model parameters of analogous rivers due to intrinsic differences in the conditions of runoff formation (up to 25%). Finally, the amount of precipitation obtained from the nearest meteostations might substantially differ from the actual precipitation at a given watershed, which is reflected, in particular, by low quality of the rain-flood modeling. The maximal overall uncertainty of daily discharges is determined by the most probable value of 30%. However, because we used two-month averaged discharges for calculating elementary yields, the real uncertainty of export fluxes is determined by the uncertainty of watershed delineations and ranges from 7 to 30%.

In order to calculate element export fluxes, we defined six latitudinal belts (56–58◦ N, 58–60◦ N, 60–62◦ N, 62–64◦ N, 64–66◦ N and 66–68◦ N). These latitude belts were selected based on: i) the permafrost map of the WSL, where the permafrost zones roughly follow latitude, and ii) necessity to integrate sufficient number of rivers in each permafrost zone and for statistical comparison. For statistical comparisons, we separated the years 2015 and 2016, because the autumn was sampled only in 2016. The number of rivers used for the latitudinal-average flux calculation ranged from 2 to 10 in 2016 and from 2 to 16 in 2015 for each latitudinal range.

#### **3. Results**

#### *3.1. Impact of the Watershed Size and Season on Element Export Fluxes*

The yields of DOC, representative major solutes (DIC, Ca, Mg) and nutrients (Si, Ptot, K, Fe) are illustrated in Supplementary Figure S1. Flux magnitudes exhibited sizable variation, ranging over one (DOC, Si and Mg) to two (DIC, Ca, K, Fe and Ptot) orders of magnitude. Generally, the variations were the highest for rivers having watersheds between 100 and 1000 km2, for both permafrost-affected and permafrost-free. The river watershed area (Swatershed) exhibited rather weak control on elementary yields, as also illustrated by statistical treatment of individual seasons and full data set for both years (Supplementary Table S1). Overall, the watershed size exhibited more pronounced correlations with element yield in the permafrost-free zone compared to the permafrost zone. The correlations were quite low during spring flood but become pronounced during summer and autumn baseflow. The highest correlations were observed in summer, when Cl, SO4, Fe, Cu, Y, Mo, Sb, REEs and Th were positively linked to Swatershed (RSpearman > 0.58, p < 0.01) in the permafrost-free zone.

Partial contribution of spring, summer, and autumn (2 months each) to the overall export of elements during the six-month open-water period (May to October) in 2016 demonstrated the dominant role of spring for DOC, Al and Fe (>60%), for both permafrost-free and permafrost-affected rivers (Figure 2). In contrast, soluble highly mobile elements (DIC, Na, K, Ca, Mg, Si) had less than 40% and 50% contribution of spring in the permafrost-free and permafrost zone, respectively, and a 2–3 times higher contribution during summer baseflow period compared to during the spring flood period. Nutrients (e.g., K, Ptot) and metals (e.g., Mn) presented an intermediate case with a half of yield occurring during the spring flood period and with negligible (<10%) role of autumn period in permafrost-free zone but comparable contribution of summer and autumn in the permafrost-affected part of the WSL.

Element export as a function of watershed area (Supplementary Figure S1) and season (Supplementary Figure S2) represent two main groups, namely (1) soluble highly mobile elements (Cl, SO4, Sr, Rb, Ba and As, Sb, Mo, U (permafrost-free zone only) and (2) DOC and low soluble elements, typically present in the form of organic- and organo-mineral colloids especially in the permafrost zone [30], such as micronutrients (Fe, Co, Ni, Cu, Zn, V), toxic metals (Cr, Cd, Pb), Nb, and trivalent (Al, Ga, Y, REE) and tetravalent (Ti, Hf, Th) hydrolysates. Generally, the share of spring flood contribution in overall open-water export of elements was 20 ± 10% higher in permafrost-affected rivers than in rivers of the permafrost-free zone.

**Permafrost-free**

**Figure 2.** Partial contribution of spring, summer and autumn 2016 to overall open-water period export of elements by WSL rivers, located in the permafrost-free (**A**) and permafrost-affected (**B**) regions.

#### *3.2. Riverine Element Export Fluxes Across the Latitudinal Profile (Permafrost and Climate Gradient)*

Because both concentrations [33,34,37] and seasonal fluxes (Supplementary Figure S1 and Table S1) of elements were not strongly impacted by the river size, the watershed-averaged fluxes can be calculated as a function of the watershed latitude. Furthermore, we used ternary molar diagrams (Ca − Mg − (Na + K) and Cl − SO4 − HCO3) to reveal the role of season, river size and permafrost coverage of major elementary composition (Supplementary Figure S3), following traditional geochemical classifications [54]. Regardless of the season, water chemical composition from permafrost-free rivers was distinctly different from water from permafrost-affected rivers and strongly enriched in Ca and HCO3 −. As a result, the latitude (which corresponds with the permafrost zonation) can be considered as the main factor controlling major cation concentrations in the WSL rivers. The average (±2 s.d.) fluxes of dissolved (<0.45 μm) major element export from the WSL watersheds for each latitudinal belt in 2016 and 2015 are shown in Figure 3 and Supplementary Figure S4, respectively. For the sake of scientific rigor, we illustrate the yields separately for 2015 and 2016, but the two-year average values of element export during spring and summer are given in Table 1.

**Figure 3.** Yields (watershed-area normalized export fluxes) of DOC (**A**), DIC (**B**), Ca (**C**), Mg (**D**), K (**E**), Si (**F**), Ptot (**G**) and Fe (**H**) during May-October 2016 in 33 WSL rivers across the latitudinal gradient. The vertical uncertainties represent the 2 s.d. of several rivers belonging to the same latitudinal belt. Thick horizontal bars represent the 2◦ latitudinal belts. The year is represented by open-water period (May to October), neglecting winter (November to April), when small rivers freeze solid in the north.


**Table 1.** Mean (±SD) export fluxes of major and trace elements by WSL rivers in the permafrost-free and the permafrost zone.


**Table 1.** *Cont.*

Although the maximal uncertainties on discharge do not exceed 30% (see Section 2.2), the regional element yield assessments are subjected to high variation among rivers (Figure S1) with s.d. often at 50% (see Figure 3, Supplementary Figure S4). Among possible causes of these sizable uncertainties are non-linear and hysteretic relationships between concentrations and discharge which could not be resolved due to low frequency of sampling. Another reason of this variability could be highly dynamic behavior of element concentration during summer and autumn baseflow and spring flood, reflecting source limitation, chemostatis, or transport limitation (i.e., refs. [55–57]).

The DOC yield was minimal in rivers south to 59◦N and exhibited a maximum at 61–65◦ N and 63–65◦ N in 2015 and 2016, respectively, with overall magnitude of variation by a factor of 4 (Figure S4A and Figure 3A, respectively). The yields of DIC and Ca decreased more than five-fold from south to north and achieved minimal values in the 63–65◦ N belt, for both years of observation (Figure 3B,C and Supplementary Figure S4B,C). Magnesium showed a weak minimum of yield in the 63–65◦ N which was however pronounced only in 2016; the overall variations were less than a factor of 2 (Figure 3D and Supplementary Figure S4D). Potassium yield remained fairly constant across the latitudinal profile with overall variations less than a factor of 1.5 to 2.0 between various latitudinal belts (Figure 3E and Supplementary Figure S4E). Silicon showed a three-fold increase in export fluxes from the south to the north, quite similar for two years of monitoring (Figure 3F and Supplementary Figure S4F).

The latitudinal pattern of other major and trace element export fluxes followed these three main types of behavior described above: (1) minimal in the south and a northward increase by a factor of 3 to 5 (Si, POC, P-PO4; N-NH4, Ni, REEs, Zr, Nb, Hf, and Th), with a local maximum at 63–65◦ N (DOC, Al, Ti, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, Rb, Cd, Cs, and Pb); (2) northward decrease by a factor of 3 to 5 (DIC, Ca, SO4, Sr, Mo, W, and U), and (3) overall independence of yield on latitude, or the latitudinal variations were less than a factor of 2 (Cl, N-NO3, Li, B, Mg, Ptot, K, As, Zr, Sb, and Ba). Note that some elements (Mg, Zn, As, Rb) may belong to one or another group depending on the year of observation.

The yields of all 51 dissolved elements are listed in Table 1, which presents the export fluxes of two-year average (May to August) and open-water period (May to October 2016) in permafrost-free and permafrost-affected zones. The ratio of element flux in the non-permafrost zone to that in the permafrost zone (Rabsent/permafrost, Figure 4) demonstrates a distinct order of elements whose export occurs preferentially in southern or northern part of the WSL. There are three main groups of elements - those exhibiting the highest (a factor of 2 to 10) yield in the permafrost-free zone (DIC, SO4, Ca, Mo and U) and those showing a maximum in the permafrost-affected zone (As, Ptot, Li, B, Rb, Na, Si with 0.5 ≤ Rabsent/permafrost ≤0.25, and Ni, Th, Cd, DOC, Cs, Nb, Ti, Al, Fe, Co, Cr, Mn, Hf, Co, and REEs with Rabsent/permafrost <0.25). The elements of an intermediate group showed comparable (±30%) yields in permafrost-free and permafrost-affected parts of the WSL (Sr, Mg, K, Ba, Sb, Zr, Cu).

**Figure 4.** A histogram of the ratio of element average yield in the permafrost-free rivers to that in the permafrost-affected rivers for 6 open-water months (May to October) of 2016 (based on the data listed in Table 1).

#### **4. Discussion**

#### *4.1. E*ff*ect of Seasons, Watershed Size, Latitude and Permafrost Coverage on Elementary Yields*

There are several distinct groups of elements defined according to the latitudinal patterns of their yields. These groups reflect the relative mobility of elements, consistent with general knowledge of river hydrochemistry in the high latitude regions [58]. The DOC, organically-bound metals (V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Cd and Pb) and many low soluble TE–geochemical traces (Al, REEs, Nb, Ti, Zr, Hf, and Th) exhibited similar 1st-type latitudinal pattern (minimum in the south and maximum in the permafrost-affected zone). The overall open-water period transport of these elements was dominated by spring flood, as it was also observed in other boreal and subarctic rivers (i.e., the Severnaya Dvina River, [59]). The nutrients (K, Si, P-PO4, Ptot, and N-NH4 <sup>+</sup>) and TE—analogous of macronutrients (Rb)—exhibited 50 ± 10% share of annual export during spring flood and a northward increase. Finally, the yield of geochemically mobile elements (DIC, Cl−, SO4, Li, B, Mg, Ca, Sr, Mo, and U) decreased northward, with less than 40–50% of overall contribution provided by spring flood period.

