Seasonality of Dissolved Rare Earth Elements in the Lower Mississippi River

Transcription

1 The University of Southern Mississippi The Aquila Digital Community Faculty Publications Seasonality of Dissolved Rare Earth Elements in the Lower Mississippi River Alan M. Shiller Follow this and additional works at: Part of the Chemistry Commons Recommended Citation Shiller, A. M. (2002). Seasonality of Dissolved Rare Earth Elements in the Lower Mississippi River. Geochemistry,, Geosystems, 3. Available at: This Article is brought to you for free and open access by The Aquila Digital Community. It has been accepted for inclusion in Faculty Publications by an authorized administrator of The Aquila Digital Community. For more information, please contact

2 Geosystems G 3 AN ELECTRONIC JOURNAL OF THE EARTH SCIENCES Published by AGU and the Geochemical Society Article Volume 3, Number November , doi: ISSN: Seasonality of dissolved rare earth elements in the lower Mississippi River Alan M. Shiller Department of Marine Science/Center for Trace Analysis, University of Southern Mississippi, Stennis Space Center, Mississippi, 39529, USA (alan.shiller@usm.edu) [1] Dissolved rare earth element (REE) concentrations were determined in a 27-month time series of the lower Mississippi River. Overall, the results agree with limited previous investigations; that is, the river shows enrichment of heavy REEs relative to light REEs and also has a significant Ce anomaly. However, the previous investigations relied on only single samples from the river. This seasonal investigation reveals significant temporal variations in the river s REE chemistry. In particular, large (approximately fivefold) variations in light REE concentrations are observed. The light REEs follow a seasonality similar to particle-reactive trace elements. Also, the Ce anomaly shows a corresponding seasonality with greatest fractionation when dissolved concentrations are lowest. The time series encompassed a period of extreme flooding in the U.S. Midwest. However, this event did not appear to affect dissolved REE concentrations. The results of this study are compatible with previous observations of dissolved trace elements in the lower Mississippi River which suggested the importance of redox processes within the river system in controlling seasonal concentration variability for many of these elements. Additionally, speciation modeling indicates that carbonate complexes likely dominate the solution chemistry of REEs in the lower Mississippi. The observed behavior of the dissolved REEs is likewise compatible with this speciation. Components: 5965 words, 6 figures, 3 tables. Keywords: River; trace elements; rare earth elements; redox; Mississippi River. Index Terms: 1806 Hydrology: Chemistry of fresh water; 1065 Geochemistry: Trace elements (3670); 1871 Hydrology: Surface water quality; 1045 Geochemistry: Low-temperature geochemistry. Received 30 April 2002; Revised 6 August 2002; Accepted 29 August 2002; Published 16 November Shiller, A. M., Seasonality of dissolved rare earth elements in the lower Mississippi River, Geochem. Geophys. Geosyst., 3(11), 1068, doi:, Introduction [2] Time series of dissolved trace elements in rivers are important for various reasons. For instance, seasonal data can help elucidate mechanisms controlling trace element concentrations whereas single samples may be biased by unusual conditions. Time series data are also necessary for accurate annual flux calculations as well as for understanding how to design and implement research and monitoring programs. Previously, I examined a suite of dissolved trace elements in a 27-month time series from the lower Mississippi River [Shiller, 1997]. Figure 1 summarizes some of those previous observations. In that study, it was found that Mn and Fe as well as the particle-reactive elements Pb and Zn had their highest dissolved concentrations when water temperature was coolest. Seasonal variation Copyright 2002 by the American Geophysical Union 1 of 14

3 Figure 1. Dissolved trace elements in the lower Mississippi River [from Shiller, 1997]. in the concentrations of these trace elements could not be explained by variations in discharge, tributary mixing ratios, ph, DOC, or suspended load. The effect of temperature on the rate of microbial Mn oxidation was suggested as at least a partial explanation for this seasonal cycle [Shiller, 1997; Shiller and Hebert, 1998]. According to this hypothesis, warmer temperatures would result in faster rates of Mn oxidation which would have the following consequences: a) increased formation of freshly precipitated, adsorbing Mn oxide surfaces, b) increased adsorption of particle-reactive elements such as Pb and Zn, and c) decreased solution concentrations of Mn, Pb, and Zn. In contrast to these elements, dissolved V and Mo show an opposite (to Mn) seasonal cycle in the lower Mississippi that was suggested to result from the reductive removal of these elements in upstream 2of14

