Geochemical behavior of rare earth elements (REE) in urban reservoirs: the case of Funil Reservoir, Rio de Janeiro State, Brazil

Rare earth elements (REE) have unique chemical properties, which allow their use as geochemical tracers. In this context, the present study aims to assess the role of Funil Reservoir on REE biogeochemical behavior. We collected water samples upstream of the reservoir (P-01) in the city of Queluz, inside the reservoir (P-02), and downstream of Funil Reservoir (P-03) in the city of Itatiaia, RJ. In the field, physicochemical parameters were measured using a probe (pH, temperature, electrical conductivity, and dissolved oxygen). In the laboratory, water samples were filtered (0.45 µm) and properly packed until chemical analysis. Chlorophyll a concentrations were determined by a spectrophotometric method and suspended particulate matter (SPM) by a gravimetric method. Ionic concentrations were determined by ion chromatography technique and REE concentrations were determined by ICP-MS. Chlorophyll a concentrations were higher in Funil Reservoir. Ionic concentrations in Queluz (P-01) suggest anthropic contamination. The sum of REE in the dissolved fraction ranged from 2.12 to 12.22 µg L−1. A positive anomaly of La in Queluz indicates anthropic contamination. The observed patterns indicate that Funil Reservoir acts as a biogeochemical barrier, modifying the fluvial transport of REE. Nonetheless, another factor that probably influences REE behavior is the algal bloom that occurs in reservoirs during the rainy season. The seasonal behavior of algae can influence REE biogeochemistry through the incorporation and release of trace metals.

Many studies have been conducted in recent decades to better understand REE composition and fractionation in distinct marine environments, including estuaries, oceans, and hydrothermal veins (Haley et al., 2004), aquatic environments (Goldstein & Jacobsen, 1988;Elderfield et al., 1990;Dupré et al., 1996;Gailladert et al., 1997;Douglas et al., 1999;Viers et al., 2000;Tosiani et al., 2004;Xu & Han, 2009). Nonetheless, the geochemical controls on REE distribution in tropical environments are relatively scarce compared to their temperate counterparts. In the modern world, REE is largely employed in several technological applications owing to its unique magnetic, phosphorescent, and catalytic properties. The universal REE may result in their releasing and contamination in the environmental, while other human interventions in the landscape can alter natural erosion and runoff dynamics of rivers and estuaries, leading to changes in physicochemical properties, such as grain size and texture, and eventually influence the distribution of trace elements ( Gomes et al., 2013;Bisi et al., 2012Patchineelam et al., 2011. The anthropogenic factor in the natural cycling of the REE deserves better mitigation, particularly in fluvial courses modified by dams (Tang & Johannesson, 2010).
Human actions in drainage basins affect the pulse, magnitude, and nature of water flow and the transport of elements in fluvial waters. As an example, human activities such as logging, agriculture, and livestock production increase surface runoff and transport in drainage basins. Dams increase breath and water supply, as well as sediments and nutrients. Dams retain and transform materials, altering the natural flux of material, changing the capacity of transport (Souza et al., 2011).
Considering the influence of dams on biogeochemical cycles, this study aims to assess the hole of Funil Reservoir on REE biogeochemistry in this section of the Paraíba do Sul drainage basin. Indeed, REE has been frequently used to evaluate anthropogenic influences and sources for river waters or sediments (Benabdelkader et al., 2019;Bau & Dulski, 1996;Fuganti et al., 1996;Xu et al., 2012;Gallello et al., 2013). The natural distribution of REE in water, soil, and sediment from densely industrialized and populated regions can be altered by anthropogenic influences (Di Leonardo et al., 2009;Elbaz-Poulichet et al., 2002;Kulaksiz & Bau, 2007;Nozaki et al., 2000;Oliveira et al., 2003;Rabiet et al., 2009).
The geochemical interest in studying the behavior of ETR is justified by the history of contamination of the surroundings, which began in the seventeenth century with the development of the Paraíba do Sul Valley region and the installation of industrial activities in the region, marked by the establishment of the Companhia Siderúrgica Nacional (CSN).