The division of elements into these groups depends on (1) the share of different seasons in overall 6-month open water period export, (2) the shape of the latitudinal pattern (Figure 3 and Figure S4), and (3) the difference in total 2-year averaged riverine yields between permafrost-free and permafrost-affected zones (Table 1, Figure 4). Such a distinction is consistent with two main factors controlling the element export (transport) in the WSL rivers. First, this is source-limitation, when the input of elements from soils to the river is controlled by river connectivity to deep and shallow groundwater reservoirs [27]. Due to the presence of carbonate mineral concretions in clay-silt bedrocks, especially in the south of the WSL [33], the groundwater is enriched in soluble elements such as alkalis and alkaline-earth metals, oxyanions and U(VI). The second is transport-limitation, when the export of an element is controlled by the availability of its carrier such as organic and organo-ferric, organo-aluminum colloids [30,34,45]. This factor reflects the superposition of surface source (topsoil, vegetation) providing organic colloids, and deep soil (mineral horizons), together with Fe(II)-rich groundwaters, providing soluble TE. In addition to these two main groups of elements, the mineral nutrients are not limited by transport and exhibit quite complex pattern which reflects their deep (groundwater, bedrock lithology) and surface (plant litter, atmospheric deposition) sources. These elements are strongly impacted by their biotic uptake in the river channel or seasonal release from decaying aquatic macrophytes, plankton and periphyton [37,60]. Various internal factors, operating in soils and riparian zone of the river, are capable of modifying the export of nutrients. For example,

low export fluxes of both total P (phosphate and organic P) and P-PO4 in southern (56–60◦ N) latitudinal belt compared to northern, permafrost-affected rivers (see Figure 3G, Figure S4G) may be due P retention via adsorption onto and coprecipitation with Al, Fe hydroxides in the deeper part of soil profile [61,62] as well as P removal in the form of Ca phosphate minerals [63] occurring in the unfrozen mineral soils, exposed to surface fluids. Furthermore, P uptake by abundant terrestrial and aquatic vegetation is most pronounced in the south. In the north, the mineral soils are essentially frozen, and the requirements of aquatic and terrestrial biota for this nutrient are much lower [64].

The effect of watershed size on DOC, DIC, cations, and Si export fluxes was rather minor. This is at odds with strong river size control on element yields as it is known in temperate and mountainous regions [65,66]. In particular, small catchments of wetlands exhibit generally lower runoff than the medium and large rivers, and the runoff is one of the major controlling factors of chemical weathering and element export [46,67–70]. Season also played a secondary role in determining element yield pattern. However, sizable correlations of element fluxes with Swatershed in summer and autumn, observed solely in the permafrost-free zone (Supplementary Table S1) can be explained by more pronounced impact of deep underground waters in large rivers compared to small ones. These waters typically contain a high concentration of soluble, labile elements (e.g., Cl, SO4, Na, Mo, As) [33,34] but also Fe(II), those oxidation in the riparian and hyporheic zones create large amount of organo-ferric colloids [45] capable to provide enhanced concentrations of typically insoluble low mobile elements such as trivalent and tetravalent hydrolysates.

The latitude was revealed to be the primary governing parameter of elementary yields and clearly marked the difference between permafrost-free, discontinuous and continuous permafrost regions. The most northern regions of the WSL exhibited rather high yields of DOC, DIC, Si and cations. It is possible that in continuous permafrost zone of frozen peat bogs, the underlining mineral layer is protected by the permafrost. As a result, the active (seasonally unfrozen) layer is located within the organic layer which comprises live vegetation, plant litter, and upper peat layer. This organic matrix is extremely reactive, and capable of releasing sizable amount of DOC, P, Si, Ca and nutrients over very short periods of time during contact with surface waters [71,72]. Water temperature (between 4 and 25 ◦C) has only minor effect on C and element release from both thawed and frozen peat [72]. Therefore, even short-term contact of water with surface peat and vegetation is capable mobilizing sizable amount of DOC and nutrients, despite low temperature in the northern regions. In this regard, the elementary export by the WSL rivers is strongly controlled by dynamics of peat formation/decay across the territory. The particularity of the WSL is that, currently, this region is recovering from the last glaciation. As a result, the ecosystems are highly non-stationary: the peat actively accumulates in the south [73], while in the north, the frozen peat is subjected to thawing and degradation [74–76]. The uptake of elements from groundwater, river and forest tree litter by growing peat in southern mires counteracts with DOC and element release from thawing/degrading peat in the northern palsas. The elements affected by these processes are those that exhibit the highest concentration in peat relative to undelaying mineral (silt, sand) horizons. According to a previous assessment of elementary peat composition in the WSL [77], the peat is sizably enriched in C, (V, Cr), trivalent (TE3<sup>+</sup>) and tetravalent (TE4<sup>+</sup>) trace metals (Al, Y, REEs, Ti, Zr, Hf, Th) and U, Zn, Pb and depleted in highly mobile alkalis and alkaline-earths metals, As, and Mo. Therefore, enhanced riverine yield of low-soluble TE3+, TE4+, U, and some divalent metals in the north relative to the south is possibly due to these elements being tightly linked to peatland evolutionary pattern across the WSL. Similarly, depletion of peat relative to underlying mineral horizons in soluble, highly mobile elements is consistent with enhanced export of these elements by southern rivers, where the peat accumulation occurs.

#### *4.2. Comparison of Major Cation, DIC, Si and DOC Export Fluxes in the WSL with Other Boreal Regions*

The average total dissolved cation flux (TDS\_c = Na + K + Ca + Mg) over May-October 2016 ranged from 4.40 <sup>±</sup> 0.55 t km−<sup>2</sup> y−<sup>1</sup> in the permafrost-free zone to 2.64 <sup>±</sup> 0.59 t km−<sup>2</sup> y−<sup>1</sup> in the permafrost-affected zone, which is lower than the fluxes of Central Siberian rivers of the same latitude, draining basaltic rocks (5 to 8 t km−<sup>2</sup> y<sup>−</sup>1, [78]), large Siberian rivers such as Yenisey and Lena (6.2 and 6.8 t km−<sup>2</sup> y−1, respectively, [79]), the Ob River in its middle course (6.0 t km−<sup>2</sup> y−1, [80]), and the permafrost-free Eurasian Arctic rivers draining sedimentary rocks (Sev. Dvina, 9.5 t km−<sup>2</sup> y−1, [59]; Pechora, 6.6 t km−<sup>2</sup> y−1, [79]). The TDS\_c yield of permafrost-affected WSL rivers is, however, comparable with previous estimations of that in the middle-size Siberian rivers (2.8, 2.5, and 2.3 t km−<sup>2</sup> y−<sup>1</sup> for Kolyma, Indigirka and Anabar, respectively [79]). The main reason for relatively low cationic fluxes of small WSL rivers compared to other large rivers of Northern Eurasia of similar runoff are i) low connectivity of WSL rivers with underlying bedrocks and groundwater, due to thick peat layer and permafrost, and ii) essentially weathered character of silicate rocks (sands, clays) in western Siberia. Note also that the transport of major cations in permafrost-free rivers is strongly pronounced during winter (e.g., 35–40% of total annual yield in the Severnaya Dvina River [59]), so it is possible that cationic fluxes of WSL rivers during May-October are somewhat underestimated relative to annual export. At the same time, the majority of small (<4000 km2 watershed area) rivers in the northern part of the WSL freeze solid during the winter [33], so the winter flux may be non-negligible only for southern, permafrost-free rivers.

In contrast to major cations, no difference in the export fluxes between small WSL rivers and large and medium size Eurasian rivers was detected for the DIC. The DIC export of small WSL rivers in the permafrost-free zone (2.2 <sup>±</sup> 0.4 t km−<sup>2</sup> <sup>y</sup><sup>−</sup>1) is in agreement with recent estimations of DIC export in the middle course of the Ob River (2.9 t km−<sup>2</sup> y<sup>−</sup>1, [80]) and with the mean riverine DIC yield of the entire Eurasian Arctic basin (2.2 t km−<sup>2</sup> y<sup>−</sup>1, [79]). However, DIC export by permafrost-affected WSL rivers is somewhat lower (1–2 t km−<sup>2</sup> y−1) and comparable to medium size rivers of Central and Eastern Siberia (0.6–2.2 t km−<sup>2</sup> y<sup>−</sup>1, [79]). A tentative explanation of elevated DIC (but not TDS\_c) flux in the WSL rivers is CO2 and HCO3 − generation by microbial (and photolytic) processing of peat soil organic carbon (both DOC and POC), which is the main cause of very high CO2 emission from WSL inland waters [81,82]. The light isotopic composition of DIC in the WSL rivers (−<sup>25</sup> <sup>≤</sup> <sup>δ</sup>13CDIC<sup>&</sup>lt; <sup>−</sup><sup>10</sup> ‰, [33]) is consistent with this hypothesis.

The riverine Si fluxes in the southern part (<61◦ N) of theWSL (0.5–1.0 t km−<sup>2</sup> y<sup>−</sup>1) are comparable to those of the Ob River (0.62 t km−<sup>2</sup> y<sup>−</sup>1, [80] and small rivers of the northern Sweden (0.9 t km−<sup>2</sup> y<sup>−</sup>1, [83]). In contrast, the fluxes of the permafrost-affected rivers (1.0–1.5 t km−<sup>2</sup> y<sup>−</sup>1, Figure 3F and Figure S4F and Table 1) contradict the expected trend of decreasing flux with the increasing latitude and decreasing temperature, given that the chemical weathering of silicate rocks is much slower in colder climates [67–70]. In fact, we noted that the northern fluxes of WSL are comparable with those of the temperate rivers such as Mississippi and Yangtze [84]. Moreover, if the silicate rock weathering significantly controls element delivery from the soil to the river, such a northward increase would occur for cations (Ca, Mg, Na) as it is known for typical silicate terrains of the boreal zone [85], but this is not observed in the WSL territory. As such, we hypothesize that two- to three-fold increase in Si yield of northern rivers relatively to southern rivers is linked to combination of (1) enhanced mobilization from plant litter via suprapermafrost flow in the permafrost zone, (2) limited silicate secondary mineral formation in shallow, essentially frozen northern soils compared to southern soils, and (3) strong uptake of Si by both terrestrial plants and aquatic macrophytes and periphyton in the southern rivers.

