(Al, Fe, Mn, Zn, Cu, and Cr) Elements Along the Lower Orinoco River

Received: 3 August 2016 Accepted: 5 October 2016 DOI 10.1002/hyp.11051

RESEARCH ARTICLE

Dynamics of dissolved major (Na, K, Ca, Mg, and Si) and trace (Al, Fe, Mn, Zn, Cu, and Cr) elements along the lower Orinoco River

1 Centro del agua para América Latina y el Caribe, Tecnológico de Monterrey, Av. Abstract Eugenio Garza Sada Sur No. 2501, CP 64849 This study addresses the changes in dissolved major and trace element concentrations along the Monterrey, Nuevo León, México Orinoco River, including the mixing zone between the Orinoco and Apure Rivers. Water samples 2 Escuela Técnica Superior de Ingenieros de from the Apure and Orinoco Rivers were collected monthly in four sectors over a period of Minas, Universidad Politécnica de Madrid, 15 months. Auxiliary parameters (pH, dissolved oxygen, conductivity, and temperature), total Madrid 28003, Spain suspended sediments, dissolved organic carbon (DOC), and major (Na, K, Ca, Mg, and Si) and trace 3 GET–UMR CNRS/IRD/UPS–UMR 5563 du CNRS, UR 234 de l’IRD; OMP, 14 Avenue (Al, Fe, Mn, Zn, Cu, and Cr) element concentrations were measured in all sectors. The relative con- Edouard Belin, 31400 Toulouse, France tribution of both rivers after the Apure–Orinoco confluence was determined using Ca as a tracer. 4 Centro de Oceanología y Estudios Antárticos, Moreover, a mixing model was developed to determine whether dissolved species exhibit a con- Instituto Venezolano de Investigaciones servative behavior during mixing. The results indicate that DOC is removed from waters during Científicas (IVIC), Caracas 1020‐A, Venezuela the Apure–Orinoco mixing, probably due to absorption of DOC on mineral phases supplied by 5 Laboratorio de Fisicoquímica, Estación de Investigaciones Hidrobiológicas de Guayana, the Apure River. Dissolved Na, Ca, and Mg behave conservatively during the mixing processes, Fundación La Salle de Ciencias Naturales, San and their concentrations are controlled by a dilution process. The anomaly in the temporal pattern Félix 8051, Venezuela of K in the Orinoco is caused by the input of biogenic K originating from the Apure River during Correspondence the high‐water stage. The loss of dissolved Si during the low‐water stage can be explained by the Abrahan Mora, Centro del Agua para América ‐ Latina y el Caribe. Tecnológico de Monterrey, uptake of Si by diatoms. Dissolved Mn, Zn, Al, and Fe showed a non conservative behavior during Av. Eugenio Garza Sada Sur No. 2501, CP the Apure–Orinoco mixing. The removal of Mn and Zn from the dissolved phase can be explained 64849 Monterrey, Nuevo León, México by the formation of Mn‐oxyhydroxides and the scavenging of Zn onto Mn oxides. Dissolved Fe is Email: [email protected] controlled by redox processes, although the removals of Fe and Al due to the preferential adsorption of large organometallic complexes by mineral surfaces after the Apure–Orinoco confluence can affect the mobility of both elements during transport. The conservative behavior shown by Cu and Cr can be related to the tendency of both elements to be complexed with small organic colloids, which are not preferentially adsorbed by clays.

KEYWORDS

major elements, mixing zone, organic matter, Orinoco River, trace elements

1 | INTRODUCTION tributaries have been classified based on their optical appearance and their physicochemical properties in “whitewater,”“clearwater,” and The Orinoco River ranks third in terms of water discharge to the “blackwater” rivers (Vegas‐Vilarrubia, Paolini, & Miragaya, 1988). oceans, with an annual mean discharge of 36.000 m3 s−1 (Lewis & Whitewater rivers originating from the Andes show near neutral pH Saunders, 1989). Like the Amazon River tributaries, the Orinoco River values and high concentrations of total suspended sediments (TSS)

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. © 2016 The Authors Hydrological Processes Published by John Wiley & Sons Ltd.

Hydrological Processes 2017; 31: 597–611 wileyonlinelibrary.com/journal/hyp 597 598 MORA ET AL. and major cations such as Na, K, Ca, and Mg. These characteristics are the processes controlling the content of dissolved trace elements in due to the reaction‐limited weathering regime and the diverse mixture these tropical environments. Moreover, the behavior of organic matter of silicates, carbonates, and evaporites in the Andes (Edmond, Palmer, and dissolved elements in mixing zones has been documented in large Measures, Brown, & Huh, 1996). Conversely, “blackwater” rivers orig- rivers such as the Amazon and Yangtze (Aucour et al., 2003; Ran, Yu, inating from the Guayana Shield have low concentrations of major ions Yao, Chen, & Mi, 2010). However, this issue has not yet been docu- and TSS and low pH values, mainly due to the presence of organic mented in the Orinoco. Because of the lack of this information, we acids originating from forest soils that have very low buffering capacity can formulate the following questions: Why do the concentrations of because of the igneous rocks that predominate in the blackwater K and Si in the lower Orinoco River not show a unimodal seasonal pat- drainages. Because of the marked difference in water chemistry tern, as shown by the other major elements? Could major and between both typologies of river waters, there is a chemical heteroge- dissolved trace elements show a conservative behavior during the neity between the left and right banks within the Orinoco mainstream mixing process between the Orinoco and a “whitewater” river originat- because “whitewater” rivers flow to the left bank and “blackwater” riv- ing from the Andes? Which processes could control the content of ers flow to the right bank. This lateral asymmetry, which has been dissolved trace elements in this riverine mixing zone? Considering that noted by many authors (Lewis & Saunders, 1984; Laraque et al., mixing is an important natural phenomenon that can affect the chem- 2013; Stallard, 1987), is maintained in the Orinoco from the conflu- ical composition of water due to different reactions taking place ence with the Guaviare River in the upper Orinoco to the river mouth between the dissolved and particulate phases, understanding the due to the continuous and heterogeneous deliveries from one river major and trace element dynamics in the Apure–Orinoco mixing zone bank to the other (Figure 1). during a hydrological cycle is in prime importance in understanding The temporal variability of dissolved major elements in the lower the biogeochemical cycling of these elements in tropical environments. Orinoco River has been well documented by several authors. Elements Therefore, the objective of this work was to investigate the tem- such as Na, Ca, and Mg are mainly controlled by a dilution process, and poral variation and behavior of dissolved organic carbon (DOC) and they show a unimodal regime, with high concentrations during low dissolved major (Na, K, Ca, Mg, and Si) and trace (Al, Fe, Mn, Zn, Cu, water and low concentrations during the high‐water stage (Lewis & and Cr) elements along the lower Orinoco River mainstream. We Saunders, 1989). However, elements such as K and Si have shown a emphasized the mixing zone between the Apure River (a “whitewater” non‐defined seasonal pattern, with little temporal variations along river originating from the Andes) and the Orinoco River in order to the hydrological cycle. Despite the numerous works regarding the geo- identify the diverse processes that may affect the geochemistry of chemistry of major ions in the Orinoco waters (Edmond et al., 1996; the studied dissolved elements. Thus, as a novel issue in comparison Lewis & Saunders, 1989, 1990; Stallard, Koehnken, & Johnsson, with other works carried out in the Orinoco, we compared the mea- 1991), the geochemistry of dissolved trace elements in the Orinoco sured concentrations of elements after the mixing zone with those pre- and its tributaries has been poorly investigated because most of the dicted through a conservative mixing model to determine whether the published works, based on one sampling campaign, do not describe dissolved elements show a conservative behavior during mixing.