Study area
Paraíba do Sul drainage basin, in southeastern Brazil, occupies 57,000 km 2 and covers important Brazilian states such as São Paulo, Rio de Janeiro, and Minas Gerais. This drainage basin is in a rural and metropolitan area, with the original Atlantic Forest restricted to parks and reserves. The course of Paraíba do Sul River began to be modified by fluvial transport in 1950, with the creation of Santa Cecília Reservoir, where waters from Paraíba do Sul were pumped (160 m 2 /s) to supply the metropolitan area of Rio de Janeiro (INEA, 2013). The hydro energy systems of Furnas are in this drainage basin area, represented by Funil Reservoir and Light Company, which control five reservoirs: Santa Cecília, Vigário, Santana, Tocos, and Lajes.
In Brazil, the main industrialized area of the country is in the states of São Paulo and Rio de Janeiro. A large percentage of the country's population, around 21%, lives in the metropolitan regions of these states (Torres et al., 2002). The Paraíba do Sul River is located between these two Brazilian regions, being the main river crossing the State of Rio de Janeiro. It is the main source of urban supply for the entire metropolitan region of Rio de Janeiro, which is home to about 10 million inhabitants. The river receives a lot of industrial and domestic waste from many cities and is also crossed by major roads and railways (Brito et al., 2005). The river is marked by successive dams designed to provide water and electricity to meet the energy and supply demands of the population of the Rio de Janeiro metropolitan region.
Funil Reservoir (Fig. 1) is in Paraíba do Sul River, in the city of Itatiaia, Rio de Janeiro (State). The Granite of Funil characterizes the geology of this region. This area is 50 km 2 in size and corresponds to a type I granite, incipiently deformed to not deform, monzogranite composition (Chappell & White, 2001). This granite is characterized by the presence of magnetite, pyrite, and molybdenite. The study area is distinguished by yellow dystrophic oxisols, sedimentary alluvial soils, and podzolic lithosols. The soil of the region is also recognized by its clay minerals typical of acid soils with high Al contents and low permeability (Governo do Estado do Rio de Janeiro, 1997).
The Funil Reservoir receives water from São Paulo from areas renowned for their industrial activities. The former, an important pole of the war industry, which concentrates metallurgical activities, hosts the largest aerospace complex in Latin America. The region houses large industries such as Johnson & Johnson, General Motors, Petrobras, Sony Eriksson, and Embraer, among others. Taubaté stands out as a large industrial and commercial hub, which houses companies such as Volkswagen, Ford, LG, Alston, and Usiminas.

Experimental settings
We collected subsurface water samples using a Van Dorn water sampler at three different sampling points: (P-01) located 18 km upstream of the reservoir, in the city of Queluz, São Paulo (State); (P-02) located in Funil Reservoir; and (P-03) located 6 km downstream of the reservoir, in the city of Itatiaia, Rio de Janeiro (State). In situ, physicochemical data were measured using a probe (Hanna Instruments -HI9829). The samples were taken at subsurface, 1.5 m depth. At each of the points, 5 L of water was collected and stored in previously sanitized flasks. Then, water samples were correctly packed into Teflon flasks and transported to the laboratory in ice coolers.
In the laboratory, water samples were filtered through cellulose acetate membranes (0.45 µm) to obtain the respective dissolved (< 0.45 µm) and particulate fractions (> 0.45 µm). After the filtration step, ion concentrations in dissolved fractions were determined using ion exchange chromatography (Dionex ICS-2000 Ion Chromatograph with Dionex AS40 Automated Sampler). The anionic species measured were fluoride (F − ), chloride (Cl − ), bromide (Br − ), nitrate (NO 3 − ), sulfate (SO 4 2− ), and phosphate (PO 4 3− ). Gradient separations were performed with KOH eluent, AS-19 column, AG-19 guard column, and ASRS-300 electrolytic suppressor (all Dionex and 2 mm). A Fluka multi-element standard diluted to a final concentration of 6 mg/L using ultrapure water was used as the standard for certifying the analyses using ultrapure water and analyzed under the same conditions as the samples. The concentrations of the   175 Lu, and 232 Th. The analysis was run with Octopole Reaction System (ORS) equipped with a low-flow MicroMist nebulizer connected to a refrigerated quartz Scot mist chamber. Nickel samplers and skimmer cones were used. Rhodium (Rh, 5 µg.L −1 ), luterius (Lu, 5 µg.L −1 ), and indium (In, 5 µg.L −1 ) elements were used in the internal standardization test from single-element standards of concentration 1000 mg L −1 (SPC Science, Canada). The determination of Fe isotope m/z 56 was made using the ORS system with the introduction of He gas for the elimination of ArO interference.