The DOC fluxes in the permafrost-affected regions of the WSL territory were quite high (2 to 6 t km−<sup>2</sup> y−1) compared to the middle and lower reaches of the Ob River (0.64 and 1.2 t km−<sup>2</sup> y−1, respectively, [79,80]). At the same time, these fluxes are comparable with those in large boreal river draining permafrost-free wetlands (the Severnaya Dvina River, 4.2 <sup>±</sup> 0.8 t Corg km−<sup>2</sup> <sup>y</sup><sup>−</sup>1, [59]) and are the highest among all known rivers flowing to the Arctic Ocean. Indeed, the Ob, Yukon, Lena, Yenisey, and Mackenzie rivers exhibit a DOC yield from 0.5 to 2.5 t km−<sup>2</sup> y−<sup>1</sup> ([79,86]). We suggest that the main factor responsible for such high DOC yield in small WSL rivers is high proportion of peatlands on their watershed (i.e., typically from 40 to 60%, according to the GIS data [34]). The peatland-draining Taz (watershed = 150,000 km2), Pur (112,000 km2), and Nadym (64,000 km2) rivers, located entirely in the discontinuous permafrost zone, also have a DOC yield of 1.9, 2.1, and 4.4 t Corg km−<sup>2</sup> y−1, respectively ([79] and calculations based on data from the RHS).

The northward increase in DOC flux possibly reflects strong leaching of OM from the plant litter and organic-rich topsoil (Histosol). In the north, the adsorption of DOM on underlying mineral horizons is minimal because these horizons are frozen. As a result, the riverine DOC export in the permafrost zone of the WSL is controlled by water travel time through the peat layer and underlying mineral horizons and the water residence time necessary for DOC leaching from upper vegetation layer (moss, lichen, litter). However, quantitative modeling of DOM and element reactive transport in the WSL peatlands, on the scale of a small watershed or large river, is beyond the scope of this study.

#### *4.3. Possible Evolution of Western Siberia Rivers Elementary Yields Under Climate Change Scenario*

The space for time approach employed in the present work provides some future projections for riverine element behavior. However, it exhibits a number of shortcomings whose analysis goes beyond the scope of this work. In particular, this approach does not address the time scale, necessary for the northern ecosystem to reach the new "more southern" state; it ignores possible shift in the structure of vegetation and soil microbial community that can indirectly impact the carbon and nutrient, in terms of landscape storage and removal via rivers, and it does not include information on the altered seasonality such as extended hydrologic seasons, earlier snowmelt, higher precipitation that will likely occur as a result of climate change. Nevertheless, as a first order empirical assessment, the following predictions can be made. The first consequence of climate warming in western Siberia is thawing of frozen peat and underlying mineral horizons. The thickness of the active layer (ALT) is projected to increase more than 30% during this century across the tundra area in the Northern Hemisphere [87–89]. In the WSL, this increase will be most dramatic in the north, where the peat deposits are thinner than those in the discontinuous permafrost zone [48,90–92]. The main consequences of the ALT increase may be the involvement of mineral (clay) horizons into water infiltration within the soil profile [90]. As a result, the DOC originated from the leaching of the upper peat layers and plant litter degradation will be retained on mineral surfaces via adsorption onto and incorporation into clay interlayers [93–97]. For inorganic solutes, the effect of ALT increase will be lower than that of DOC, given much lower affinity of HCO3, cations and Si to clay surfaces and the lack of unweathered (primary) silicate rocks underneath the peat soil column. However, if the thawing will open new water paths between deep groundwater reservoirs and the river, this may increase the riverine export of major cations and DIC [9,33].

The second consequence of climate change in western Siberia is a shift of the permafrost zone boundaries further north [98–100]. Within the substituting space for time scenario, such a permafrost boundary change can be considered equivalent to the northward shift of the river latitudes as shown in Figure 3. Based on latitudinal pattern of major and TE (as illustrated in Figure 3 and Figure S4), the following groups of the river water components likely to change their riverine yield over open-water period in case of anticipated shift in the permafrost zones: (1) Elements those yields likely to increase by a factor of 2 to 3: DIC, Ca, SO4, Sr, Ba, Mo, U; (2) Elements likely to decrease the yields by a factor of 2 to 5: DOC, Fe, Si, Ptot, P-PO4, N-NH4, divalent heavy metals, trivalent and tetravalent hydrolysates, and (3) Elements weakly affected by the permafrost boundary change and the ALT increase (change of yield by a factor of 1.5 to 2.0): B, Li, K, Rb, Cs, N-NO3, Mg, Zn, As, Sb, Rb, Cs).

The impact of climate warming on riverine fluxes in the WSL is not restricted to the shift of permafrost zones and the change of water flow path (deep versus surface). It has to be placed in the context of changing precipitation, plant biomass productivity and modification of the seasonality [101]. Complex evaluation of these factors goes beyond the scope of this study and it requires ecosystem-level regional modeling. Overall, the WSL is likely to have increased lateral export of DIC but decreased export of DOC. However, the changes of both fluxes (+2 and <sup>−</sup>3 to <sup>−</sup>4 t C km−<sup>2</sup> y−<sup>1</sup> for DIC and DOC, respectively) are dwarfed compared to possible magnitude of C emission from WSL inland waters (rivers and lakes) to the atmosphere: 10–20 t km−<sup>2</sup> y−<sup>1</sup> in permafrost-free and isolated zone and 20–40 t C km−<sup>2</sup> y−<sup>1</sup> in discontinuous and continuous permafrost zone [102].

#### **5. Conclusions**

Based on a two-year seasonal sampling of 32 western Siberian rivers of various size (from 10 to 10<sup>5</sup> km2 in watershed area) draining through a sizable permafrost gradient, we measured riverine export fluxes of dissolved (<0.45 μm) C, nutrients, major and TE over a six-month open-water period (May to October). The export fluxes of DOC, DIC, major cations, macro- and micro-nutrients, toxicants, and geochemical tracers were weakly dependent on the size of the river. The primary parameter of export fluxes control was latitude, which marked the position of the permafrost zones.

There are several distinct groups of elements defined according to the latitudinal patterns of their yields. These groups reflect the relative mobility of elements, consistent with general knowledge of river hydrochemistry in high latitude regions. The DOC, organically-bound metals (V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Cd and Pb) and many low soluble TE (Al, Ti, Zr, Hf, Th and REEs) exhibited similar latitudinal pattern with a minimum in the south and a maximum in the permafrost zone). A northward increase in Si export flux may be due to a decrease in Si uptake by plants in the north and strong Si retaining by mire and forest vegetation in the south.

An increase in DOC export fluxes from the south to the north (by a factor of 3 to 4 depending on the years of observation) could be due to leaching of OM from the plant litter and organic-rich topsoil (Histosol). The removal of DOC by adsorption on mineral horizons was hypothesized to be very low in the north. As a result, the riverine DOC export in the permafrost zone of the WSL may be strongly controlled by the water residence time necessary for DOC leaching from upper vegetation layer (moss, lichen, litter). This calls a need for quantitative modeling of DOM and element reactive transport in WSL peatlands, both at the scale of small watershed and large rivers. Furthermore, the peculiarity of western Siberia is that the elementary export by WSL rivers is strongly controlled by dynamics of peat formation and decay across the territory. Because the WSL is currently recovering from the last glaciation, this territory is dominated by non-stationary ecosystems with strong latitudinal contrast: the fresh peat is accumulating in the south whereas the old frozen peat is thawing and degrading in the north. As a result, uptake of elements from groundwater, river and forest tree litter by growing peat in southern bog competes with DOC and element release from thawing/degrading peat in the northern palsas. Note that, unlike many other permafrost-affected regions in the world whose CO2 uptake rate during weathering is likely to increase under climate warming, the WSL may increase its riverine export of DIC but decrease the export of DOC.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4441/12/6/1817/s1, Figure S1: Correlation between elementary fluxes and watershed area for permafrost-free and permafrost-affected regions; Figure S2: Partial contribution of spring, summer and autumn 2016 to overall open-water period export of anions and trace elements by WSL rivers; Figure S3: Ternary molar diagrams of major cations and anions in the WSL rivers. Figure S4: Yields of DOC (A), DIC (B), Ca (C), Mg (D), K (E), Si (F), Ptot (G) and Fe (H) during May-August 2015 in 33 WSL rivers across the latitudinal gradient. Table S1: Spearman correlation coefficients (p < 0.05) between element export flux (yield) and watershed area.

**Author Contributions:** O.S.P. designed the study and wrote the paper; R.M.M., I.V.K., and S.V.L. performed sampling, analysis of cations and anions, and their interpretation; S.N.V. and S.N.K. were responsible for the choice of sampling objects and statistical treatment; S.V.L. provided the background information on soil, peat, and permafrost active layer; L.S.S. was in charge of DOC, DIC, and anion measurements and their interpretation; L.G.K. provided GIS-based interpretation, mapping, and identification of river watersheds; S.G.K. performed all primary hydrological data collection, and their analysis and interpretation. All 10 authors participated in field expeditions. All authors have read and agreed to the published version of the manuscript.

**Funding:** Russian Science Foundation: No 18-17-00237 and 18-77-10045. Russian Fund for Fundaental Research: 19-55-15002, 20-05-00729\_a.

**Acknowledgments:** We acknowledge main support from RSF grant No 18-17-00237 and RFBR grants No 19-55-15002, 20-05-00729\_a, and RSF grant No 18-77-10045 for field work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **The Main Features of Phosphorus Transport in World Rivers**

**Vitaly S. Savenko <sup>1</sup> and Alla V. Savenko 2,\***


**Abstract:** Data on the geochemistry of phosphorus in the continental runoff of dissolved and solid substances were systematized and generalized, with a separate consideration of the processes of runoff transformation in river mouth areas. It has been established that atmospheric deposition, which many authors consider to be an important source of phosphorus in river runoff and not associated with mobilization processes in catchments, actually contains phosphorus from soil-plant recycling. This is confirmed by the fact that the input of phosphorus from the atmosphere into catchments exceeds its removal via water runoff. An analysis of the mass ratio of phosphorus in the adsorbed form and in the form of its own minerals was carried out. It was shown that the maximum mass of adsorbed phosphorus is limited by the solubility of its most stable minerals. The minimum concentrations of dissolved mineral and total phosphorus were observed in the rivers of the Arctic and subarctic belts; the maximum concentrations were confined to the most densely populated temperate zone and the zone of dry tropics and subtropics. In the waters of the primary hydrographic network, the phosphorus concentration exhibited direct relationships with the population density in the catchments and the mineralization of the river water and was closely correlated with the nitrogen content. This strongly suggests that economic activity is one of the main factors in the formation of river phosphorus runoff. The generalization of the authors' and the literature's data on the behavior of phosphorus at the river–sea mixing zone made it possible to draw a conclusion about the nonconservative distribution of phosphorus, in most cases associated with biological production and destruction processes. The conservative behavior of phosphorus was observed only in heavily polluted river mouths with abnormally high concentrations of this element.