FIGURE 1 Map showing the sampling sectors in the Apure and Orinoco Rivers MORA ET AL. 599

2 | MATERIALS AND METHODS The sampling sites were chosen inside the four sectors distributed between the Apure and Orinoco Rivers (Figure 1). Sector 1 was 2.1 | Study area, hydrological settings, and sampling located in the Apure River, Sector 2 was located in the Orinoco before the confluence with the Apure River, Sector 3 was located in the Ori- sites noco after the confluence with the Apure River, and Sector 4 was 6 2 The Orinoco River basin covers an area of 1 × 10 km . This basin can be located in the lower Orinoco River after the confluence with the divided into three geological regions: (i) the Andes, which comprises oro- Caura River (a “blackwater” tributary originating from the Guayana genic mountain belts; (ii) the Llanos, an extensive region built by sedi- Shield) and before the Puerto Ordaz City. Given the chemical hetero- ments originating from the Andes; and (iii) the Guayana Shield, a deeply geneity shown by the Orinoco between the left and right banks, two eroded crystalline bedrock zone dominated by intrusive Precambrian sampling sites were chosen for each sector in the Orinoco River. igneous rocks (Figure 1). The Orinoco River’s main channel is fringed by These two sampling sites were located approximately 800 m from a network of permanent lagoons and fluvial wetlands called the flood- each bank. The approximate width of the Orinoco River’s main chan- plain, which has an average width of 9 km and covers an area of nearly nel was 3.7, 5, and 2.2 km for Sectors 2, 3, and 4, respectively. Table 1 2 7,000 km when the Orinoco River reaches high water discharges (Ham- shows the codification and the location of the seven sampling sites. ilton & Lewis, 1990). The climate in the basin is seasonally tropical, with a rainy season extending from May to November and a dry season from | December to April. This rainfall regime produces a unimodal regime in 2.2 Collection, treatment, and preservation of the hydrological cycle of the Orinoco River and its main tributaries, with samples. ‐ high discharge (high water stage) between July and September and low Water samples were collected monthly from a small boat between ‐ flow (low water stage) between February and April (Figure 2). June 2007 and August 2008. Given the large distance between the sampling sites, each sampling campaign was carried out over 3 days in the third week of each month. Sectors 1 and 2 were sampled on (a) Tuesdays, Sector 3 was sampled on Wednesdays, and Sector 4 was sampled on Thursdays. The Caura River was also sampled during this period. However, although data of the Caura and Apure Rivers have been published previously (Mora, Alfonso, Baquero, Balza, & Pisapia, 2010a; Mora, Baquero, Alfonso, Pisapia, & Balza, 2010b), we used those data in order to carry out comparisons between the Orinoco and its tributaries. The water samples were collected 50 cm below the surface in prewashed polyethylene bottles (4 L), which were placed in a cooler with ice for transportation. Auxiliary parameters (pH, conductivity, temperature, and dissolved oxygen concentrations) (b) were measured in situ using pre‐calibrated electrodes. Alkalinity was also measured in the field by titration to pH 4.5. All samples were filtered in the laboratory on the day following sampling through pre‐ washed 0.2‐μm acetate cellulose filters. The first 50 ml of filtrate was discarded. The filtered samples were placed in sampling kits and then preserved with ultrapure nitric acid for major and trace element analyses. The samples for Si analyses were preserved with a solution of 1 M of NaOH. For Cr determinations, 500 ml of filtrate was pre‐ concentrated to 50 ml at 80 °C by adding 3 ml of ultrapure nitric acid to obtain samples with concentrations higher than the detection limit given for the analytical technique. All bottles used for sampling and (c) storage were pre‐cleaned with nitric acid and MilliQ deionized water prior to use. The samples for the DOC analyses were preserved with phosphoric acid and stored in dark glass bottles pre‐washed with sul- furic acid and MilliQ deionized water.

2.3 | Analytical measurements

Certified standard solutions were used to prepare five‐point calibration curves for major and trace element determinations. All calibration curves covered a concentration range of one order of magnitude. Na, K, Ca, Mg, FIGURE 2 Temporal variation of water discharge in the (a) Orinoco, (b) and Si were measured in triplicate by flame atomic absorption spectrom- Apure, and (c) Caura Rivers during the years 2007 and 2008 etry using a GBC Avanta instrument (model 908G). Na and K were 600 MORA ET AL.

TABLE 1 Codification and location of the water sampling sites in the Apure and Orinoco Rivers

Sector River Bank Sampling Site Latitude Longitude

1 Apure Center APU 7°39′20″N 66°27′49″W 2 Orinoco Right OR1 7°35′52″N 66°25′16″W 2 Orinoco Left OR2 7°34′53″N 66°25′16″W 3 Orinoco Right OR3 7°38′47″N 64°55′03″W 3 Orinoco Left OR4 7°40′13″N 64°54′29″W 4 Orinoco Right OR5 8°16′25″N 62°54′46″W 4 Orinoco Left OR6 8°16′56″N 62°54′43″W