All reagents used in ion chromatography and ICP-MS analysis were of analytical grade. The solutions were prepared with ultrapure water (resistivity < 18.2 MΩ cm) obtained from a Milli-Q reference system (Millipore).
For validation of the analytical method for the determination of REE concentrations, the certified reference material SPS-SW1 (surface water level 1-Batch 117, LGC Standards) was used. A MCR was prepared with a concentration of 0.5 µg/L, the mean value measured 0.620 µg/L, standard deviation of 0.008, and relative error of 29%.
The acetate cellulose membranes with particulate fraction were designed to quantify the SPM by using gravimetric method (Carmouze, 1994). Chemical determination of chlorophyll a in filters was performed by a spectrophotometric method (Shimadzu UV-1800) after 90% acetone extraction (Carmouze, 1994).
The data were submitted to the non-parametric statistical test of Kruskal-Wallis and post hoc Dunn's test to establish the existence of space-time variability among the measurements performed at the sampling points. A 5% significance level was adopted for the tests. Additionally, a simple linear correlation matrix was drawn up between the hydrochemical variables (p = 0.05) to explain the biogeochemical behaviors identified in the study.   Table 1 summarizes all physicochemical data. Overall, a similar electrical conductivity and TDS concentration behavior was observed from upstream (P-01) to downstream (P-03). According to the Kruskal-Wallis statistical test, there is a significant difference between the variables evaluated during the dry and rainy seasons in point Queluz (P-01) (p < 0.05, p = 0.0277); this data suggests that there are differences between the biogeochemical behaviors of this point in the dry and rainy seasons. This difference in biogeochemical behavior was not observed for points P-02 Funil Reservoir and P-03 Itatiaia. The suspended particulate matter (SPM) concentrations were highest at P-01 Queluz (35.0 mg.L −1 ), intermediate at the P-02 Funil Reservoir (32.0 mg. L −1 ) and lowest at P-03 Itatiaia (30.0 mg.L −1 ). The highest chlorophyll a concentrations were found in P-03 Funil Reservoir (16 µg.L −1 ) and lower concentrations were found at upstream (4.0 µg.L −1 ) and downstream (5.4 µg.L −1 ) points. Chlorophyll a concentration detaches from the other physicochemical data with concentrations four times occurring in Funil Reservoir (P-02). Furthermore, according to CONAMA 357/2005 (the Brazilian resolution that establishes the physicochemical parameters for water quality), values of chlorophyll a concentration within Funil Reservoir (P-02) are above the reference value (30.0 µg.L −1 ). This behavior probably reflects seasonal algal blooms in Funil Reservoir, as reported by other authors (Barbosa, 2005;Rudorff et al., 2007;Londe, 2008;Costa, 2009).
Analyzing the statistical correlation matrix, we identified a strong negative correlation between pH, TDS, and ionic species present in the medium and a strong positive correlation with REE and chlorophyll a concentration. This pattern can be explained by the influence of pH on the biogeochemical behavior of the chemical species present in the medium, by the action of ionic strength, and by the process of ion desorption that may be happening in the SPM.
Ion concentrations are presented in Table 2. Comparing the three sample stations, the high concentrations of chloride nitrate and sulfate found in Queluz (P-01) suggest an anthropic influence, considering the different sources of chemical elements of anthropogenic origin that are found in the region, and described in study area. In Funil Reservoir (P-02), ionic concentrations reduced, likely due to ion incorporation by algae or water column scavenging, which increases seasonally in the reservoir (Barbosa, 2005;Rudorff et al., 2007;Londe, 2008;Costa, 2009). The strong positive correlation between the electrical conductivity of waters and the concentrations of different ionic species, as well as the negative correlation with REE concentrations, indicates that the concentration of chemical species present in the solution exert control over conductivity.
In line with previous studies (Gomes et al., 2013), iron and aluminum concentrations in the dissolved phase (Fig. 2) seem to be derived from the weathering of aluminosilicate mineral assemblies of the material from the middle Paraíba do Sul watershed geology. These high concentrations are probably due to the entrainment of soil and erosion of fluvial margins. However, the lowest concentrations of Fe, Al, and Mn in Funil Reservoir (P-02) indicate that this reservoir could acts as a biogeochemical barrier, promoting the removal of these elements via sorption on surface particles and subsequent  decantation (Von Sperling, 1999). Concentrations of Fe, Al, and Mn in Itatiaia (P-03) indicate the influence of drainage basin washing, as was found in Queluz (P-01) (Gomes et al., 2013). Evaluating the SPM fraction (Fig. 3), Al concentration was more elevated upstream (P-01) and downstream (P-03) of the reservoir, and there is a significant reduction of Al inside Funil Reservoir (P-02). Accordingly, Al data probably indicates clay mineral deposition inside the Funil Reservoir (P-02).