**Keywords:** geochemistry of phosphorus; continental runoff; river mouth

#### **1. Introduction**

In the second half of the last century, the uncontrolled growth of economic activity led to a significant disruption in the natural migration of chemical elements, which can be eliminated or optimized only by controlling the fluxes of matter in the environment. In this regard, knowledge of the basic laws and physicochemical mechanisms of chemical element migration in the global hydrological cycle, which links the objects of the biosphere into an integrated dynamic system, is of paramount importance. Here, based on numerous former studies of phosphorus transport within the global hydrological cycle, we searched for the general patterns and physicochemical mechanisms of the aqueous migration of chemical elements in the global hydrological cycle. The objective of this work is to present the general features of the phosphorus biogeochemical cycle and describe the physicochemical mechanisms controlling phosphorus migration in the aqueous systems of the earth's surface, notably river runoff within the context of the global hydrological cycle.

#### **2. Phosphorus Mobilization at the Stage of River Runoff Formation**

The initial stage in the formation of the chemical composition of surface waters is often associated with atmospheric precipitation on the earth's surface and their subsequent

**Citation:** Savenko, V.S.; Savenko, A.V. The Main Features of Phosphorus Transport in World Rivers. *Water* **2022**, *14*, 16. https://doi.org/ 10.3390/w14010016

Academic Editor: Liudmila S. Shirokova

Received: 27 November 2021 Accepted: 19 December 2021 Published: 22 December 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

interaction with soil and vegetation cover and rocks. However, due to the constant presence of terrigenous aerosols (mainly the products of the wind erosion of soils) in the surface air layers, this interaction begins already in the atmosphere immediately after the condensation of water vapor. Therefore, it is expedient to divide the mobilization of dissolved substances at the initial stage of river runoff formation into the mobilization in the atmosphere and in the catchments.

#### *2.1. Phosphorus Mobilization in the Atmosphere*

Chemical elements are delivered from the atmosphere to the catchments in the form of wet (rain, snow) and dry (aerosols) precipitation. The chemical composition of wet precipitation is due to leaching from the atmosphere and the partial dissolution of aerosols, which are represented by substances of terrigenous and marine genesis. The contribution of marine aerosols to the transport of phosphorus into the land is apparently small. This is indicated by an exponential decrease in the content of aerosol phosphorus in the lower atmosphere when moving from the coast to the central regions of the ocean and a significantly lower content of phosphorus in the rains over the ocean compared to land [1–3]. The estimates of phosphorus input into the atmosphere from various sources confirm this conclusion and show that the main role is played by the aeolian erosion of the soil cover and the combustion of terrestrial vegetation (Table 1). Another source associated with the products of plant metabolism (spores, pollen, volatile organic compounds, and small particles of plant residues) is currently not quantifiable, but observations unambiguously indicate the widespread occurrence of plant metabolism products present in atmospheric aerosols.

**Table 1.** Sources of phosphorus in the atmosphere [4].


<sup>1</sup> By the composition of terrestrial vegetation at 10% ash content.

The total phosphorus concentration in aerosols varies from 600 to 4700 μg/g, averaging ~2000 μg/g [5], which is 2–5 times higher than the phosphorus content in the rocks of the earth's crust and soils, the main sources of terrigenous material in the atmosphere. The increased phosphorus concentrations in atmospheric aerosols are logically explained by the presence of the solid products of plant biomass combustion in the amount of 0.6–1.1% of the total aerosol mass [4]. A significant part of the phosphorus in aerosols is present in water-soluble form, which, as a rule, accounts for 20–50% of its total content [6]. Apparently, the soluble forms of phosphorus in aerosols are associated with the products of combustion of plant biomass and its destruction.

In atmospheric precipitation, the concentrations of mineral and total phosphorus (Pmin and Ptotal) are distributed in accordance with the lognormal law. The average median concentrations of these forms are 15 and 33 μg/L and the values of their input with atmospheric precipitation to the earth's surface are 0.11 and 0.25 kg/ha yr (Table 2). The percentage of the soluble forms of the total phosphorus in atmospheric precipitation is in the range of 20–80%, with an average value of 55% [6], which is in good agreement with the percentage of soluble phosphorus in aerosols, the main source of dissolved substances.


**Table 2.** Concentration of phosphorus in atmospheric precipitation and its input on the earth's surface with wet precipitation [7].

Atmospheric precipitation is considered by many authors as an important source of phosphorus in river runoff, which is not associated with the processes of its mobilization in the catchments. However, the balance of total phosphorus in the catchments shows that the input of this element with atmospheric precipitation usually exceeds the removal with water runoff [4]. The opposite situation, when the phosphorus runoff exceeds its input, is observed, as a rule, under the conditions of a strong anthropogenic load. The positive value of the difference between the phosphorus input from the atmosphere and its removal from the catchments is an artifact that is associated with the lack of reliable methods for quantifying the masses of substances remobilized from the earth's surface into the atmosphere and returned back as a part of atmospheric precipitation (Figure 1).

**Figure 1.** Scheme of phosphorus fluxes in the catchments.

#### *2.2. Phosphorus Mobilization in the Catchment Areas*

The primary sources of phosphorus are igneous, metamorphic, and sedimentary rocks, which differ significantly in the content of this element (Table 3). The maximum phosphorus concentrations are characteristic of basic and intermediate magmas; with an increase and decrease in acidity, the phosphorus content in igneous rocks decreases. In sedimentary rocks, the phosphorus concentration does not vary so much, and in general, for the sedimentary deposits it is slightly higher than in granites. In metamorphic processes, phosphorus behaves as an inert component, and its content is inherited from the parent rocks.



<sup>1</sup> Based on results of [9].

Apatite is the main phosphorus mineral in all types of igneous, metamorphic, and sedimentary rocks. The abundance of two other important phosphorus minerals, xenotime YPO4 and monazite CePO4, is 100–1000 times lower than that of apatite and can reach 10% only in acid rocks [10]. According to mineralogical analysis, in igneous rocks, apatite accounts for 1.7–5.7% of the total phosphorus, whereas in sedimentary carbonate, clayey, and sandy rocks, apatite contains 22.9, 0.5, and 7.1% of phosphorus, respectively [10,11]. In magmatic and metamorphic silicates, phosphorus can isomorphically replace silicon with charge compensation (Na<sup>+</sup> + P5+ = Ca2+ + Si4+ and Al3+ + P5+ = 2 Si4+) or with the formation of cation vacancies. In the Critical Zone, the bulk of phosphorus is in the sorbed state, as well as in the form of apatite and various iron and aluminum phosphates. The composition of apatite is different for various types of rocks. Fluorapatite predominates in igneous and metamorphic rocks, with fluoro-carbonate-apatite pervading in sedimentary rocks, and bone phosphate is represented by hydroxyl-apatite and carbonate-hydroxyl-apatite.

Biological metabolites and the products of dead organisms' destruction are an important source of phosphorus in continental runoff. The phosphorus content in land plants (on average 1500–2000 μg/g dry weight) is almost an order of magnitude lower than the content in animals and bacteria [12–14]. Therefore, the destruction of animal and bacterial biomasses can lead to the emergence of high local concentrations of dissolved phosphorus.

The most obvious factor in phosphorus mobilization is the solubility of the phosphoruscontaining mineral phases. According to calculations [15], the concentration of dissolved mineral phosphorus in the waters of the Rhine and Rhone rivers is controlled by the solubility of hydroxylapatite. However, under the conditions of the earth's surface, hydroxylapatite is unstable and transforms into a less soluble fluoro-carbonate-apatite. The dissolution of fluorapatite in fresh waters leads to a concentration of dissolved mineral phosphorus at the level of 14 ± 3 μg/L [16], which, as will be shown below, approximately corresponds to the average median value for the world rivers.

The acidity of the aquatic environment is apparently the main factor controlling the stability of the mineral forms of phosphorus. In a moderately alkaline medium, the stable phase is fluoro-carbonate-apatite; in a moderately acidic medium, iron (III) and aluminum phosphates are stable under oxidizing conditions and iron (II) and aluminum phosphates are stable under reducing conditions [6]. According to the experimental data [17], in waters with a reaction close to neutral, the monophosphates of iron (III) and aluminum transform into more stable iron-calcium and aluminum-calcium phosphates with the hypothetical chemical formulas CaFe(OH)3HPO4 and CaAl(OH)3HPO4. In the neutral medium, the dissolved iron (III) and aluminum are mainly in the form of the electroneutral hydroxocomplexes Fe(OH) 0 <sup>3</sup> and Al(OH) 0 <sup>3</sup>, and the bulk of the phosphorus is represented by HPO2<sup>−</sup> <sup>4</sup> . Therefore, in accordance with the dissolution reactions

$$\text{CaFe(OH)}\_{3}\text{HPO}\_{4} = \text{Ca}^{2+} + \text{Fe(OH)}\_{3}^{0} + \text{HPO}\_{4}^{2-},\tag{1}$$

$$\text{CaAl(OH)}\_{3}\text{HPO}\_{4} = \text{Ca}^{2+} + \text{Al(OH)}\_{3}^{0} + \text{HPO}\_{4}^{2-} \tag{2}$$

an inverse relationship between the logarithms of the concentrations of mineral phosphorus and calcium is observed (Figure 2).

In equilibrium with iron-calcium and aluminum-calcium phosphates, the concentration of dissolved mineral phosphorus is significantly higher than its content in river and ground waters. Therefore, it should be assumed that the presence of these solid phases is possible only where high local concentrations of dissolved phosphorus can be maintained for a long time. These can be bottom sediments with an extremely slow rate of water exchange or soils in which a high concentration of dissolved phosphorus is provided by the destruction of organic matter during the biological cycle. In all other cases, ironcalcium and aluminum-calcium phosphates must be replaced by hydroxides containing adsorbed phosphorus.

**Figure 2.** Relationship between logarithms of the concentrations of dissolved phosphates and calcium in the interaction of FePO4 and AlPO4 with fresh waters [17]. FePO4: (*1*) water from the Moscow River, (*2*) water from the Don River mouth; AlPO4: (*3*) water from the Moscow River, (*4*) water from the Don River mouth.