measured by air‐acetylene flame, whereas Ca, Mg, and Si were measured of the Caura and Apure Rivers have already been published in separate by nitrous oxide–acetylene flame. A total of 2,000 mg L−1 of Cs was papers on the geochemistry of these rivers (Mora et al., 2010a; Mora added to the samples and standards used for the determination of Na, et al., 2010b). These results, together with the studies performed in K, Ca, and Mg to prevent analyte ionization. Additionally, K (2,000 mg L the Orinoco, Caura, and Apure Rivers in the 1980s (Lewis, Hamilton, −1) was used as an ionization suppressor for Si determinations. Trace ele- Jones, & Runnels, 1987; Lewis & Saunders, 1990; Saunders & Lewis, ments (Al, Fe, Mn, Zn, Cu, and Cr) were measured in triplicate by graphite 1989), represent a unique time‐series of data of major and trace ele- furnace atomic absorption spectrometry (GBC Avanta GF3000) using a ments measured along the Orinoco River and its tributaries, and they deuterium lamp for non‐atomic correction. A certified standard solution show the changes in the chemical properties of the Orinoco down- was used to prepare DOC standards for the calibration curve. The stream of the Apure–Orinoco confluence. It can be seen that our data DOC concentrations were measured in standards and filtered samples of dissolved major elements in Sector 4 are in agreement with those using a Tekmar Apollo 9000 combustion TOC analyzer. TSS concentra- reported by Lewis and Saunders (1989). However, the trace element tions were calculated by weight difference with filters dried at 60 °C concentrations in Sector 4 were lower than those reported previously for 24 hr. The relative standard deviations of the element analyses (Mora et al., 2009), probably due to the differences in the pore size of ranged 0.3–5% for Na, Ca, and Mg; 0.5–6% for K, Mn, and Fe; 0.6– the filters used in both studies (0.45 vs. 0.2 μm), which can affect the 10% for Si; and 1–10% for Al, Zn, Cu, and Cr. The DOC determinations trace element content in filtered waters. showed an uncertainty better than 10%. Figures3–5depictthespatiotemporalvariabilityofthechemicalvar- The international geostandard SRM‐1643e (trace elements in iables measured in the Orinoco River during this study. The dissolved water certified by the National Institute of Standards and Technology) oxygen concentrations (Figure 3c) did not show clear spatial differences; was used to check the accuracy of the atomic absorption analyses. The however, a high temporal variation was found in all sectors. This variable −1 results of the geostandard measurements and the detection limits showed lower values at the high‐water stage (close to 4.5 mgO2 L ) reported for the atomic absorption analyses have been published else- probably due to the consumption of dissolved oxygen by bacterial respi- where (Mora et al., 2010a). ration during the degradation of large pools of organic matter (Amon & Benner, 1996) provided by the floodplain at this stage. The seasonal 2.4 | Statistical analysis TSS pattern (Figure 3d) is in agreement with those noted by other authors (Lewis & Saunders, 1989; Meade, Weibezahn, Lewis, & Pérez, ’ ‐ The Student s t test (Miller & Miller, 1989) was used to compare concen- 1990), with low values at the low‐water stage and the maximum value trations of the studied elements between banks in each sector of the Ori- in May, due to the resuspension of fine‐grained sediments stored in the noco River. A principal component analysis (PCA) of the results obtained channel beds when the river begins to rise. in the Orinoco River waters was performed to investigate the relation- The spatiotemporal pattern of DOC depicted in Figure 3e shows ships among the variables. Prior to multivariable analysis, the data were that it had maximum concentrations at the high‐water stage due to ‐ log transformed, andthe variables werestandardized bycalculating their DOC in the Orinoco mainly being derived from microbial decay of veg- ‐ standard scores (z scores). This approach approximates normality and etation in the forests and savannas that are flooded during this period gives the same weight to all variables. For geochemical data, an orthogo- (Medina, Francisco, Sternberg, & Anderson, 2005). The pH; conductiv- nal method should be chosen for factor rotation during PCA (Reimann, − ity; and Na, Ca, Mg, and HCO3 concentrations (Figures 3a and b, and Filzmoser, & Garrett, 2002). Therefore, the varimax method was used. 4a, c, d, and f, respectively) showed their maximum values during the The PCA was conducted using the Statistica 5.0 statistical program. low‐water stage and their lowest values during the high‐water stage, suggesting that these variables are controlled by a dilution process. 3 | RESULTS AND DISCUSSION Sector 2 showed a slight heterogeneity between banks (OR1 and OR2), with significant differences between both banks (p < 0.05) for Na, K, Ca, and Mg concentrations. This heterogeneity is due to the 3.1 | Temporal and spatial variability of selected continuous input of high quantities of Na, K, Ca, Mg, and HCO− from variables along the Orinoco River 3 whitewater tributaries such as the Meta and Guaviare Rivers toward The results of the measured variables in all selected sectors of the the left bank of the Orinoco (Figure 1). The highest chemical heteroge- Orinoco River throughout this study are provided in Table S1. Data neity between banks was found in Sector 3. This sector showed MORA ET AL. 601

(a) (b)

(e) (f)

FIGURE 3 Temporal and spatial variability of (a) pH, (b) conductivity, (c) dissolved oxygen, (d) total suspended sediments (TSS), (e) dissolved organic carbon (DOC), and (f) temperature in the Orinoco River waters at Sectors 2 (OR1 and OR2), 3 (OR3 and OR4), and 4 (OR5 and OR6) during the studied period significant differences (p < 0.05) in Na, K, Ca, and Mg concentrations (p < 0.05) after the Apure–Orinoco confluence (Sector 3). Dissolved due to the Apure River considerably increasing the values of the afore- Mn and Zn showed a high heterogeneity between banks in Sectors 2 mentioned variables in the Orinoco. Unlike the other major ele- and 3 (significant differences were found for both elements in ments, the Si and DOC concentrations did not show significant these sectors) during most of the studied months. differences between the right and left banks in Sectors 2 and 3. Sector 4 exhibits high homogeneity between banks (no significant 3.2 | Principal component analysis (PCA) differences were found for major elements and DOC concentrations) due to this sector being located after Ciudad Bolívar, where PCA is a statistical method that reduces the dimensionality of the data a reduction of the main channel and bedrock outcrops of the while retaining most of its variation. It carries out this reduction by Guayana Shield provoke flow vortexes and turbulences that identifying vectors (principal components) in which the variation in cause a certain homogenization of the channel waters (Laraque the data is maximal. In this study, PCA captures the important et al., 2013). features inherent in the variability of dissolved elements along the Ori- All of the studied trace elements showed their minimum noco River and helps to recognize patterns controlling the content of concentrations during the low‐water stage. Dissolved Mn, Zn, and Al these elements. Applying PCA to the large dataset of the Orinoco (Figure 5a, b and c, respectively) had their maximum concentrations River, only two important components were significant (Figure 6). before the Apure–Orinoco confluence (OR1 and OR2), whereas the The first, which describes most of the sample variance (40 %), had high Cu and Cr concentrations (Figure 5e and f) only presented slight positive loads for the variables controlled by a dilution process (Na, K, − variations between the sampling locations. The Al, Cu, Cr, and Fe Ca, Mg, HCO3, pH, and conductivity) and high negative loads for Mn concentrations did not show significant differences between banks in and Zn. The second component describes 28% of the sample variance Sector 2 in most of the studied months. However, the Al and Cu and has high negative loads for DOC, Cr, and Cu, suggesting that DOC concentrations showed significant differences between banks controls the content of both elements. Although Fe seems not to be 602 MORA ET AL.