4 seasonally reducing reservoirs and lakes [Shiller, 1997]. [3] The rare earth elements (REEs) form a unique chemical set wherein the gradual decrease in ionic radius across the series leads to systematic changes in geochemical behavior. In this report I examine the seasonal behavior of dissolved REEs in the same lower Mississippi River time series described above. The objective of this exercise is to see if REE behavior is in accord with the trace element behaviors described above and the processes hypothesized to account for those behaviors. [4] Previous analyses of dissolved REEs in single samples from the lower Mississippi River have shown the shale-normalized dissolved REE distribution to be enriched in the heavy REEs relative to the light REEs [Sholkovitz, 1995; Goldstein and Jacobsen, 1988]. (The shale normalization provides an approximate normalization to the average weathering source.) This heavy REE enrichment is reasonable given what is known about REE complexation and adsorption as well as the physical-chemical nature of the Mississippi River. Specifically, solution complexation of REEs in a moderately alkaline river like the Mississippi (ph 7.8; alkalinity 2 meq/l) is likely dominated by carbonate complexes and the stability constants of REE-carbonate complexes increase with increasing atomic number. In contrast, adsorption of REEs by d-mno 2 [Ohta and Kawabe, 2001], Fe-oxyhydroxide [Ohta and Kawabe, 2001], and surface carboxylates [Byrne and Kim, 1990] appears to change less strongly with atomic number. It is these contrasting trends in solution complexation versus surface adsorption across the REE series that are thought to lead to the heavy REE enrichment of seawater [Byrne and Kim, 1990] and presumably also in lower Mississippi River water. [5] If, as is hypothesized above, the observed seasonal concentration variability of dissolved Pb and Zn in the lower Mississippi is due to seasonal changes in the adsorption of these elements, then one might expect the lighter REEs to show a seasonal concentration variability similar to Pb and Zn due to a dominant effect of adsorption. In contrast, seasonal trends for the heavier REEs should be more muted due to solution complexation being more important for their physicalchemical behavior. [6] The previous work [Sholkovitz, 1995; Goldstein and Jacobsen, 1988] has also shown a significant negative Ce anomaly in the Mississippi River. Ce, unlike its neighboring REEs, can be oxidized from the typical REE +III oxidation state to the more readily scavenged +IV oxidation state, resulting in Ce s preferential removal from solution. The Ce anomaly is usually calculated as the ratio of the observed Ce concentration to the concentration predicted based on La and Nd concentrations (with all of the concentrations shale-normalized as an input source correction): Ce/Ce* = 3[Ce] n /(2[La] n + [Nd] n ), where n refers to shale-normalized values. (Ce anomalies are typically calculated this way rather from La and Pr because monoisotopic Pr generally has not been analyzed in the past.) If redox processes are important in determining the seasonal variability of dissolved Mn and Fe in the lower Mississippi, then it is expected that the Ce anomaly will show its lowest values (i.e., strongest oxidation effect) when water temperature is warm and dissolved Mn and Fe are low. 2. Methods [7] Approximately monthly samples were collected from the Mississippi River above Baton Rouge, Louisiana from October 1991 to December Samples from tributaries of the Mississippi River were collected during the same time period in conjunction with the U.S. Geological Survey s Mississippi River Project [Meade, 1996]. Clean sampling and filtration (0.4 mm Nuclepore polycarbonate) protocols were followed as described previously [Shiller, 1997]. [8] REEs were determined by ICP-MS (Thermo- Finnigan Element 2) using a desolvating microconcentric nebulizer (CETAC Aridis). The desolvating nebulizer minimizes oxide formation which can cause severe interferences with ICP-MS analysis of REEs. Calibration was by isotope dilution utilizing enriched isotope spikes (Oak 3of14