At sufficiently high concentrations of phosphates, arising, for example, during the destruction of animal or bacterial biomass, the silicate phosphatization reaction can occur, in which the silicon of the solid phase is replaced by phosphorus from the solution. This process was experimentally studied by us, using rock-forming minerals of different structural types (hornblende, orthoclase, labradorite, kaolinite, and montmorillonite) and background buffer solutions with variable concentrations of orthophosphates (0.25–6.0 mM), maintaining the pH at ~1.8, 3.7, 4.9, 6.8, 7.8, and 8.8 [18–20].

The results of the experiments demonstrated the following features. First, all the samples were characterized by approximately equivalent variations in the concentrations of phosphorus and silicon in the solution in the pH range of 3.7–8.8:

$$
\Delta[\text{Si}] \approx -\Delta[\text{P}]\_\prime \tag{3}
$$

whereas at pH 1.8 the supply of the dissolved silicon was 1.3–2 times higher than the removal of the phosphates (Figure 3), which was likely explained by the change in the stoichiometry of the phosphatization reaction. Second, the amount of phosphorus absorbed by the silicates was linearly dependent on its final concentration in the solution,

$$-\Delta[\mathbf{P}] = k[\mathbf{P}]\_{\text{final}} \tag{4}$$

with almost the same values of the proportionality coefficient *k* for the different minerals, slightly decreasing with a decrease in the acidity of the medium (Table 4).

**Table 4.** Proportionality coefficient *k* in Equation (4) as a function of solution pH.


According to the data in Table 5, the amount of silicon removed from the studied silicates and replaced by phosphates at pH 3.7–8.8 reached 6.5–11.0% of the initial silicon content in the minerals. Even more silicon (up to 9.4–19.9%) entered the solution at pH 1.8, when the process of phosphatization was accompanied by the acid leaching of silicates, which led to an additional release of silicon and violation of equivalence (3). Such large amounts of removed silicon and absorbed phosphorus, which were much higher than the

limiting values of the sorption removal of phosphates, definitely indicated the occurrence of a chemical reaction which replaced the silicate with a phosphate mineral.

**Figure 3.** Correlation between variations in the concentrations of phosphorus and silicon in the solution upon phosphotization of silicates [19]. (**a**) pH 1.8: (*1*) hornblende, (*2*) orthoclase, (*3*) labradorite; pH 3.7–8.8: (*4*) hornblende, (*5*) orthoclase, (*6*) labradorite. (**b**) pH 1.8: (*1*) kaolinite, Glukhovetsk, (*2*) the same, Podol'sk, (*3*) montmorillonite, Askania, (*4*) the same, near Askania; pH 3.7–8.8: (*5*) kaolinite, Glukhovetsk, (*6*) the same, Podol'sk; (*7*) montmorillonite, Askania, (*8*) the same, near Askania.

**Table 5.** Amount of silicon passed into solution in the experiments on phosphatization of silicates at the maximum initial concentration of phosphates 1, % of the initial concentration in the mineral [19].


<sup>1</sup> 5 mM for hornblende, orthoclase, and labradorite; 6 mM for kaolinite and montmorillonite.

In previous studies [21–25], the negative correlation between the variations in the concentrations of dissolved phosphates and silicon was associated with the adsorption exchange of phosphate ions and silica on the surface of silicates. Since the duration of the experiments did not exceed several days, this time was sufficient to establish the adsorption equilibrium but was not long enough for noticeable progress in the phosphatization reaction of the bulk silicate phase. Our experiments proceeded for more than one year, so the amount of phosphorus absorbed from the solution and the silicon displaced from the solid phase indicated the participation in the process of not only the surface layer, but also the volume of the solid phase.

The same quantitative characteristics of the process of phosphatization for all the studied silicates, corresponding to different structural types and with different chemical compositions, were an unusual result. It can be assumed that the initial minerals were not subject to phosphatization, but that the secondary silicate phases formed during the interaction of the silicates with water and were stable in a certain pH range. The parameters of the phosphatization reaction at pH 1.8 varied due to the stability under these conditions of the surface silicate phase, which was different to that in the area of higher pH values.

A powerful factor of the phosphorus mobilization in the Critical Zone is the activity of living organisms. Primary producers annually synthesize about 140 Gt of dry organic matter on land, 98–99% of which is mineralized. With the average phosphorus content in plants equal to ~1500 μg/g dry matter, about 210 Mt of phosphorus participates in the biotic cycle, which forms soluble phosphates at the stage of mineralization and becomes a potential source of dissolved phosphorus in continental runoff. However, mineralized phosphorus is almost completely reincluded in the biotic cycle and used to create new organic matter. The highest degree of completeness of the biotic cycle is inherent in mature biogeocenoses (Table 6).

**Table 6.** Phosphorus input with litter and removal with subsurface runoff in forest biogeocenoses [26].


If all mineralized phosphorus was a part of river runoff, the volume of which is 41,700 km3/yr [27], its concentration due to this source alone would be 5 mg/L. Such high concentrations of dissolved phosphorus are extremely rare and usually associated with the reducing conditions of the environment or anthropogenic pollution. The average concentration of dissolved phosphates in unpolluted river waters is equal 30–50 μg P/L [28–30], which is 0.6–1.0% of the calculated value of 5 mg P/L. This means that continental runoff contains a very small portion of the labile phosphorus that is formed as a result of organic matter degradation.

It is known that when phosphorus fertilizers are applied to soils, the behavior of the phosphorus differs significantly depending on the properties of the soil and the fertilizers themselves. Poorly soluble phosphorite flour increases the content of biologically available phosphorus if the soil conditions are conducive to the transformation of apatite into more soluble forms. With the addition of highly soluble fertilizers, over time, phosphorus immobilization occurs due to chemosorption and the formation of poorly soluble compounds, including apatite phases. It is assumed that phosphates of iron, aluminum, and calcium make up ~90% of the immobilized phosphorus of fertilizers [31].

Whereas the final products of the transformation of fertilizers are represented by poorly soluble mineral phases, there is usually no direct relationship between the amount of applied phosphorus and its removal. The amount of removed phosphorus from fertilizers, as a rule, does not exceed 1–2% [32–35].

Formally, the mobility of chemical elements in the Critical Zone is characterized by the coefficients of water migration *Ki*, equal to the ratio of the concentrations of element *i* in the dry residue of water (*ai*) and in drained rocks (*mi*):

$$K\_i = a\_i / m\_i. \tag{5}$$

Phosphorus belongs to the group of low-mobility elements with 0.01 < *Ki* < 0.1 [36].

In (5), it is implicitly assumed that all the substances in the dry residue of water enter it as a result of the dissolution of drained rocks. However, there are two other powerful sources of dissolved matter: cyclic sea salts, transported from the ocean to land through the atmosphere, and anthropogenic substances. Taking into account the contribution of these sources leads to a significant change in the values of the coefficients of water migration, in particular, to an approximately tenfold increase of this coefficient for phosphorus (Table 7). **Table 7.** Coefficients of water migration of chemical elements in the Critical Zone taking into account the contribution of cyclic sea salts and anthropogenic substances [37].


#### **3. Phosphorus in River Runoff**

#### *3.1. Phosphorus in the Waters of the Primary Hydrographic Network*

The primary hydrographic network consists of small catchments, which are characterized by the significant spatial variability of the chemical composition of the waters, caused by the territorial heterogeneity of geomorphological, lithological, and biological soil conditions. The enlargement of rivers and pooling of small catchments leads to the "averaging" of the local conditions for runoff formation. Therefore, the larger-scale regularities associated with the implementation of the periodic law of geographic zonality are acquiring decisive importance.

The lithological characteristics of the catchments have a strong influence on the phosphorus concentration in the waters of the primary hydrographic network, because rocks are the main source of dissolved phosphorus. The highest concentrations of dissolved mineral phosphorus are found in catchments located on basalts, in which the phosphorus content is greater than in other types of rocks (Table 8). The runoff of dissolved phosphorus from drainage basins composed of sedimentary rocks is usually greater than for igneous rocks. In the small, almost completely forested catchments on the Canadian Crystalline Shield, the dissolved phosphorus runoff was 4.8 (2.5–7.7) mg/m<sup>2</sup> yr for igneous rocks and twice as large (10.7 (6.0–14.5) mg/m2 yr) for sedimentary rocks [38].

**Table 8.** Relationship between the concentration of dissolved mineral phosphorus in the waters of the primary hydrographic network and phosphorus content in the catchment rocks.


Another important factor of phosphorus migration is the climate, which affects the rate of weathering and, consequently, the intensity of the phosphorus mobilization from rocks. For example, the runoff of dissolved phosphorus from the territory of Karelia (NW Russia, temperate climate) due to the pure weathering of crystalline rocks is 2 mg/m<sup>2</sup> yr [39], while the average intensity of dissolved phosphorus removal during the weathering of crystalline rocks for three catchments in Brazil (humid tropical climate) is 5 times higher: 10 (5–14) mg/m<sup>2</sup> yr [40].

The presence of areas with slow water exchange in catchments leads to a decrease in the phosphorus content in the waters of the primary hydrographic network. Indeed, Conley et al. [41] showed an exponential dependence of the concentration of total dissolved phosphorus ([Ptotal], μg/L) on the relative area of lakes (*S*, %) in catchments:

$$\left[\mathrm{P}\_{\mathrm{total}}\right] = 2.18e^{-0.096S}.\tag{6}$$

Data on the content of dissolved phosphorus in the waters of the primary hydrographic network, to which catchments with an area ≤50 km<sup>2</sup> were assigned, were collected

during observations that lasted for at least a year and were systematized in [42]. Based on the differences in the sources of phosphorus input, the conditions of runoff formation, and the processes in the catchments, all catchments were divided into four groups: (1) natural (forest) catchments; (2) mixed agricultural–forest catchments with land use <50%; (3) agricultural catchments with land use >50%; (4) urban catchments. For a number of catchments, the group could not be determined due to the lack of the necessary data.

Table 9 shows that the values of the arithmetic and median mean concentrations of mineral and total phosphorus in solution for all the accounted catchments differed several times, indicating the positive asymmetry of the probability distribution functions, which corresponds to the lognormal law. When the small catchments were combined into groups, the asymmetry of the probability distribution functions for the phosphorus concentrations remained. Therefore, the average median concentrations of mineral and total phosphorus, equal to 31 and 95 μg/L, can be considered as the global average concentrations of these forms of dissolved phosphorus in the waters of the primary hydrographic network under modern conditions.


**Table 9.** The average content and concentration range of dissolved phosphorus (μg/L) in the waters of the primary hydrographic network [42].