(a) (b)

(e) (f)

FIGURE 4 − Temporal and spatial variability of dissolved (a) Na, (b) K, (c) Ca, (d) Mg, (e) Si, and (f) HCO3 concentrations in the Orinoco River at Sectors 2 (OR1 and OR2), 3 (OR3 and OR4), and 4 (OR5 and OR6) during the studied period associated with a particular group of variables, it tends to be positively assessment was carried out in all months during this study, demon- correlated with redox‐sensitive elements such as Mn and with DOC. strating no marked input from other water sources such as floodplain lakes and small streams during the hydrological cycle. Contributions f3 and f4 of the Apure waters to sites OR3 and OR4 3.3 | Conservative mixing model (respectively) of the Orinoco River can be calculated using Ca as a Although major elements have shown a conservative behavior during tracer through the following equations: transport and mixing (Ran et al., 2010), Ca could be precipitated during C ¼ C :f3 þ C :ðÞ1−f3 (1) transport. However, there is no tendency toward calcite precipitation OR3 APU OR1 in the Orinoco and Apure Rivers as shown by our calculations based ¼ : þ :ðÞ− on the water quality data of both rivers. Additionally, because the Ca COR4 CAPU f4 COR2 1 f4 (2) content in the particulate phase can be removed by the weathering of clays during the mixing process, the enhancement of Ca in the where COR1,COR2,COR3,COR4, and CAPU are the Ca concentrations dissolved phase by this erosive process is negligible (Aucour et al., at sites OR1, OR2, OR3, OR4 and in the Apure River, respectively. The 2003). Indeed, these facts indicate that dissolved Ca shows a conserva- uncertainties associated with the contribution of the Apure River tive behavior during the Apure–Orinoco mixing and can be used as a toward the Orinoco River at sites OR3 (Δf3) and OR4 (Δf4) can be tracer. Because the Orinoco River in Sector 2 is mainly composed of calculated using the partial derivative method: waters originating from the Guayana Shield, there is a marked chemical 1 C −C Δf3 ¼ :ΔC þ OR3 APU :ΔC (3) difference between the Orinoco and Apure Rivers. Compared with the − OR3 2 OR1 CAPU COR1 ðÞCAPU−COR1 Orinoco, the Apure River is more enriched in major ions such as Ca and − – COR1 COR3 Na. However, after the Apure Orinoco confluence, the Apure waters þ :ΔCAPU ðÞC −C 2 are strongly diluted by the Orinoco waters. The theoretical mixing line APU OR1 between waters from the Apure and Orinoco Rivers during June 2007 1 C −C Δf4 ¼ :ΔC þ OR4 APU :ΔC (4) is shown in Figure 7. The mixing line is defined by the composition of − OR4 2 OR2 CAPU COR2 ðÞCAPU−COR2 two end members and can be used to assess the conservative behavior COR2−COR4 of dissolved species. Water samples taken after the mixing zone during þ :ΔCAPU ðÞC −C 2 June 2007 (sites OR3 and OR4) align well on the mixing line. This APU OR2 MORA ET AL. 603

(a) (b)

(e) (f)

FIGURE 5 Temporal and spatial variability of dissolved (a) Mn, (b) Zn, (c) Al, (d) Fe, (e) Cu, and (f) Cr concentrations in the Orinoco River at Sectors 2 (OR1 and OR2), 3 (OR3 and OR4), and 4 (OR5 and OR6) during the studied period

FIGURE 6 Results obtained from principal component analysis carried FIGURE 7 Ca versus Na concentrations in water samples taken from out on the dataset of the Orinoco River. TSS, total suspended the Apure and Orinoco Rivers before and after the Apure/Orinoco sediments; DOC, dissolved organic carbon confluence where ΔC , ΔC , ΔC , ΔC , and ΔC are the standard OR1 OR2 OR3 OR4 APU element. The uncertainties of the predicted concentrations for each deviations of the Ca concentrations measured at sites OR1, OR2, OR3, element were calculated through the following equations obtained OR4 and the Apure River, respectively. from the partial derivative method: Once the contributions f3 and f4 are obtained, the predicted concentrations for each studied element at sites OR3 (COR3) and OR4 ΔCOR3 ¼ jj1−f3 :ΔCOR1 þ jjf3 :ΔCAPU þ jjCAPU−COR1 :Δf3 (5)

(COR4) can be computed using the same equations (1 and 2), substitut- Δ ¼ jj− :Δ þ jj:Δ þ jj− :Δ ing the concentration of Ca for the concentration of the studied COR4 1 f4 COR2 f4 CAPU CAPU COR2 f4 (6) 604 MORA ET AL.

− where CAPU,COR1, and COR2 are the concentrations of each stud- because HCO3 is generated during the weathering of silicates and ied element in the Apure River and in The Orinoco River at sites OR1 carbonates, increasing the pH values in waters. The concentrations and OR2 (respectively); ΔCAPU, ΔCOR1, and ΔCOR2 are the associated of dissolved Mg after the Apure–Orinoco confluence align well with standard deviations for these measurements; and Δf3 and Δf4 are the theoretical mixing line (Figure 8a). Similarly, Figure 10a and b the uncertainties associated with the contribution of the Apure River shows that there are no significant differences between the pre- to the Orinoco River at sites OR3 and OR4. dicted and measured concentrations of dissolved Mg and Na after The Apure River contribution (f) at site OR3 of the Orinoco the Apure–Orinoco mixing in all months, indicating that both ele- River (between 0.04 and 0.15) was higher than that based on ments are conservative during the mixing. Indeed, these elements discharge, indicating that there is no complete mixing of waters have shown a conservative behavior in river mixing zones (Aucour at Sector 3. Figure 8 shows the DOC concentrations and the stud- et al., 2003; Ran et al., 2010), mainly due to their high mobility dur- ied element compositions in the Apure and Orinoco rivers at site ing weathering and transport. OR1 (site OR2 for DOC concentrations), and it compares the measured concentrations with those predicted by the conservative mixing model after the Apure–Orinoco confluence during the 3.5.2 | Nutrients (K and Si) increasing branch of the hydrograph (June, July, and August). The Because the dissolved K in the Orinoco River is derived from errors bars on the mixing line indicate the uncertainties associated silicate weathering, the concentrations of this element should be with the calculation of the predicted concentrations of the studied the highest in all sampling locations during the low‐water stage. elements. However, this element only showed high concentrations during the low‐water stage at sites OR1, OR2, and OR3, with little variation along the hydrological cycle at sites OR4, OR5, and OR6 3.4 | Non‐conservative behavior of DOC (Figure 4b). Studies carried out in the Apure River have shown that Figure 9 compares the measured concentrations of DOC at sites OR3 a high proportion of K comes from biogenic sources (Mora et al., and OR4 with those predicted by the conservative mixing model dur- 2010b). The Apure River’s deltaic floodplain (internal delta type) ing the 15 months of the study. The error bars indicate the standard covers a large extent of lowlands areas. During the rainy season, deviations resulting from three determinations (for the measured shallow flooding occurs in these large expanses of savanna concentrations) and the uncertainties associated with the predicted (Saunders & Lewis, 1988; Hamilton & Lewis, 1990), and a vast area concentrations. This figure shows that the predicted DOC concentra- with high vegetation density is inundated, promoting the tions are significantly higher than the measured concentrations at site decomposition of large amounts of submerged vegetation. During OR3, indicating that DOC is not conservative in the mixing zone. This this process, the K content in plants is leached from the vegetation process has also been observed in the Negro–Solimões mixing zone to the river waters (Chaudhuri, Clauer, & Semhi, 2007), increasing (Aucour et al., 2003; Moreira‐Turcq, Seyler, Guyot, & Etcheber, 2003) the content of dissolved K in the Apure River (Mora et al., and is due to the high‐molecular‐weight organic matter coming from 2010b), which is provided later to the Orinoco. However, this Negro River being adsorbed on mineral phases supplied by the process is not extensive in the Orinoco River due to the Solimões River by ligand exchange reactions governing sorption to differences in the morphology of its floodplain. Unlike the Apure clays (Pérez, Moreira‐Turcq, Gallard, Allard, & Benedetti, 2011). The floodplain, the Orinoco River floodplain consists of a network of DOC loss seems to only be extensive at site OR3 at the high‐water permanent lagoons that fringe the main channel. Therefore, at the stage because during this stage, there is a flushing of accumulated low‐water stage, the major proportion of dissolved K content in pools of high‐molecular‐weight organic matter originating from Shield the lower Orinoco River (Sector 4) is derived from the weathering rivers at site OR1 (Lewis, Saunders, Levine, & Weibezahn, 1986; Mora of silicates and K‐feldspars in the Andes and in the Guayana Shield. et al., 2010a), which could be adsorbed onto mineral particles supplied Nevertheless, during the high‐water stage, the dissolved K derived by the Apure River. from rock weathering decreases due to a dilution process, whereas the biogenic K supplied by the Apure River increases considerably, producing a non‐well‐defined K concentration pattern in Sector 4. 3.5 | Behavior of major and trace elements Although Si can be derived from silicate weathering, it did not 3.5.1 | Elements derived from water–rock interactions (Na, show a positive relationship with Ca (Figure 6). This fact, together Ca, and Mg) with the depletion of Si concentrations in February and March and Figure 4a, c, and d shows that the concentration patterns of Na, Ca, the significant loss of this element in the same months (Figures 4e, and Mg are opposite to the annual evolution of the water discharge 8c and 10d), indicates that Si can be taken up by the biomass, of the Orinoco River. These elements are mainly provided by specifically during the low‐water stage. Figure 3d and e shows that whitewater tributaries (Guaviare, Meta, and Apure) due to the the minimum values of TSS and DOC occur in February, so the light weathering of silicates, carbonates, and evaporites in the Andes path through the water column is increased during this month. (Edmond et al., 1996). Their temporal variations are controlled by a Similarly, Lewis (1988) indicates that during low water, there is a dilution process due to mineralized groundwater being diluted by peak of primary production in the Orinoco mainstream, where the non‐mineralized superficial waters during the rainy season. These dominant taxa include the diatoms Melosira and Rhizosolenia.Itis − elements show positive relationships with pH and HCO3 (Figure 6) well known that during favorable conditions (light and nutrients), MORA ET AL. 605