5 Ridge National Laboratory) of Ce-142, Nd-145, Sm-149, Eu-153, Gd-155, Dy-161, Er-167, and Yb-171. These spikes not only allowed for calculation of the concentrations of 8 spiked elements but also permitted the detector response to be calibrated across the entire REE series. Thus, the spikes also served as internal standards for the determination of the other REEs. This calibration procedure is similar to that described by Field and Sherrell [1998]. [9] Some samples were extracted using an organophosphate chelator column (RE-Spec; Eichrom); however, most analyses were performed without the extraction step. No difference was noted in the results for extracted and unextracted samples except that a slight (1 pm) correction needed to be made to the unextracted Eu data due to the large amounts of Ba in Mississippi River water. This correction was based both on the comparison of the extracted and unextracted samples as well as measurements of the effect of Ba additions on the Eu signal. Precision was estimated at better than ±4% based on replicate analyses. Additionally, two U.S. Geological Survey Standard Reference Water Samples (SRWS) for REE determination (PPREE1 and SCREE1 [Verplanck et al., 2001]) were analyzed six times each. For these analyses, the majority of REE results were within 5% of the published most probable value s (MPV). Additionally, there was only one result with more than 7% deviation from the reported MPV (Gd in SCREE1 was 16% lower than the MPV but only 4% lower than the MPV for PPREE1). [10] Shale normalization of results was performed using REE values for the North American Shale Composite (NASC) [Taylor and McLennan, 1985]. [11] Speciation modeling was performed using PHREEQC Interactive version RC1 [Charlton et al., 1997]. The llnl.dat database included with this version of the program was utilized. This database includes stability constants for REE complexes with CO 3 2,PO 4 3,SO 4 2,OH, and Cl. Carbonate and phosphate stability constants in the database are similar to those given by Lee and Byrne [1992]. (Note that the llnl database as obtained had an error for the reaction: Yb 3+ + HPO 4 2 = YbPO 4 +H + with the logk given as rather than ) Stability constants for REE-oxalate complexes [Schijf and Byrne, 2001] were added to this database. Major ion-oxalate complexation constants were taken from Martell and Smith [1989]. [12] For the speciation modeling, a typical composition of Mississippi River water was used based on Briggs and Ficke [1977] as well as ph and PO 4 3 data from Shiller [1997]: [Ca 2+ ] = 870 mm; [Mg 2+ ] = 425 mm; [Na + ] = 735 mm; [K + ]=70mM; [Cl ]= 530 mm; [SO 4 2 ] = 480 mm; [PO 4 3 ]=3mM; ph = 7.8; Alk = 1.9 meq/l. Rare earth concentrations used were based on mean concentrations for Mississippi River water as listed in Appendix A; thus, for the Nd and Yb calculations presented here [Nd T ] = 100 pm and [Yb T ] = 25 pm. 3. Results and Discussion 3.1. Seasonal Variability of Dissolved REE Concentrations [13] Dissolved REE concentrations for the time series are listed in the appendix (Table A1). Figure 2 shows a comparison of the time series results for shale-normalized dissolved REEs in the lower Mississippi River with those of previous workers [Sholkovitz, 1995; Goldstein and Jacobsen, 1988]. Goldstein and Jacobsen s sample shows slightly higher La than is found in the time series, but otherwise their data fall within the range of what is reported here. Sholkovitz s 0.22 mm filtered sample was collected at Vicksburg, MS on August 20, 1993 and his results are typically within 7% of the average of two time series samples collected above Baton Rouge on August 13 and 25, The shale-normalized pattern shows the previously mentioned enrichment of heavy REEs compared with the lighter REEs. Note that the choice of a different shale composite for normalization will somewhat affect the patterns observed here [Sholkovitz, 1990]. For example, use of the Post-Archean Australian Shale composite (PAAS) results in slightly less heavy REE enrichment, no anomaly at Ho, and a slight maximum in the shale normalized distribution at Er. However, the discussion below utilizing NASC shale-normalized ratios and anomalies is not significantly affected by normalization to PAAS. 4of14