The sum of REE concentrations in the dissolved fraction in the Funil Reservoir sample (Table 3) was four to six times higher than other sample stations. In all of them, light REE concentrations were higher than heavy REE, especially in Funil Reservoir (P-02). Such a pattern agrees with experimental studies conducted in lacustrine environments, where hydrodynamical conditions favor a depletion of medium and heavy REE typically observed (Sultan & Shazili, 2009).
Total REE concentration normalized by PAAS (Post Archean Australian Shales) in water samples reveals geochemical fractionation, characterized by a slight enrichment of heavy REE (Fig. 4).
Together, particulate and dissolved fractions evidence processes occurring in the water column of the Funil Reservoir that modify the dynamics of these elements. The formation of aqueous complexes, biological uptake, and adsorption to colloids are potential processes responsible for REE fractionation between dissolved and particulate phases (Ronnback et al., 2008).  In dissolved fractions, REE curves overlap in shape and magnitude, while in the SPM, Funil Reservoir samples distinguish for neighbor sampling stations with lower normalized values. This pattern could be explained by two reasons: the first is the incorporation of light REE by microalgae (explained by the strong positive correlation between chlorophyll a and REE concentration) that develop seasonally in Funil Reservoir according to many authors (Novo et al., 2004;Barbosa, 2005;Rudorff et al., 2007;Londe, 2008;Costa, 2009); and the second is particle sedimentation in Funil Reservoir due to hydrodynamic processes and physicochemical conditions. The role of REE incorporation by microalgae present in the Funil Reservoir can be confirmed by the strong positive correlation identified between chlorophyll and REE concentrations.
Algae are great accumulators and have been experimentally used for REE recovery, which are metals with high technological value destined for the use of high technology industries. The metal bioavailability for algae depends upon many factors: physicochemical parameters, pH, salinity, luminosity, SPM concentration, and organic matter contents (Karez et al., 1994). Furthermore, according to Karez et al. (1994), competition for dissolved metals in algal cell sites depends on biological factors.
Notwithstanding, Valitutto et al. (2006) demonstrated that biota have a keyhole in REE removal from the water column. Other authors identified correlations between REE concentrations in the dissolved phase and biota. In this way, some authors identify that the capacity of REE absorption from the soluble phase is probably related to physicochemical similarities between REE and Ca, which is an essential element for plants (Lakatos et al., 1999).
Observing the graphic in Fig. 5, we can identify two anomalies: Tm and Eu. Tm anomaly could be justified by the geological settings of the study area, whose minerals are sources of REE, and thus reflect weathering activity in drainage basins (Pereira et al., 2001).
According to Sultan and Shazili (2009), Eu anomalies are observed in most rivers and reflect the redox potential chemistry of this element (Ryu et al., 2007). This anomaly was higher in Funil Reservoir (P-02) when compared with other sample points; consequently, in Funil Reservoir (P-02) Eu is released and associated with the dissolved phase. We hypothesize that algal blooms leading to anoxic conditions in sediments may favor the Ce releasing through water pore diffusion.
Positive anomalies of Ce occur typically in tropical environments, where high temperatures are constant throughout the year, combining with precipitation, according to Sultan and Shazili (2009) (Table 4), probably due to reducing conditions that favor Ce anomalies. And analyzing La anomalies, high positive anomaly of La found in Queluz (P-01) suggests anthropic contamination (Bau, 1999), considering that this point receives water from a densely populated area of Brazil and with significant industrial activity.

Conclusion
According to the presented results, Funil Reservoir acts as a biogeochemical barrier, modifying the fluvial transport of rare earth elements associated with particulate and dissolved fractions of waters from the middle Paraíba do Sul River. Another factor that might influence REE biogeochemical behavior is the seasonality of algae. These algae can act through the uptake and release of REE, which behave exactly like other metals. Moreover, hydrochemical conditions can influence the dynamics of REE along this section of Paraíba do Sul River. The positive La anomaly suggests anthropic influence in Queluz (P-01).