FIGURE 8 Comparison between measured concentrations of dissolved Mg, K, Si, dissolved organic carbon (DOC), Mn, Zn, Fe, Al, Cu, and Cr and those predicted by a conservative mixing model between the Apure and Orinoco Rivers during June, July, and August 2007 (February and March 2008 for Si) there is a “bloom” of phytoplankton communities, mainly diatoms Kociolek, 2011). Because the increase in primary production in the (Furnas, 1990), which have the ability to metabolize silicic acid from Orinoco River is driven by the high transparency values during the the environment for the formation of their exoskeletons (Seckbach & low‐water stage (Lewis, 1988), large quantities of dissolved Si can 606 MORA ET AL.

FIGURE 9 Comparison between measured and predicted concentrations of dissolved organic carbon (DOC) at sampling sites OR3 and OR4 (after the Apure–Orinoco confluence) be taken up by a diatom bloom, thus explaining the depletion of Si at this stage. Given these results, the behavior of Mn in the Orinoco concentrations during the low‐water stage and the Si removal after River and its tributaries is controlled by a pH‐dependent redox process the Apure–Orinoco confluence at this stage. (Figure 11d), where Mn‐oxyhydroxides can be incorporated into the particulate phase by coprecipitation. The positive relationship between Mn and Zn (Figure 6) and the 3.5.3 | Redox‐sensitive elements (Mn and Fe) drastic removal of both elements after the Apure–Orinoco confluence Like several worldwide rivers (Gammons, Nimick, & Parker, 2015), the (Figures 8f and 10f) suggest that a specific process must be involved to changes in pH and redox conditions play a key role in the temporal explain the behavior of both elements. It is well known that Mn oxides and spatial variability of dissolved Mn along the Orinoco River. The scavenge trace elements in aquatic environments (Tebo et al., 2005) inverse relationship between pH and Mn shown in Figure 6 and and that the oxidation of Mn may affect the cycling and transport of the drastic removal of Mn from the dissolved phase after the Zn because Zn can be adsorbed by Mn oxides during Mn oxidation Apure–Orinoco mixing (Figures 8e and 10e) indicate that the Mn (Tebo et al., 2004). Thus, the removal of Zn from the dissolved phase removal can be produced by the formation of Mn‐oxyhydroxides after the Apure–Orinoco mixing can be due to the scavenging of Zn because under aerobic conditions, the oxidation of soluble Mn2+ to by Mn oxides, where the oxidation of Mn can be catalyzed by bacteria insoluble Mn‐oxyhydroxides (such as Mn2O3, MnOOH, and MnOx)is in response to increased pH. This is in agreement with studies carried catalyzed by a diverse group of bacteria in aquatic environments, out by Shiller and Boyle (1985) in tributaries of the Yangtze, Amazon, where the large activation energy of Mn2+ oxidation can be overcome and Orinoco Rivers, who found that dissolved Zn shows a strong by increased pH (Tebo, Johnson, McCarthy, & Templeton, 2005). decrease with increasing pH. Moreover, because the significant tem- Thus, due to the recent discovery of Mn‐oxidizing bacteria in the poral variations of pH and Zn are evident in the Orinoco (high values Mn‐enriched rock coatings of the Orinoco (Dorn, Krinsley, & Ditto, of pH and low Zn concentrations during low water), the adsorption 2012), the increase in pH values in the Orinoco during the low‐water process of Zn by Mn oxides can also control the temporal changes of stage and after the Apure–Orinoco confluence can favor the oxidation Zn along this river. of Mn2+ by the action of bacteria, promoting the removal of Mn from The PCA results (Figure 6) show that dissolved Fe tends to be the dissolved phase. associated with redox‐sensitive elements (Mn) and with DOC. Figure 11 shows the relationships among the pH, dissolved oxy- Additionally, Figures 8g and 10g show that dissolved Fe behaved gen concentrations and the dissolved Mn in the Apure, Caura, and Ori- conservatively in several months and non‐conservatively in other noco Rivers. Although the Apure River showed the lowest values of months after the Apure–Orinoco mixing. For example, due to Fe dissolved oxygen at the high‐water stage, the high pH values of this showing a strong affinity with DOC to form Fe‐bearing organic river can catalyze the oxidation of Mn2+ by bacteria, thereby limiting colloids in tropical rivers (Dupré et al., 1999; Benedetti, Mounier, the solubility of this element (Figure 11a). Similarly, even though the Filizola, Benaim, & Seyler, 2003a), the removal of Fe from the Caura River showed low pH values, the Mn content in the dissolved dissolved phase after the Apure–Orinoco confluence in several phase is limited because this river shows high concentrations of months can be attributed to the adsorption of Fe‐organic complexes dissolved oxygen through the whole year (Figure 11b), which also onto mineral surfaces provided by the Apure River. Also, the forma- favors the biogenic Mn2+ oxidation (Tebo et al., 2004). The maximum tion of Fe‐oxyhydroxides due to the input of the Apure alkaline concentrations of dissolved Mn were found before the Apure–Orinoco waters promotes the precipitation of authigenic Fe, increasing the confluence at the high‐water stage due to this sector showing a com- amount of Fe in the sands of the Orinoco River (Johnsson, Stallard, bination of low concentrations of dissolved oxygen and low pH values & Lundberg, 1991). MORA ET AL. 607