The lowest concentrations of the dissolved forms of mineral and total phosphorus in the surface waters (7 and 28 μg/L) were observed in the forest landscapes with the least anthropogenic impact. As the economic activity intensified, the phosphorus content increased. For the mixed agricultural–forest catchments, the average concentrations of Pmin and Ptotal were 48 and 90 μg/L, while for the agricultural catchments they increased to 116 and 250 μg/L. An even higher content of dissolved phosphorus was characteristic of the urban catchments, where the average concentrations of Pmin and Ptotal reached 700 and 1500 μg/L. In general, there was a tendency towards an increase in the concentrations of Pmin and Ptotal in the surface waters of the small catchments as the population density increased (Figure 4).

#### *3.2. Phosphorus in River Waters*

The rivers of the world carry into the ocean ~3 Gt/yr of dissolved matter and 15–20 Gt/yr of solid matter. The phosphorus runoff in the form of particulate suspended matter significantly exceeds its dissolved flux, which plays an extremely important role for biota and biogeochemical processes.

**Figure 4.** Relationship between the average annual concentrations of dissolved forms of mineral (*1*) and total (*2*) phosphorus in the waters of the primary hydrographic network and the population density *D* [6].

#### 3.2.1. Phosphorus of Suspended Matter and Bed Load

The distribution function of the phosphorus content in suspensions of 77 large, medium, and small rivers of the world corresponds to a lognormal law; the arithmetic and geometric mean concentrations of phosphorus equal 1500 and 1000 μg/g [43], respectively, which is close to the estimate [44]: 1270 μg/g. About 3% of the phosphorus in river suspended matter is represented by bioavailable soluble/exchangeable forms that can be used by living organisms [45,46].

Phosphorus runoff in the form of suspended solids is affected by the ratio of fine and coarse fractions. The phosphorus content in the fine fractions of the suspended matter and bottom sediments of rivers is 2–10 times higher than that in the coarse fractions. The consequence of this is apparently a decrease in the phosphorus concentration in river suspensions, with an increase in the total content of suspended solids (turbidity), which is accompanied by an increase in the proportion of the coarse fractions (Figure 5). The highest phosphorus concentrations (~4000 μg/g) are observed at a turbidity <20 mg/L, while at a turbidity >100 mg/L, the phosphorus concentration begins to decline sharply, reaching 400 μg/g at a suspended matter content of 1000 mg/L. The same reason leads to an inverse relationship between the concentration of phosphorus in suspended matter and the water discharge or erosion rate. At small discharges during the low-water period, the relative contribution of fine suspensions increases and the phosphorus concentration reaches its maximum values, while in the high-water period, the bulk of suspended solids are represented by coarse suspensions with low phosphorus content.

The use of fertilizers is accompanied by an immobilization of the phosphorus in the upper soil horizons, which are the main supplier of suspended matter. As a result, the phosphorus content in suspensions denudated from cultivated lands is approximately 2 times higher than in the runoff of solids from forest catchments: 2500 and 1100 μg/g, respectively [47].

Forests prevent the erosion of the earth's surface and should reduce phosphorus runoff. This is confirmed by the data for seven small catchments in Southern Quebec [48], where the relationship between the concentration of suspended phosphorus ([Psusp], μg/L) and the degree of forest coverage of the territory (*X*, %) was established:

$$[P\_{\text{susp}}] = 10.2 - 0.056X. \tag{7}$$

Deforestation should lead to an increase in suspended phosphorus runoff on a global scale, but it is still very difficult to quantify this effect.

**Figure 5.** Relationship between the phosphorus content in suspended matter and the turbidity (*s*) of river waters [43].

It is estimated that 10 to 30% of the river runoff of solid matter is carried in the form of bed load, in which the phosphorus content is on average 800 μg/g [49]. This value is lower than the phosphorus content in river suspended matter (1000 μg/g), which corresponds to the larger hydraulic size of the bed load.

#### 3.2.2. Dissolved Phosphorus

In [50,51], the average annual and long-term average annual data on the content of the dissolved forms of mineral and total phosphorus in 179 rivers of the world (>200 observation stations) are summarized. The arithmetic and median mean concentrations of dissolved mineral phosphorus are 113 and 28 μg/L, respectively, and those of total dissolved phosphorus are equal to 241 and 85 μg/L (Table 10). The distribution of the concentrations of dissolved phosphorus obeys the lognormal law; therefore, the median mean concentrations are preferred for obtaining average values.


**Table 10.** The average content of dissolved phosphorus (μg/L) in the river waters of different geographic zones [50,51].

Most of the natural factors affecting the content of chemical elements in river water are closely related to the geographic zonality, which determines the features and intensity of the weathering processes, biological activity, etc. In this regard, to analyze the spatial distribution of phosphorus content, all rivers were divided into four groups according to their geographical zones: Arctic region and subarctic zone, temperate zone, humid tropics and subtropics, and dry tropics and subtropics.

The minimum median mean concentrations of the dissolved forms of mineral and total phosphorus were observed in the rivers of the Arctic and subarctic belts, where the biological cycle of elements is much slower and the anthropogenic impact on the aquatic environment is not pronounced, given that there are no extensive sources of phosphorus input associated with agricultural industries, fewer large cities, and, therefore, less industrial and domestic wastewater. The highest median mean concentrations of dissolved mineral phosphorus were characteristic of the rivers of the temperate zone and the zone of dry tropics and subtropics. This is explained by the powerful anthropogenic impact on the nature of these regions, as well as the favorable conditions for the involvement of phosphorus in the biological cycle and its rapid turnover therein. A similar situation is typical for total dissolved phosphorus.

The average annual concentrations of the dissolved forms of mineral and total phosphorus for the rivers of the world correlate with the mineralization of river water (*r* = 0.94 and 0.89, respectively) and with the concentration of total nitrogen (*r* = 0.81 and 0.79, respectively) [51]. The cycles of nitrogen and phosphorus are closely linked in the biological cycle of matter and liable to similar anthropogenic changes. Like phosphorus, nitrogen is used in mineral fertilizers and its concentration in wastewater also increases tens and hundreds of times. A rather close correlation between dissolved phosphorus and the mineralization of waters is interesting. It can be assumed that it arises due to an increase in the mineralization of river water in the north–south direction parallel to an increase in the population density, which is an indicator of the anthropogenic load and, in particular, of the intensity of anthropogenic phosphorus sources. Indeed, the average concentrations of mineral and total phosphorus in the river water regularly increase with an increase in the population density in the catchments (Table 11).


**Table 11.** The average content of dissolved phosphorus in the water of rivers with different population densities in their catchments [51].

According to [52], for large rivers there is only a weakly expressed tendency towards an increase in the runoff of dissolved mineral phosphorus with an increase in the population density in the catchments. However, if one takes into account the presence of a directly proportional dependence of phosphorus removal from catchments on the value of specific water discharge, a significant correlation (*r* = 0.78) is found between the runoff of dissolved mineral phosphorus and the population density in the catchments, normalized to the specific water discharge.

The intensification of economic activity is accompanied by an increase in the phosphorus content in river runoff. Systematic observations carried out in 1936–1980 on the territory of the USSR showed a noticeable increase in the concentration and runoff of dissolved mineral phosphorus over time (Table 12). The same was established for other large rivers of the world, including the coastal parts of the sea basins into which these rivers flow [53].


**Table 12.** Change in water runoff (*Q*, km3/yr), concentration ([Pmin], μg/L) and runoff (*J*Pmin, thous. t/yr) of dissolved mineral phosphorus in the USSR in 1936–1980 [54].

Environmental protection measures can not only stop the increase in dissolved phosphorus concentrations but also cause its significant decrease. In particular, due to a reduction in the volumes of municipal wastewater and the use of phosphorus-containing detergents, the total phosphorus runoff into Lake Erie decreased from 27.9 to 10.5 thous. t/yr during 1968–1981 [55]. The deepening of wastewater treatment and a decrease in its volume led to a decrease in the phosphorus runoff into the Rhine and Elbe rivers from 51.1 and 20.5 thous. t/yr, respectively, in 1983–1987 to 20.5 and 12.5 thous. t/yr in 1993–1997 [56].

#### *3.3. Phosphorus in Groundwater in the Zone of Active Water Exchange*

The surface waters of the primary hydrographic network, rivers and lakes, are in direct hydrodynamic connection with groundwater, which plays an important role in the formation of the chemical composition of the continental runoff of dissolved matter. The greatest influence is exerted by the groundwater of the zone of active water exchange, the discharge of which is the main source of river runoff during the low-water period. The phosphorus content in groundwater is of the same order of magnitude as in the waters of the primary hydrographic network.

The average content of dissolved mineral phosphorus in the groundwater of the Critical Zone varies within the same order of magnitude: from 18 to 191 μg/L (Table 13). The maximum concentrations (191 and 127 μg P/L) were found in the waters of bog landscapes and steppes (dry savannah). The lowest phosphorus content was observed in the waters of permafrost zones and mountainous areas, in which the fluorine mobilization from rocks is impeded by the low temperature and relatively high water velocity, respectively. The concentrations of dissolved mineral phosphorus in the groundwater in areas of leaching and continental salinization, despite the significant difference in their mineralization, are relatively equal, amounting to 56.9 and 62.6 μg/L, respectively.


**Table 13.** The content of dissolved phosphorus in groundwater of the Critical Zone [57].


**Table 13.** *Cont.*

#### *3.4. Integral Characteristic of the Phosphorus River Runoff*

3.4.1. Phosphorus Runoff in the Composition of Suspended Matter and Bed Load

The average phosphorus concentrations in the suspended matter and bed load of world rivers are 1000 and 800 μg/g, respectively [49]. The most detailed calculations of the global runoff of suspended matter give a value of 15.5 Gt/yr [58,59]. The mass of the bed load, according to various estimates, is from 10 to 30% of the mass of suspended matter, and 20% can be taken as an average value. Hence, the total continental runoff of suspended and drawn phosphorus is equal to 18.0 Mt/yr. This value is in close agreement with earlier estimates, 16.1 [44] and 20.4 [60] Mt P/yr; however, these did not take into account the runoff of bed load.

Phosphorus is also carried out from land via ice runoff and coastal abrasion. Here, phosphorus is mainly contained in the lithogenic material, while the contribution of its dissolved forms is negligible. The phosphorus content in the products of glacial erosion and coastal abrasion can be taken to equal that in the rocks of the land surface: 690 μg/g. A.P. Lisitsyn [61] estimates the removal of the solid products of ice runoff and coastal abrasion to be 1.5 and 0.5 Gt/yr, which corresponds to a phosphorus mass of 1.4 Mt/yr.