6 Figure 2. Shale-normalized dissolved rare earth element concentrations in the lower Mississippi River: comparison of this work with previously published results. [14] Figure 3 shows the time series of selected dissolved REEs in the lower Mississippi River. Figure 4 shows the relationships between some REEs and certain other dissolved trace elements. Light REEs show a seasonality similar to Mn, Fe and Pb (i.e., highest dissolved concentrations in winter). However, this variability becomes tempered along the REE series and the heaviest elements show much lower seasonal variability than the light elements. Correlations between light REEs and other particle-reactive dissolved trace elements (Figure 4) are all statistically significant (p < 0.01), as would be expected from the redox and adsorption control mechanism outlined above. Clearly, though, there is much scatter to these relationships. However, the scatter is not surprising given differences in complexation and other physical-chemical behaviors of these elements as well as possible differences in sources among the various elements. Correlations among the light REEs show little scatter as would be expected from their similar chemistries and sources (e.g., r 2 = 0.93 for Pr vs. Ce and 0.99 for Pr vs. Nd, see Figure 4). No significant correlations between heavy REEs and particle-reactive trace elements (including light REEs such as Pr; see Figure 4) were observed nor were there significant correlations between heavy REEs and other relatively soluble trace elements such as Mo (Figure 4), V or U. However, heavy REEs did vary significantly with other heavy REEs (e.g., r 2 = 0.97 for both the Yb-Lu and Yb- Tm correlations) which likewise reflects the similarities of their chemistries and sources. [15] In the tributaries that were sampled (Table A2), REEs span a concentration range similar to that seen in the lower Mississippi River time series. This makes it unlikely that variable contributions of water from any one of these tributaries results in the seasonality shown in Figure 3. This is exemplified by examining the relationship between Nd and the percent contribution of the Ohio River to the discharge of the lower Mississippi (Figure 5). The Ohio, which is generally the major contributor of water to the lower Mississippi, has some of the lowest light REE concentrations of these tributaries. 5of14

7 Figure 3. Selected dissolved rare earth element concentrations in the lower Mississippi River. Thus, if changes in tributary mixing ratios were to be an important determinant of the lower Mississippi s seasonal REE concentration variability, then one would expect low Nd (as well as other light REE) concentrations when the percentage of water coming from the Ohio River is highest. Clearly this is not the case (Figure 5). [16] Changes in dissolved REE concentrations are not correlated with changes in river discharge (Figure 1) or suspended load (not shown, but see Shiller [1997]). ph changes only slightly as does DOC [see Shiller, 1997]. Thus, it seems unlikely that changes in ph, suspended load or organic complexation could be the dominant factors accounting for most for the variability of dissolved REEs in the lower Mississippi River REE Ratios and Anomalies [17] Shale-normalized ratios of lighter-to-heavier REEs can shed light on the fractionation of these 6of14

8 Figure 4. Dissolved rare earth element concentrations in the lower Mississippi River versus concentrations of other dissolved trace elements. Ce/Ce* is the cerium anomaly. elements during weathering and transport. A ratio of one indicates similar mobilization of the two elements relative to shale (the assumed weathering input). A ratio less than one suggests a lower degree of dissolved phase mobilization of the lighter REE. Figure 6 shows the Yb n /Lu n ratio is nearly one and minimally variable (0.98 ± 0.03). This is reasonable given the similar nature of these elements and their position at the heavy end of the series where solution complexation is most important. The low degree of variability helps validate the quality of the data. In contrast, the Nd n /Yb n ratio is significantly lower than one and seasonally variable (Figure 6). The low ratio reflects the overall heavy REE enrichment and the variability mainly reflects the greater seasonal variability of Nd. [18] There is a significant Ce anomaly (Figure 2, Figure 6); that is, Ce concentrations are typically 0.38 of that expected based on the shale-normalized concentrations of La and Nd. Ce anomalies calculated using La and Pr are strongly correlated (r 2 = 0.997) with the more traditional La/Nd calculated values, and only slightly higher (mean = 0.39). Seasonal variability is observed in the Ce anomaly (Figure 6), with winter values being typically 0.15 higher than summer values. Thus, Ce shows its least fractionated behavior relative to its neighbors during winter. This is in accord with the observations of other transition elements indicating either greater importance of upstream reducing sources during winter (e.g., V and Mo) or higher rates of oxidation during summer (e.g., Mn [Shiller and Hebert, 1998]). Ce fractionation also is greatest (i.e., Ce anomaly is lowest) when light REEs appear to be most strongly removed to the particulate phase. Indeed, Ce concentrations are strongly correlated with the cerium anomaly as is dissolved 7of14