FIGURE 10 Comparison between measured and predicted concentrations of dissolved (a) Mg, (b) Na, (c) K, (d) Si, (e) Mn, (f) Zn, (g) Fe, (h) Al, (i) Cu, and (j) Cr at sampling sites OR3 and OR4 (after the Apure–Orinoco confluence)

Although the question of what processes control the Fe behavior Because dissolved Fe and Al can be bounded to two different pools along the Orinoco is difficult to answer, redox processes seem to play a of organic colloids (high‐ and low‐molecular‐weight organic colloids) key role in the cyclicity of Fe in the Orinoco River and its tributaries. in tropical rivers (Dupré et al., 1999; Benedetti, Ranville, Ponthieu, & 608 MORA ET AL.

(a) (b)

(d)

(c)

FIGURE 11 Scatter plots of (a) dissolved Mn versus dissolved oxygen, (b) dissolved Mn versus pH, and (c) dissolved oxygen versus pH in the Apure, Caura, and Orinoco Rivers. (d) 3D graphic showing the dissolved oxygen and pH dependence of Mn concentrations. Data on the Caura and Apure rivers are from Mora et al. (2010a) and Mora et al. (2010b), respectively

Pinheiro, 2002) and because the binding capacity of organic colloids (Figure 5d) are due to the high pH values, the low DOC concentrations, for Fe and Al increases with the molecular weight of the colloids and the high concentrations of dissolved oxygen found at this stage, (Benedetti, Ranville, Allard, Bednar, & Menguy, 2003b), the inverse whereas the enhancement of DOC concentrations and the decrease relationship between the Fe/Al ratios and the dissolved oxygen con- in the pH and dissolved oxygen concentrations during high water centrations shown in Figure 12 indicates that under reducing condi- promote the increase of dissolved Fe concentrations in all sectors of tions, there is an increase in Fe concentrations with respect to Al and the Orinoco River. DOC. In other words, there is an increase in the concentrations of “truly” dissolved Fe2+, which has a low affinity to form organic com- | plexes with DOC (Van Schaik, Persson, Berggren Kleja, & Gustafsson, 3.5.4 Elements associated with organic matter (Al, Cu, and Cr) 2008; Allard et al., 2011). Indeed, the low concentrations of Fe found in all sectors of the Orinoco River during the low‐water stage Although dissolved Al concentrations in river waters have been associated with Al‐organic complexes, dissolved Al was not correlated with DOC (Figure 6). Dissolved Al has been shown to be complexed with two different pools of organic colloids (high‐ and low‐molecular‐ weight organic colloids) in tropical river waters (Dupré et al., 1999; Benedetti et al., 2002, 2003a). The high‐molecular‐weight organic pool is highly aromatic and more reactive and tends to be adsorbed by mineral surfaces, whereas fulvic substances of low aromaticity and with abundant oxygen groups are less reactive (Bardy et al., 2008; Alasonati, Slaveykova, Gallard, Croué, & Benedetti, 2010). Figures 8h and 10h show that the predicted concentrations of Al in the Orinoco River were significantly higher than the measured concentrations at sites OR3 and OR4. This indicates a removal of Al after the Apure–Orinoco mixing, which is in agreement with previous studies carried out in the Amazon River system (Benedetti et al., 2003a). Because high‐molecu- FIGURE 12 Scatter diagram showing the relationship between Fe/Al ‐ ratios and dissolved oxygen concentrations in the Orinoco, Apure, lar weight humic substances are preferentially adsorbed onto mineral and Caura Rivers. Data on the Caura and Apure Rivers are from Mora phases during mixing processes (Pérez et al., 2011), the content of et al. (2010a) and Mora et al. (2010b), respectively Al‐organic complexes of high molecular weight in the waters of the MORA ET AL. 609

Orinoco are preferentially adsorbed onto mineral phases supplied by can be due to the coprecipitation of Mn‐oxyhydroxides and the scav- the Apure River. However, less‐reactive Al‐organic complexes of low enging of Zn onto Mn‐oxides. Overall, dissolved Mn is a function of a molecular weight are not adsorbed onto mineral surfaces and can pH‐dependent redox process, and the Zn content in the dissolved behave conservatively during mixing. Thus, the distribution of phase can be controlled by the oxidation of Mn2+. Redox reactions dissolved Al between the different pools of DOC and the preferential can also play an important role in the behavior of dissolved Fe in the removal of large organic molecules bound to Al during the Apure– Orinoco River and its tributaries. However, other processes, such as Orinoco mixing can explain the loss of DOC and Al during the mixing pH variations and the adsorption of large organometallic complexes and thus the absence of the relationship between DOC and dissolved of Fe onto mineral surfaces, can also control the mobility of dissolved Al, as shown in Figure 6. Fe during mixing and transport. Dissolved Cu and Cr showed a positive relationship with DOC The temporal variations of dissolved Al, Cr, and Cu are controlled by (Figure 6), suggesting that they are complexed by organic matter and the temporal variability of DOC concentrations in the Orinoco River due that their temporal variations depend on the temporal variability of to these elements being complexed with dissolved organic matter. DOC concentrations along the hydrological cycle. In fact, like the Dissolved Al behaves in a non‐conservative way during the Apure– DOC concentrations, both elements showed their minimum concentra- Orinoco mixing. However, Cu and Cr can behave conservatively because tions at the low‐water stage and their maximum concentrations during both elements tend to be complexed with small organic colloids, the high‐water stage (Figure 5e and f). Figure 8i and j also shows that which can be not adsorbed on the clays provided by the Apure River. the concentrations of Cu and Cr after the Apure–Orinoco confluence align well with the mixing line between both endmembers, indicating ACKNOWLEDGMENTS that they are the only two studied trace elements that showed a con- The authors thank the ORE/HYBAM project (www.ore‐hybam.org) servative behavior after the Apure–Orinoco mixing. Ultrafiltration for providing data on the daily water discharges of the Orinoco River studies carried out in whitewater rivers (like the Apure and Orinoco at Ciudad Bolívar. The help of Jorge and Francisco Medina in the Rivers) of the Amazon basin have shown that 75% of the bulk organic field is greatly appreciated. The authors also acknowledge the carbon is composed of small organic colloids (Benedetti et al., 2003a), constructive comments of two reviewers. This work was entirely which do not tend to be adsorbed onto mineral surfaces (Pérez et al., supported by Conoco‐Phillips Latinoamerica under LOCTI (Project 2011; Alasonati et al., 2010). Additionally, several studies have shown CONOCO‐EDIHG‐23938). that lower‐molecular‐weight fractions of organic colloids exhibit the highest complexing capacity for Cu and Cr (Nifant’eva, Burba, Fedorova, Shkinev, & Ya Spivakov, 2001; Benedetti et al., 2003a). This REFERENCES can suggest that dissolved Cu and Cr can be mainly bounded to small‐ Alasonati, E., Slaveykova, V. I., Gallard, H., Croué, J., & Benedetti, M. F. sized organic matter in the Orinoco River. Thus, the conservative (2010). Characterization of the colloidal organic matter from the Ama- zonian basin by asymmetrical flow field‐flow fractionation and size behavior of both elements is induced by the low capacity of these exclusion chromatography. Water Research, 44, 223–231. small organometallic complexes to be adsorbed onto mineral surfaces. doi:10.1016/j.watres.2009.09.010 However, over several months, there is a small but significant loss of Allard, T., Weber, T., Bellot, C., Damblans, C., Bardy, M., Bueno, G., … Cu from the dissolved phase (Figure 10i), most likely due to a minor Benedetti, M. F. (2011). Tracing source and evolution of suspended particles in the Rio Negro basin (Brazil) using chemical species of amount of Cu associated with large‐sized organic colloids, which are iron. Chemical Geology, 280,79–88. doi:10.1016/j.chemgeo. adsorbed onto clays supplied by the Apure River. 2010.10.018 Amon, R. M. W., & Benner, R. (1996). Photochemical and microbial consumption of dissolved organic carbon and dissolved oxygen in 4 | CONCLUSIONS the Amazon River system. Geochimica et Cosmochimica Acta, 60, 1783–1792. doi:10.1016/0016-7037(96)00055-5