#### 3.4.2. Dissolved Phosphorus in River Runoff

A detailed assessment of river phosphorus runoff was made in [50,51], where data for more than 100 medium and large rivers of the world were used and a correction for the value of the accounted water runoff for each continent was applied (Table 14). The total volume of continental water runoff in these works was taken to be equal to 38,500 km3/yr. The more correct value is 41,700 km3/yr [27], which would mean that the river runoff of dissolved mineral and total phosphorus increases to 1.6 and 4.5 Mt/yr, respectively.

#### 3.4.3. Dissolved Phosphorus in Groundwater Runoff

Groundwater phosphorus runoff is difficult to estimate due to the limited amount of available information. According to calculations [62], the proportions of the dissolved forms of mineral and total phosphorus in the groundwater and river runoff into the seas of the Russian Arctic are approximately the same, at 11–13% (Table 15). A similar proportion of dissolved mineral phosphorus in groundwater and river runoff also follows from the global estimates [63]. With an average concentration of phosphorus in the groundwater of the Critical Zone of 58 μg/L [57] and a groundwater runoff value of 2200 km3/yr, the phosphorus removal into the ocean is 0.13 Mt/yr, or ~8% of the river runoff of dissolved mineral phosphorus.


**Table 14.** River runoff of dissolved forms of mineral and total phosphorus [50,51] 1.

<sup>1</sup> Values in parentheses are calculated for water runoff of 41,700 km3/yr.

**Table 15.** The proportion of dissolved phosphorus in groundwater and river runoff into the marginal seas of the Russian Arctic [62].


<sup>1</sup> Weighted mean for water runoff.

#### **4. Phosphorus in the Mixing Zone of River and Sea Waters**

The final stage of the transformation of the river runoff of dissolved and suspended matter is carried out in the mouth area of rivers, as a result of which the ratio of the dissolved and suspended forms of chemical elements entering the ocean changes.

The dissolved components with conservative behavior are involved in intrabasin chemical and biological processes to an insignificant extent, and their content linearly depends on the ratio of the proportions of the sea and river water masses in the mixing zone. The components with nonconservative behavior are also added into the solution or are removed from it as a result of their involvement in different processes occurring in the water column or at the water–air and water–bottom boundaries. In this case, the linearity of the relationship between the component concentration and the ratio of the water mass proportions is violated. The best indicator of the ratio of the sea and river water mass proportions is the isotopic composition of water; however, the concentration of chemically inert chlorides, which is more accessible for measurements, is used more often.

The conservative behavior of the dissolved component *i* in the mixing zone of river and sea waters is described by the linear relationship between its concentration [*i*]*mix* and chloride content [Cl]*mix*:

$$[i]\_{mix} = a + b[\mathbf{C}1]\_{mix} \tag{8}$$

where *a* = [*i*]*rw*[Cl]*sw*−[*i*]*sw*[Cl]*rw* [Cl]*sw*−[Cl]*rw* <sup>≈</sup> [*i*]*rw* is a constant parameter; *<sup>b</sup>* <sup>=</sup> [*i*]*sw*−[*i*]*rw* [Cl]*sw*−[ Cl]*rw* is the slope ratio taking positive or negative values at a higher or lower concentration of the component *i* in seawater in comparison with river runoff; and the subscripts "*rw*" and "*sw*" denote the concentrations in river and sea water, respectively. If the component *i* is removed from the solution or, on the contrary, its internal source is present, the line showing the actual distribution of the concentrations of the relevant component is located, respectively, below or above the calculated line of conservative behavior (Figure 6). Equation (8) is widely used to determine the type of behavior of chemical components in the mixing zones of river and sea waters.

**Figure 6.** Relationships between the concentration of dissolved component *i* and chloride content by the conservative behavior (*1*) and availability of processes of additional input (*2*) or removal (*3*) of this component in the mixing zone of river and sea waters: (**a**,**b**) are cases when the concentration of component *i* in the river water is accordingly below or above that in the seawater.

Active participation in biological processes brings about the nonconservative behavior of phosphorus, which is observed in most river mouths of the world. The consumption of dissolved mineral phosphorus by phytoplankton leads to a decrease in its content in the water down to "analytical zero". The mineralization of the precipitated organic detritus causes the input of phosphorus into a solution in the lower layers of the water column and at the water–bottom boundary. Approximately half of the phosphorus entering the bottom as part of the organic detritus, after the mineralization of organic matter, can return back to the water with circulating currents or during the stirring up of the bottom sediments [64]. In addition to the production–destruction processes, an important role in the transformation of dissolved phosphorus runoff in the mixing zone of river and sea waters is played by the transformation processes of phosphorus-containing solid phases, sorption–desorption, and coprecipitation, as well as the diagenetic processes that control the phosphorus fluxes at the water–bottom boundary [6].

Table 16 summarizes the data on the distribution of the concentrations of the dissolved forms of mineral, organic, and total phosphorus in the mouth areas of large and small rivers of the world. The behavior of phosphorus in river mouths can be both nonconservative and conservative. In some cases, a complex type of phosphorus distribution was established exhibiting different behavior in various parts of the mixing zone.

#### *4.1. Nonconservative Behavior of Phosphorus*

Biological processes are the main reason for the nonconservative behavior of dissolved phosphorus in the mixing zones of river and sea waters. This is shown in the interrelated change in the phosphorus content and the various characteristics of the biological processes: the concentrations of nutrients, oxygen, chlorophyll, and organic detritus, and the pH value.

*Water* **2022**, *14*, 16


**Table 16.** The behavior of dissolved phosphorus in the mouth areas of world rivers.







**Reference**

 " "

[78]

"

"

 " "

 [79]

 [80]

 "

 "

 " A.S.

"

"

[81]

[82]

 " "



 **Runoff**

**Supposed Cause of**

**Nonconservative**

**Behavior**

Destruction of

"

organic matter

Biological consumption Input from bottom

sediments

 As above

 "

 –

Biological consumption

Input from bottom

sediments Desorption from

"

suspended matter

Desorption from

suspended matter

 As above

 –

Desorption from

suspended matter

Biological consumption

 As above

 "

 "

 "

Input from bottom

sediments

 September 2006

"

Pmin

"

 April 2017

April 2016

 0–8

 0–5

 0–3

 1.6

 1.1

 +300

 As above

 "

 0–5

 16

 29

 +150

Ural

 As above

 0–9

 0–1

 14

 3.4

−64

 [83]

 "

 "

 " [84]

"

 "

 " "

Mineralization

 " "

Mineralization

 "

 "

 "

 " "

"

"

"

**Reference**

> **%**




*Water* **2022**, *14*, 16


The leading role of production–destruction processes in the transformation of dissolved phosphorus runoff has been established for the mouths of many rivers (Table 16): Neva, the small rivers of the Kandalaksha Bay of the White Sea, Onega, Kyanda, Northern Dvina, Mezen, Pechora, Ob, Yenisei, most of the small rivers in the Black and the Azov Sea catchments, Volga, Ural, Yangtze, the Far Eastern rivers Razdolnaya and Uda, and Peace (USA). Due to the seasonal and interannual dynamics of phytoplankton development, the distribution of nutrients in the mixing zones of river and sea waters is also subject to seasonal and interannual variability.

The consumption of dissolved mineral phosphorus by biota occurs in the surface layer or in the vertically mixing water column throughout the entire salinity range and varies from 6 to 100% of its content in river runoff. The most intensive consumption of mineral phosphorus by aquatic organisms is observed during the vegetative season and is most often accompanied by the extraction from the solution of another biogenic element, silicon [71–74,82,83,91,92]. Along with this, in the mouths of some rivers (Strelna of the Kandalaksha Bay of the White Sea, Chernaya of the Sevastopol Bay of the Black Sea, and Razdolnaya of the Amur Bay of the Sea of Japan), the removal of dissolved mineral phosphorus was found even in winter, with relatively low biological activity.

Under stratification conditions, the plots of river and sea water mixing show the influence of two processes that regulate the concentration of dissolved mineral phosphorus at different depths. Either the behavior of phosphorus in surface waters is conservative, or it is removed as a result of biological assimilation, while in the deep layers, phosphorus, on the contrary, enters the solution due to the destruction of the organic matter deposited on the bottom. As a result, the relationships between the concentration of dissolved mineral phosphorus and the chlorinity (salinity) acquire orderliness only when the points are grouped along the horizons [77,78,103,106]. This distribution is typical for the estuaries of the Ob and Yenisei Rivers, in which aquatic organisms consume 6–57% of the mineral phosphorus supplied via river runoff, and the destruction of organic matter provides an increase in its concentration by 50–150%, relative to the content in the river water. A distinctive feature of the release of mineral phosphorus into the solution during the destruction of organic matter is the simultaneous entry into the water column of the mineral forms of nitrogen [77,96].

The predominance of destruction processes over production processes is also often found throughout the entire water column, periodically occurring in the mouths of the Onega River, some rivers of the Kandalaksha Bay of the White Sea, the Ashamba River of the Black Sea coast of Russia, and the Uda River of the Sea of Okhotsk, and leading to an increase in the flux of dissolved mineral phosphorus by 35–100%. An even greater transformation of river runoff under the influence of destruction processes occurs when pore solutions from stirred up sediments enter the mixing zone. This situation is typical for the mouths of the Mezen and the Ural Rivers and in some periods is observed in the mouths of the Black Sea rivers Anapka and Ashamba, where the additional input of mineral phosphorus into the solution exceeds its removal via river runoff by 90%, 150–300%, 7.0 times, and 9.5 times, respectively. Firstly, the organic matter in the bottom sediments is remineralized with the release of phosphates, and secondly, it can reduce iron (III) phosphates to iron (II) phosphates, which causes the input of part of the phosphorus into a dissolved state:

$$3\text{ }12\text{FePO}\_4 + 3\text{CH}\_2\text{O} + 38\text{H}\_2\text{O} = 4\text{Fe}\_3(\text{PO}\_4)\_2 \cdot 8\text{H}\_2\text{O} + 4\text{HPO}\_4^{2-} + 3\text{HCO}\_3^- + 11\text{H}^+. \tag{9}$$

The behavior of dissolved organic phosphorus characterizes the process of the destruction of organic matter in the mixing zone of river and sea waters. Its additional intake indicates the excess of the rate of the release of dissolved organic forms over the rate of complete phosphorus recycling and is observed in the mouths of the Northern Dvina and Salgir Rivers. A decrease in the concentration of dissolved organic phosphorus relative to the line of the conservative mixing of water masses occurs at a higher rate of its mineralization than its release into the solution during destruction and was noted in the mouths of the Kyanda River and rivers of the Black Sea coast of Russia (Anapka, Ashamba, Mezyb, and Hotetsai). Both losses and excesses of dissolved organic phosphorus were recorded at the mouth of the Onega River in different periods and at different salinity intervals. The distribution of dissolved organic phosphorus is close to conservative at the Mezen River estuary and the Serebryanka River mouth (Sikhote Alin Reserve).