9 Figure 5. Dissolved Nd and the Ce anomaly versus an estimate of the percentage of lower Mississippi River water derived from the Ohio River. Mn (Figure 4). Oxidation of Ce(III) by d-mno 2 has been observed [Ohta and Kawabe, 2001; de Carlo et al., 1998] as has microbial Ce oxidation in conjunction with microbial Mn oxidation [e.g., Moffett, 1994]. The seasonal variability of the Ce anomaly is thus compatible with the proposed scenario in which warmer waters increase the rate of microbial Mn oxidation which thereby increases the amount of adsorbing surface of freshly precipitated Mn oxides and hence results in greater summer scavenging of particle-reactive elements such as Pb, Zn, and the light REEs. However, this interpretation should be viewed cautiously given the slight correlation (r 2 = 0.30) between the cerium anomaly and the relative contribution of the Ohio River to the discharge of the lower Mississippi River (Figure 5). Additionally, the limited tributary sampling indicates a higher Ce/Ce* in the Ohio River than the other major tributaries (compare Ohio River with the Upper Mississippi and Missouri Rivers in Table A2). Thus, some of the time series variability in Ce/Ce* may result from changes in tributary mixing ratios. [19] Another important observation is that the Ce anomaly shows less relative variation than Ce concentrations. Indeed, Ce concentrations are tightly correlated with other light REE concentrations with, for example, the Ce-Pr correlation showing only slightly more scatter than the Nd-Pr correlation (Figure 4). This suggests that adsorption of Ce(III) is a more important factor in controlling seasonal variability of dissolved Ce than oxidation/ reduction of this element. [20] A slight positive Gd anomaly is also evident in Figure 2. The Gd anomaly was calculated using Sm and Tb as a reference (anomalies calculated using Eu and Tb were typically 0.05 higher than those calculated using Sm and Tb). The Gd anomaly averaged 1.14 ± 0.07 but showed no seasonal trend. Kim et al. [1991] attributed similar slight Gd anomalies observed in seawater to differ- 8of14

10 Figure 6. Shale-normalized ratios of dissolved rare earth elements in the lower Mississippi River (Ce SN /Ce* SN is the cerium anomaly). ences in Gd solution and surface complexation relative to its neighboring REEs. However, a slight anthropogenic effect, due to gadopentetic acid which is used as a magnetic resonance imaging contrast agent, cannot be discounted [Bau and Dulski, 1996]. Indeed, the Illinois River, which receives wastewater contributions from the Chicago metropolitan area and has the highest population stress (people/river discharge) of the tributaries reported here, also was observed to have the highest Gd anomaly of these tributaries ( ; see Table A2) Speciation Modeling [21] As an aid to understanding controls on dissolved REE concentrations, speciation calculations were performed comparing Nd and Yb. A first calculation was made with the solution conditions listed above in Methods. The results (Table 1) 9of14