The concentrations of dissolved Na, Ca, and Mg along the Orinoco Aucour, A. M., Tao, F. X., Moreira‐Turcq, P., Seyler, P., Sheppard, S., & Benedetti, M. F. (2003). The Amazon River: Behaviour of metals (Fe, River are controlled by a dilution process. Thus, the unimodal sea- Al, Mn) and dissolved organic matter in the initial mixing at the Rio sonal pattern shown by these elements is driven by one unimodal Negro/Solimões confluence. Chemical Geology, 197, 271–285. water discharge regime in the river. The concentrations of dissolved doi:10.1016/S0009-2541(02)00398-4 K do not show a well‐defined unimodal seasonal pattern due to the Bardy, M., Fritsch, E., Derenne, S., Allard, T., Do Nascimento, N. R., & input of plant‐derived K during the high‐water stage, which is pro- Bueno, G. T. (2008). Micromorphology and spectroscopic characteristics of organic matter in waterlogged podzols of the upper Amazon vided by the Apure River. All of these elements behave conserva- basin. Geoderma, 145, 222–230. doi:10.1016/j.geoderma.2008.03.008 tively during the Apure–Orinoco mixing. Dissolved Si is also Benedetti, M., Ranville, J. F., Ponthieu, M., & Pinheiro, J. P. (2002). Field‐ controlled by a dilution process, although this element behaves flow fractionation characterization and binding properties of particulate non‐conservatively at the low‐water stage probably due to biological and colloidal organic matter from the Rio Amazon and Rio Negro. – removal by diatoms. Organic Geochemistry, 33, 269 279. doi:10.1016/S0146-6380(01) 00159-0 The loss of DOC during the Apure–Orinoco confluence can be mainly related to the adsorption of high‐molecular‐weight organic Benedetti, M. F., Mounier, S., Filizola, N., Benaim, J., & Seyler, P. (2003a). Carbon and metal concentrations, size distributions and fluxes in major colloids onto mineral particles supplied by the Apure River. Similarly, rivers of the Amazon basin. Hydrological Processes, 17, 1363–1377. the removal of Mn and Zn from the dissolved phase during mixing doi:10.1002/hyp.1289 610 MORA ET AL.