A significant part of the phosphorus in suspended matter in rivers is represented by reactive mineral and organic compounds, which can be sources of dissolved phosphorus in the mixing zones of river and sea waters. From the experimental data [107], it follows that the amount of phosphorus passing from river suspensions into the dissolved state increases with an increasing salinity.

Some curves of the distribution of dissolved phosphorus concentration have a deflection in the salinity range from 0 to 9‰ [103]. In this case, the similar form of such relationships for dissolved iron and aluminum in the absence of clear signs of the biological consumption of phosphorus indicates the physicochemical removal of the latter from the solution as a result of coprecipitation with iron and aluminum hydroxides.

It can be expected that the terrigenous hydroxophosphates of iron and aluminum in the marine environment will be unstable due, firstly, to an increase in the pH value, accompanied by the displacement of phosphates by hydroxyl ions, and, secondly, to an increase in the concentration of dissolved calcium, which promotes the formation of apatite phases typical for ocean sediments. This assumption is confirmed by the results of observations [108], which showed that iron–calcium hydroxophosphates in oceanic pelagic sediments decompose during diagenesis into more stable iron oxyhydroxides and apatite. The replacement of the terrigenous hydroxophosphates of iron and aluminum with apatite should begin already in the river mouth areas and cause an increase in the concentration of dissolved phosphorus, according to the reaction

$$5\text{CaMe}\_n(\text{OH})\_{3n}\text{HPO}\_4 + \text{F}^- = \text{Ca}\_5(\text{PO}\_4)\_3\text{F} + 5n\text{Me}(\text{OH})\_3 + 2\text{HPO}\_4^{2-} + 5\text{H}^+, \tag{10}$$

where Me = Fe(III), Al. Therefore, the input of iron and aluminum phosphates into the salinized portion of the river mouth area as a component of the suspended matter and bed load can lead to a partial release of dissolved phosphorus and, in some cases, be the cause of its nonconservative behavior. From these positions, one can explain the decrease in phosphorus content with the increase in salinity in the bottom sediments of the mouths of the Pamlico and Potomac Rivers [109,110].

Phosphorus removal as a result of sorption and coprecipitation processes occurs, as a rule, at the initial stage of river and sea water mixing during the period of low biological activity and is observed in the mouths of the rivers Clyde (Great Britain), Mandovi and Chilka Lake (India), Mekong (Vietnam), Raritan (USA), and Amazon, accounting for 15–100% of the removal of dissolved mineral phosphorus via river runoff (Table 16). The input of dissolved phosphorus due to desorption from river suspensions penetrating into the marine environment is also a common phenomenon established for the mouths of the Black Sea rivers Mezyb and Hotetsai, Mekong in the area of medium salinity, Yangtze, Serebryanka (Sikhote Alin Reserve), Old Mill Creek (USA), Orinoco (Venezuela), Amazon at medium salinity, and Zaire (Table 16). The maximum desorption values (13–300% of the content in river water, or 1–75 μg P/L) are reached at a salinity of 7–15‰, and the mixing curves have a convex shape.

The spatial separation of the processes of phosphorus sorption and desorption (predominance of sorption in the freshwater part of the mixing zone, and desorption in the area of intermediate salinity) confirms the distribution of phosphorus and iron in the bottom sediments of the river mouth areas. Thus, at the Pamlico River mouth, the concentrations of phosphorus and iron in the bottom sediments of the riverine part of the mixing zone closely correlate with each other (*r* = 0.98), whereas when approaching the sea boundary of the mouth area, this relationship becomes less pronounced (*r* = 0.86–0.77). The same concentrations of iron correspond to lower phosphorus concentrations in the marine part of

the mouth, which indicates the release of the latter from the bottom sediments [111]. This combination of sorption and desorption in the estuaries is called the phosphate buffer mechanism [60,112–114]. Many authors have tried to determine the concentration of dissolved mineral phosphorus at which an equilibrium between the water and bottom sediments is established. According to experiments [99,112,113], the equilibrium concentrations of dissolved phosphates are in the range of 22–46 μg P/L. Convincing results of field observations proving the existence of such a concentration limit have not yet been obtained, although in the studied estuaries, with the exception of the Mekong River mouth, the concentrations of desorbed phosphorus do not really exceed the values recorded in the experiments and amount to 1–28 μg/L [82,89,94,102,104,105].

#### *4.2. Conservative Behavior of Phosphorus*

Despite the active participation of phosphorus in the intrabasin biological and chemical processes, cases of its conservative behavior were established (Table 16), which were observed either under conditions of the severe pollution of the aquatic environment (the mouths of the European rivers Clyde, Scheldt, Rhine–Meuse, Ems, Weser, Elbe, Rhone, and Tiber), or during periods of low biological activity (the mouths of the Mississippi River; Amazon River; the rivers of the Arctic; and the Far Eastern rivers Knyazhaya, Niva, Kolvitsa, Onega, Semzha, Indiga, Pechora, and Usalgin).

The concentration of dissolved mineral phosphorus in the waters of polluted rivers (93–520 μg/L) is an order of magnitude higher than its average content in the rivers of the world. Increased concentrations of dissolved mineral phosphorus were also found at the sea boundary of the mouth areas of these rivers: up to 242 and 280 μg/L on the near-shore zone of the Clyde and Tiber River mouths. The conservative behavior of phosphorus in the mouths of these European rivers can be explained by the fact that the absolute values of the fluctuations in its concentration in the mixing zones at such a high content in river or sea waters are comparable to the amount of phosphorus involved in the intrabasin processes. In addition, most of the observations in which the conservative behavior of phosphorus was recorded were carried out in the autumn–winter period, when the intensity of production processes in the temperate zone decreases with the intensification of the biological processes, and the conservative behavior of phosphorus can turn into nonconservative behavior within several weeks, which was noted, for example, for the Rhone River delta [115]. The conservative distribution of dissolved mineral phosphorus in the mouths of the Onega, Mississippi, and Amazon Rivers that appeared in the winter– spring period is also, apparently, caused by the low activity of aquatic organisms.

Separately, we should consider the conservative behavior of dissolved phosphorus during the vegetative season in the mouths of the small Arctic and Far Eastern rivers that are not subject to strong anthropogenic impact. The concentration of suspended matter in the mouths of these rivers in spring and early summer is small due to the slow thawing of soils in the catchments and the low water temperature preventing phytoplankton development, which limits the participation of phosphorus in physicochemical and biological processes.

Thus, the conservative behavior of dissolved phosphorus in the mixing zone of river and sea waters is an atypical phenomenon and occurs in special conditions when the biological and chemical processes at river mouths are suppressed as a result of an unfavorable combination of natural and anthropogenic factors.

#### *4.3. Phosphorus Balance in the Mixing Zones of River and Sea Waters*

An analysis of the mixing curves indicates the complex nature of the dynamics of dissolved phosphorus fluxes in river mouth areas, with the combination of conservative and nonconservative distribution and the spatiotemporal variability of the latter, including a change in the direction of transformation. Therefore, the calculations of the values of the removal or input of phosphorus in the mixing zones of river and sea waters based on data for relatively short time intervals turn out to be insufficiently representative for balance estimates.

To obtain more reliable estimates, Savenko and Zakharova [116] summarized the results of balance studies carried out in river mouths and bays for a year or more (Table 17). As follows from the data presented, on average, a significant part of the phosphorus is removed per year. The maximum removal (80–94%) is characteristic of the total phosphorus, including the suspended fraction (Pdissol + Psusp), and only a third of this value is associated with physical sedimentation, while the rest of the phosphorus is removed as a result of biosedimentation [117,118]. Biological processes also play a major role in the extraction of mineral phosphorus from the solution (40–80%) due to its transfer to the composition of suspended organic matter, which is subsequently deposited at the bottom. Total dissolved phosphorus (Ptotal) is retained in river mouths in much smaller amounts (7–38%). This is corresponds to the observational data presented in Table 16, according to which the losses of the dissolved forms of mineral and total phosphorus during biological consumption in the mixing zones of river and sea waters are on average 46 and 25%.



#### **5. Conclusions**

The formation of the chemical composition of surface waters begins already in the atmosphere during the interaction of aerosols with the condensates of water vapor: cloudy water and the water of atmospheric precipitation. The average median concentrations of mineral and total phosphorus in atmospheric precipitation are 15 and 33 μg/L, respectively; the values of the input of these forms into the earth's surface are equal to 0.11 and 0.25 kg/ha yr. The content of the soluble forms of total phosphorus in atmospheric precipitation is in the range of 20–80%, with an average value of 55%.

The average median concentrations of dissolved mineral and total phosphorus in the waters of the primary hydrographic network are 31 and 95 μg/L, respectively. The concentrations of both forms increase with an increasing anthropogenic load: natural (forest) catchments < agricultural–forest catchments with land use less than 50% < agricultural catchments with land use over 50% < urban catchments. The concentration of dissolved mineral phosphorus, all other conditions being equal, increases with an increase of the phosphorus content in the catchment rocks.

The average median concentrations of dissolved mineral and total phosphorus in world rivers are 28 and 85 μg/L, respectively. The minimum values are observed in the rivers of the Arctic and subarctic zone; the maximum values are found in the most densely populated temperate zone and the zone of dry tropics and subtropics. The anthropogenic load is a dominant factor for riverine export, which is confirmed by the presence of a direct relationship between the concentrations of mineral and total phosphorus, on the one hand, and the population density, on the other hand.

The distribution of dissolved mineral and total phosphorus in the mixing zones of river and sea waters in the overwhelming majority of cases corresponds to nonconservative behavior. The conservative type of distribution is rarely observed and is found in the mouths of polluted rivers with high phosphorus concentrations, which significantly exceed the possible changes that occur as a result of biological and chemical processes. The decreases in the fluxes of dissolved mineral and total dissolved phosphorus at the river–sea geochemical barrier are 40–80% and 7–38%, respectively.

**Author Contributions:** V.S.S. conceived the study; V.S.S. and A.V.S. jointly carried out the research and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** The study of the phosphorus geochemistry in the continental runoff of dissolved and solid matter was supported by the RFBR, project 18-05-60219; the study of the transformation of phosphorus fluxes in river mouths was carried out with the support of the RFBR, project 20-05-00802.

**Data Availability Statement:** Data supporting reported results can be found in the literature cited in the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Water* Editorial Office E-mail: water@mdpi.com www.mdpi.com/journal/water

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com

ISBN 978-3-0365-6745-7