11 Table 1. Results of PHREEQC Calculations for 100 pm Nd and 25 pm Yb a Species Concentration Species Concentration Normal Mississippi River Inorganic Composition NdCO E-11 Yb(CO3)2 1.12E-11 Nd(CO3)2 2.69E-11 YbCO E-11 NdPO4 5.42E-12 YbPO4 2.83E-12 Nd E-13 Yb(PO4) E-13 Normal Mississippi River Inorganic Model, Doubled Phosphate, Low ph NdCO E-11 Yb(CO3)2 9.77E-12 Nd(CO3)2 2.54E-11 YbCO E-12 NdPO4 1.03E-11 YbPO4 4.99E-12 Nd E-13 Yb(PO4) E-13 Normal Mississippi River Inorganic Composition With 150 mm Oxalate NdCO E-11 YbOxalate2 7.33E-12 NdOxalate2 2.18E-11 Yb(CO3)2 6.77E-12 Nd(CO3)2 1.72E-11 YbCO E-12 NdOxalate+ 1.45E-11 YbOxalate+ 2.49E-12 Normal Mississippi River Inorganic Composition With 40 meq/l Humic Acid NdCO E-11 Yb(CO3)2 1.12E-11 Nd(CO3)2 2.69E-11 YbCO E-11 NdPO4 5.42E-12 YbPO4 2.83E-12 Nd E-13 Yb(PO4) E-13 NdSO E-13 YbOH E-14 NdOH E-13 Yb E-14 NdHumic E-13 YbHumic E-14 a Concentrations in molar units; for normal conditions and humic acid model, see text. indicate that carbonate complexes dominate the solution speciation under these conditions. Even with a doubling of the phosphate concentration to 6 mm and a lowering of the ph to 7.4 (the lower bound of observed ph s), carbonate complexes are still dominant. [22] To examine under what circumstances organic complexes might become important, calculations were performed with 150 mm oxalate added to the system. Under these conditions carbonate complexes were still the major REE species, though REE-oxalate complexes were significant (Table 1). However, 150 mm oxalate represents 13 mg/l DOC, significantly higher than the 3 4 mg/l DOC typically observed in the lower Mississippi [Barber et al., 1995; Garbarino et al., 1995; Leehneer et al., 1995]. Thus, for organic complexation to be a dominant factor in lower Mississippi River REE speciation, the organic ligands would need to have significantly higher REE stability constants than REE-oxalate complexes. [23] As an alternative method of examining REEorganic complexation, Viers et al. [1997] estimated the La 3+ -humic acid carboxylate stability constant based on a linear free energy relationship between other measured metal-humic stability complexes and first hydrolysis constants. Using Viers et al. s relationship and the REE-hydrolysis constants of Klungness and Byrne [2000], the following logk s for REE-humic complexes are estimated: La 3+ = 3.77, Nd 3+ = 4.03, Yb 3+ = Note that this La constant is 0.45 log units lower than that reported by Viers et al., presumably because they selected a different value of the La 3+ first hydrolysis constant. Following Viers et al., a Mississippi River DOC value of 4 mg/l yields an estimated 40 meq/l of humic/carboxylate functionality (assuming all DOC is humic acid). Using the above data plus Viers et al. s estimated major element-humic stability constants, results in only about 0.1% of the Nd and Yb being complexed by humic acids. Even doubling the humic acid concentration and increasing the REE-humic stability constants by 0.45 units 10 of 14

12 still results in humic acids accounting for only a fraction of a percent of the dissolved REE concentrations. Although the Viers et al. s approach is simplistic in ignoring electrostatic interactions, nonetheless it would seem that this calculation also suggests that organic complexation is not a dominant factor in Mississippi River dissolved REE speciation. 