Benedetti, M. F., Ranville, J. F., Allard, T., Bednar, A. J., & Menguy, N. savannas. Estuarine, Coastal and Shelf Science, 63, 527–536. (2003b). The iron status in colloidal matter from the Rio Negro, Brasil. doi:10.1016/j.ecss.2004.10.020 Colloids and Surfaces A: Physicochemical and Engineering Aspects, 217, Miller,J.C.,&Miller,J.N.(1989).Statisticsforanalyticalchemistry. (pp.53–54). – 1 9. doi:10.1016/S0927-7757(02)00553-8 Chichester: Ellis Horwood. Chaudhuri, S., Clauer, N., & Semhi, K. (2007). Plant decay as a major control Mora, A., Alfonso, J., Baquero, J. C., Balza, L., & Pisapia, D. (2010a). Caura of river dissolved potassium: A first estimate. Chemical Geology, 243, River basin: Weathering rates, CO2 consumption, and chemistry of 178–190. doi:10.1016/j.chemgeo.2007.05.023 major and trace elements in an Orinoco River tributary coming from Dorn, R. I., Krinsley, D. H., & Ditto, J. (2012). Revisiting Alexander von the Precambrian Guayana Shield, Venezuela. Geochemistry, Geophysics, Humboldt’s initiation of rock coating research. Journal of Geology, 120, Geosystems, 11, Q11005. doi:10.1029/2010GC003209 1–14. doi:10.1086/662737 Mora, A., Alfonso, J., Sánchez, L., Calzadilla, M., Silva, S., LaBrecque, J., & Dupré, B., Viers, J., Dandurand, J., Polve, M., Bénézeth, P., Vervier, P., & Azócar, J. (2009). Temporal variability of selected dissolved elements Braun, J. (1999). Major and trace elements associated with colloids in the lower Orinoco River, Venezuela. Hydrological Processes, 23, in organic‐rich river waters: Ultrafiltration of natural and spiked 476–485. doi:10.1002/hyp.7159 solutions. Chemical Geology, 160,63–80. doi:10.1016/S0009-2541 Mora, A., Baquero, J. C., Alfonso, J. A., Pisapia, D., & Balza, L. (2010b). The (99)00060-1 Apure River: Geochemistry of major and selected trace elements in an Edmond, J. M., Palmer, M. R., Measures, C. I., Brown, E. T., & Huh, Y. (1996). Orinoco River tributary coming from the Andes, Venezuela. Hydrologi- Fluvial geochemistry of the eastern slope of the northeastern Andes cal Processes, 24, 3798–3810. doi:10.1002/hyp.7801 and its foredeep in the drainage of the Orinoco in Colombia and Moreira‐Turcq, P., Seyler, P., Guyot, J. L., & Etcheber, H. (2003). Character- Venezuela. Geochimica et Cosmochimica Acta, 60, 2949–2976. istics of organic matter in the mixing zone of the Rio Negro and Rio doi:10.1016/0016-7037(96)00142-1 Solimões of the Amazon River. Hydrological Processes, 17, 1393–1404. Furnas, M. J. (1990). In situ growth rates of marine phytoplankton: doi:10.1002/hyp.1291 Approaches to measurement, community and species growth rates. Nifant’eva, T. I., Burba, P., Fedorova, O., Shkinev, V. M., & Ya Spivakov, B. Journal of Plankton Research, 12, 1117–1151. doi:10.1093/plankt/ (2001). Ultrafiltration and determination of Zn‐ and Cu‐humic 12.6.1117 substances complexes stability constants. Talanta, 53, 1127–1131. Gammons, C. H., Nimick, D. A., & Parker, S. R. (2015). Diel cycling of trace doi:10.1016/S0039-9140(00)00492-6 elements in streams draining mineralized areas—A review. Applied Pérez, M. A. P., Moreira‐Turcq, P., Gallard, H., Allard, T., & Benedetti, M. F. Geochemistry, 57,35–44. doi:10.1016/j.apgeochem.2014.05.008 (2011). Dissolved organic matter dynamic in the Amazon basin: Hamilton, S. K., & Lewis, W. M. (1990). Basin morphology in relation to Sorption by mineral surfaces. Chemical Geology, 286, 158–168. chemical and ecological characteristics of lakes on the Orinoco River doi:10.1016/j.chemgeo.2011.05.004 floodplain, Venezuela. Archiv für Hydrobiologie, 119, 393–425. Ran, X., Yu, Z., Yao, Q., Chen, H., & Mi, T. (2010). Major ion geochemistry Johnsson, M. J., Stallard, R. F., & Lundberg, N. (1991). Controls on the and nutrient behaviour in the mixing zone of the Changjiang (Yangtze) composition of fluvial sands from a tropical weathering environment: River and its tributaries in the three Gorges Reservoir. Hydrological Sands of the Orinoco River drainage basin, Venezuela and Colombia. Processes, 24, 2481–2495. doi:10.1002/hyp.7684 Geological Society of America Bulletin, 103, 1622–1647. doi:10.1130/ 0016-7606(1991)103<1622:COTCOF>2.3.CO;2 Reimann, C., Filzmoser, P., & Garrett, R. G. (2002). Factor analysis applied to regional geochemical data: Problems and possibilities. Applied Geochem- Laraque, A., Castellanos, B., Steiger, J., López, J. L., Pandi, A., Rodríguez, M., istry, 17, 185–206. doi:10.1016/S0883-2927(01)00066-X … Lagane, C. (2013). A comparison of the suspended and dissolved matter dynamics of two large inter‐tropical rivers draining into the Saunders, J. F., & Lewis, W. M. (1988). Transport of phosphorus, nitro- Atlantic Ocean: The Congo and the Orinoco. Hydrological Processes, gen, and carbon by the Apure River, Venezuela. Biogeochemistry, 5, – 27, 2153–2170. doi:10.1002/hyp.9776 323 342. doi:10.1007/BF02180071 Lewis, W. M. (1988). Primary production in the Orinoco River. Ecology, 69, Saunders,J.F.,&Lewis,W.M.(1989).Transportofmajorsolutesandtherela- 679–692. doi:10.2307/1941016 tionship between solute concentrations and discharge in the Apure River, Venezuela. Biogeochemistry, 8,101–113. doi:10.1007/BF00001315 Lewis, W. M., & Saunders, J. F. (1984). Cross‐sectional variation in the chemistry and suspended sediment load of the Orinoco River at Ciudad Seckbach, J., & Kociolek, J. P. (2011). Diatoms: General introduction. In J. Bolívar. Acta Científica Venezolana, 35, 382–385 Seckbach, & J. P. Kociolek (Eds.), The diatom world. (pp. 11–12). London‐New York: Springer Dordrecht Heidelberg. Lewis, W. M., & Saunders, J. F. (1989). Concentration and transport of dissolved and suspended substances in the Orinoco River. Biogeochem- Shiller, A. M., & Boyle, E. (1985). Dissolved zinc in rivers. Nature, 317, istry, 7, 203–240. doi:10.1007/BF00004218 49–52. doi:10.1038/317049a0 Lewis, W. M., & Saunders, J. F. (1990). Chemistry and element transport by Stallard, R. F. (1987). Cross‐channel mixing and its effect on sedimentation the Orinoco main stem and lower tributaries. In F. H. Weibezahn, H. in the Orinoco River. Water Resources Research, 23, 1977–1986. Alvarez, & W. M. Lewis (Eds.), The Orinoco River as an ecosystem. doi:10.1029/WR023i010p01977 (pp. 211–239). Caracas: Impresos Rubel CA Stallard, R. F., Koehnken, L., & Johnsson, M. J. (1991). Weathering Lewis, W. M., Hamilton, S. K., Jones, S. L., & Runnels, D. D. (1987). Major processes and the composition of inorganic material transported elements chemistry, weathering and element yields for the Caura River through the Orinoco River system, Venezuela and Colombia. Geoderma, drainage, Venezuela. Biogeochemistry, 4, 159–181. doi:10.1007/ 51, 133–165. doi:10.1016/0016-7061(91)90069-6 BF02180153 Tebo, B. M., Bargar, J. R., Clement, B. G., Dick, G. J., Murray, K. J., Parker, D., Lewis, W. M., Saunders, J. F., Levine, S. N., & Weibezahn, F. H. (1986). … Webb, S. M. (2004). Biogenic manganese oxides: Properties and Organic carbon in the Caura River, Venezuela. Limnology and Oceanog- mechanisms of formation. Annual Review of Earth and Planetary Sci- raphy, 31, 653–656. doi:10.4319/lo.1986.31.3.0653 ences, 32,287–328. doi:10.1146/annurev.earth.32.101802.120213 Meade, R., Weibezahn, F., Lewis, W. M., & Pérez, D. (1990). Suspended‐ Tebo, B. M., Johnson, H. A., McCarthy, J. K., & Templeton, A. S. (2005). sediment budget for the Orinoco River. In F. H. Weibezahn, H. Alvarez, Geomicrobiology of manganese(II) oxidation. Trends in Microbiology, & W. M. Lewis (Eds.), The Orinoco River as an ecosystem. (pp. 55–79). 13, 421–428. doi:10.1016/j.tim.2005.07.009 Caracas: Impresos Rubel CA. Van Schaik, J. W. J., Persson, I., Berggren Kleja, D., & Gustafsson, J. P. Medina, E., Francisco, M., Sternberg, L., & Anderson, W. T. (2005). Isotopic (2008). EXAFS study on the reactions between iron and fulvic acid in signatures of organic matter in sediments of the continental shelf facing acid aqueous solutions. Environmental Science & Technology, 42, the Orinoco Delta: Possible contribution of organic carbon from 2367–2373. doi:10.1021/es072092z MORA ET AL. 611

Vegas‐Vilarrubia, T., Paolini, J. E., & Miragaya, J. G. (1988). Differentiation ‐ of some Venezuelan blackwater rivers based upon physico chemical How to cite this article: Mora, A., Mahlknecht, J., Baquero, J. properties of their humic substances. Biogeochemistry, 6,59–77. doi:10.1007/BF00002933 C., Laraque, A., Alfonso, J. A., Pisapia, D., Balza, L. Dynamics of dissolved major (Na, K, Ca, Mg, and Si) and trace (Al, Fe, SUPPORTING INFORMATION Mn, Zn, Cu, and Cr) elements along the lower Orinoco River, Additional Supporting Information may be found online in the Hydrological Processes. 2017;31:597–611. doi: 10.1002/ supporting information tab for this article. hyp.11051