4. Summary and Conclusions [24] These REE observations are in accord with previous interpretations of seasonal variability of dissolved trace elements in the lower Mississippi River [Shiller, 1997]. In that previous work, hydrological factors and adsorption variability resulting from changes in ph and suspended load were found to be unable to account for the observed concentration variability (e.g., Figure 1). Redox processes were suggested to be the most likely cause of dissolved trace element seasonal variability in the lower Mississippi River. Specifically, for Mn and Fe, changes in the local dynamic balance of reduction (input) and oxidation (removal), as affected by water temperature, might account for the observed seasonal variability of these two elements. Seasonal changes in the dissolved-particulate partitioning of Mn and Fe could then seasonally alter the dissolved concentrations of strongly sorbed trace elements such as Pb and Zn. For elements such as Mo and V, seasonal variability in dissolved concentrations was suggested to result from seasonal flushing of upstream elementdepleted reducing environments (e.g., stratified lakes and bogs) into the river system. [25] In this report it is observed that the light REEs show a seasonality similar to other particle-reactive trace elements whereas there is much less variability in the concentrations of the heavy REEs. The light REE seasonal cycles do not co-vary with discharge, suspended load, ph, or the proportion of water in the lower Mississippi that has come from any of the major tributaries. Calculations suggest that carbonate complexes dominate the dissolved REE speciation and that organic complexes are probably of minimal importance. Previous work [e.g., Sholkovitz, 1995, and references therein] has concluded that in waters dominated by carbonate complexation, there should be preferential adsorption of light REEs relative to the heavies. It is thus reasonable to conclude that the light REE variability in the lower Mississippi results from the same mechanism that controls the variability of particle-reactive Pb and Zn. Likewise, given the dominance of carbonate complexation in the Mississippi River as well as the aforementioned stronger increase in REE solution stability constants than sorption constants with increasing atomic number, the observed limited variability of the heavy REEs is also reasonable. [26] The time series shows that there is greater fractionation of Ce relative to its neighboring REEs during summer when light REEs appear to be most strongly removed to the particulate phase as well as when dissolved Mn is lowest (Figures 3, 4, and 6). This greater fractionation (i.e., lower Ce/Ce*) implies more Ce oxidization during summer. As discussed above, Ce oxidation has been linked both with oxidized Mn [Ohta and Kawabe, 2001; de Carlo et al., 1998] and microbial Mn oxidation [Moffett, 1994]. The seasonal variability of the Ce anomaly is thus consistent with the hypothesis that warmer waters increase the rate of microbial Mn oxidation which thereby increases the amount of adsorbing surface of freshly precipitated Mn oxides and hence results in greater summer scavenging of particle-reactive elements such as Pb, Zn, and the light REEs. This interpretation should be viewed cautiously given the slight correlation between the cerium anomaly and the relative contribution of the Ohio River to the discharge of the lower Mississippi River (Figure 5). [27] The time series included samples collected when waters from record floods on the Missouri and upper Mississippi Rivers reached the sampling location at Baton Rouge during late summer However, no effect was observed on the seasonal patterns of REE concentrations. Other dissolved trace elements likewise showed no obvious effect of this extreme hydrologic event [Shiller, 1997]. These observations confirm the dominance of biogeochemical factors, as opposed to hydrological factors, in determining dissolved REE concentrations in the lower Mississippi River. 11 of 14

13 Appendix A Table A1. Dissolved Rare Earth Element Concentrations in pm a Date La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 10/17/ /11/ /5/ /19/ /21/ /11/ /5/ /16/ /8/ /5/ /25/ /27/ /24/ /17/ /7/ /6/ /9/ /15/ /22/ /7/ /12/ /23/ /28/ /13/ /25/ /24/ /21/ /18/ /17/ Mean a Mean values are discharge-weighted. Table A2. Dissolved Rare Earth Element Concentrations in pm a Tributary La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Ce/Ce* Gd/Gd* Upper Mississippi River (Keokuk, IA) 4/23/92 Des Moines River 4/24/ Upper Mississippi River (Winfied, Mo) 10/30/91 Upper Mississippi River (Winfied, Mo) 4/26/92 Illinois River (Valley City, IL) 10/30/91 Illinois River (Hardin, IL) 4/26/92 Kaskaskia River 4/30/ Missouri River (St. Charles, MO) 11/3/91 Missouri River (St. Charles, MO) 4/29/92 Ohio River (Olmstead, OH) 11/6/91 Ohio River (Olmstead, OH) 5/3/ of 14

14 Table A2. (continued) Tributary La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Ce/Ce* Gd/Gd* Arkansas River 5/5/ Yazoo River 11/10/ Yazoo River 5/7/ Mississippi River (St. Francisville, LA) 5/8/ a Tributary and mainstem locations are arranged in downstream order. Where no location is given, tributary was sampled within 2 km of its confluence with the Mississippi. Ce/Ce* and Gd/Gd* are the cerium and gadolinium anomalies, respectively, calculated as described in the text. Acknowledgments [28] I thank the Louisiana DEQ for providing sampling time on the Mississippi River and R. Meade (USGS) for the opportunity to sample the tributaries. I also thank Z. Chen for performing the analyses, R. Hannigan for encouraging this work, and H. E. Taylor for supplying the USGS reference waters. The manuscript was improved by the comments of the reviewers and the editor. Sample collection and processing was funded by the US EPA (Grant R817007); analysis and interpretation were funded by the US NSF (EAR ). References Barber, L. B., J. A. Leehneer, C. F. Tabor, G. K. Brown, T. I. Noyes, and M. C. Noriega, Organic compounds and sewagederived contaminants, in Chemical Data for Water Samples Collected During Four Upriver Cruises on the Mississippi River Between New Orleans, Louisiana and Minneapolis, Minnesota, May 1990 April 1992, edited by J. A. Moody, chap. 5, U.S. Geol. Surv. Open File Rep., , , Bau, M., and P. Dulski, Anthropogenic origin of positive gadolinium anomalies in river waters, Earth Planet. Sci. Lett., 143, , Briggs, J. C., and J. F. Ficke, Quality of rivers of the United States, 1975 water year Based on the National Stream Quality Accounting Network (NASQAN), U.S. Geol. Surv. Open File Rep., , Byrne, R. H., and K. H. Kim, Rare earth element scavenging in seawater, Geochim. Cosmochim. Acta, 54, , Charlton, S. R., C. L. Macklin, and D. L. Parkhurst, PHREEQ- CI A graphical user interface for the geochemical computer program PHREEQC, U.S. Geol. Surv. Water Resour. Invest. Rep., , 9 pp., de Carlo, E. H., X.-Y. Wen, and M. Irving, The influence of redox reactions on the uptake of dissolved Ce by suspended Fe and Mn oxide particles, Aquat. Geochem., 3, , Field, P. H., and R. M. Sherrell, Magnetic sector ICPMS with desolvating micronebulization: Interference-free subpicogram determination of rare earth elements in natural samples, Anal. Chem., 70, , Garbarino, J. R., R. C. Antweiler, T. I. Brinton, D. A. Roth, and H. E. Taylor, Concentration and transport for selected inorganic constituents and dissolved organic carbon in water collected from the Mississippi River and some of its tributaries, July 1991 May 1992, U.S. Geol. Surv. Open File Rep., , Goldstein, S. J., and S. B. Jacobsen, Rare earth elements in river waters, Earth Planet. Sci. Lett., 89, 35 47, Kim, K.-H., R. H. Byrne, and J. H. Lee, Gadolinium behavior in seawater: A molecular basis for gadolinium anomalies, Mar. Chem., 36, , Klungness, G. D., and R. H. Byrne, Comparative hydrolysis behavior of the rare earths and yttrium: The influence of temperature and ionic strength, Polyhedron, 19, , Lee, J. H., and R. H. Byrne, Examination of comparative rare earth element complexation behavior using linear free-energy relationships, Geochim. Cosmochim. Acta, 56, , Leehneer, J. A., T. I. Noyes, and P. A. Brown, Data on natural organic substances in dissolved, colloidal, suspended-silt and -clay, and bed-sediment phases in the Mississippi River and some of its tributaries, , U.S. Geol. Surv. Water Resour. Invest. Rep., , Martell, A. E., and R. M. Smith, Critical Stability Constants, vol. 1 6, Plenum, New York, Meade, R. H., (Ed.), Contaminants in the Mississippi River, , U.S. Geol. Surv. Circ., 1133, Moffett, J. W., The relationship between cerium and manganese oxidation in the marine environment, Limnol. Oceanogr., 39, , Ohta, A., and I. Kawabe, REE(III) adsorption onto Mn dioxide (d-mno 2 ) and Fe oxyhydroxide: Ce(III) oxidation by d- MnO 2, Geochim. Cosmochim. Acta, 65, , Schijf, J., and R. H. Byrne, Stability constants for mono- and dioxalato-complexes of Y and the REE, potentially important species in groundwaters and surface freshwaters, Geochim. Cosmochim. Acta, 65, , Shiller, A. M., Dissolved trace elements in the Mississippi River: Seasonal, interannual, and decadal variability, Geochim. Cosmochim. Acta, 61, , Shiller, A. M., and T. L. Hebert, Biogeochemical controls on dissolved trace elements in rivers, Mineral. Mag., 62A, , Sholkovitz, E. R., Rare earth elements in marine sediments and geochemical standards, Chem. Geol., 88, , of 14

15 Sholkovitz, E. R., The aquatic chemistry of rare earth elements in rivers and estuaries, Aquat. Geochem., 1, 1 34, Taylor, S. R., and S. M. McLennan, The Continental Crust: Its Composition and Evolution, 312 pp., Blackwell Sci., Malden, Mass., Verplanck, P. L., R. C. Antweiler, D. K. Nordstrom, and H. E. Taylor, Standard reference water samples for rare earth element determinations, Appl. Geochem., 16, , Viers, J., B. Dupré, M. Polvé, J. Schott, J.-L. Dandurand, and J.-J. Braun, Chemical weathering in the drainage basin of a tropical watershed (Nsimi-Zoetele site, Cameroon): Comparison between organic-poor and organic-rich waters, Chem. Geol., 140, , of 14