Environ. Sci. Technol. 2003, 37, 1766-1774
Geochemical Signature of Contaminated Sediment Remobilization Revealed by Spatially Resolved X-ray Microanalysis of Annual Rings of Salix nigra T R A C Y P U N S H O N , * ,† P A U L M . B E R T S C H , ‡ ANTONIO LANZIROTTI,§ KEN MCLEOD,‡ AND JOANNA BURGER† Consortium for Risk Evaluation With Stakeholder Participation, Environmental and Occupational Health Sciences Institute, Division of Life Sciences, Rutgers University, 604 Allison Road, Piscataway, New Jersey 08854, Savannah River Ecology Laboratory, University of Georgia, Drawer E, Aiken, South Carolina 29802, and Consortium for Advanced Radiation Sources, The University of Chicago, 5640 S. Ellis Avenue, Chicago, Illinois 60637
An X-ray microprobe was used to determine the concentration and distribution of Ni, U, and other metals within annual rings of willows (Salix nigra L.) from a former de facto radiological settling basin (Steed Pond; SP) and a depositional environment downstream (Tims Branch; TB) on the Savannah River Site (SRS). Geochemical and historical information about both areas are well documented. Following spillway breaches at SP in 1984 and the early 1990s, TB is inundated with contaminated sediments during storms. Bulk elemental composition of tree cores was determined using ICP-OES. Synchrotron X-ray fluorescence (SXRF) analysis showed that the metal contents of SP and TB cores were an order of magnitude higher than those from a reference site. TB cores were enriched with Ni in 1984 and 1991, corresponding with SP spillway breaches (containing 790 mg kg-1 Ni in 1991). Cores from SP exhibited an extremely high Ni peak in 1996, approximately 5000 mg kg-1, even though contaminant levels at SP did not change. The geochemical signature of contaminants recorded in TB annual rings reflected the significant sediment remobilization events consistent with the detailed history of the site, and at concentrations relative to their proximity to the source term. However, physiological processes occurring within impacted trees strongly influence the chronological accuracy of dendroanalysis and must be investigated further.
Introduction Contradictory findings in the literature have questioned the validity of dendroanalysis (1). A greater understanding of * Corresponding author phone: (803)725-5956; fax: (803)7253309; e-mail:
[email protected]. Mailing address: Savannah River Ecology Laboratory, University of Georgia, Drawer E, Aiken, South Carolina 29802. † Consortium for Risk Evaluation With Stakeholder Participation, Rutgers University. ‡ Savannah River Ecology Laboratory, University of Georgia. § Consortium for Advanced Radiation Sources, The University of Chicago. 1766
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 9, 2003
influential factors requires the use of analytical techniques with greater spatial resolution than traditional solution techniques, such as inductively coupled plasma-optical emission spectroscopy (ICP-OES) (2). A range of spatially resolved microspectroscopic techniques have been applied to dendrochemical analysis, including laser ablation inductively coupled plasma-mass spectrometry (3-5), secondary ion mass spectrometry (6), particle-induced X-ray emission (7), and more recently microbeam-synchrotron X-ray fluorescence (8-10). The spatial distribution of metals and metalloids has in the past been assumed to be conservative enough within individual annual rings to allow retrospective biomonitoring (11) and to make assumptions about past environmental conditions (12-14). This assumption has been criticized, however, after recognition of radial mobility of metal and radionuclide elements (15-17), indicating the significance of physiological processes to metal distribution in lignified vascular tissue. Current scientific opinion is therefore divided on the ability of dendroanalysis to provide accurate information about the exposure history and severity of past contamination events (6). This arises from a lack of understanding of mechanisms of metal transport and partitioning within the vascular systems of woody plants. Synchrotron X-ray fluorescence (SXRF) microanalysis is a nondestructive technique that can provide the spatially resolved information required to investigate the physiological influences that affect dendroanalysis, and is herein applied to a situation where both the magnitude and duration of contamination events are notable, and have been documented in sufficient detail. The Steed Pond (SP) former radiological settling pond offers a unique setting in which to evaluate these issues because the levels of contaminants, primarily uranium (U) and nickel (Ni), have been rigorously monitored. SXRF was used to determine spatial metal distribution within annual rings of black willow (Salix nigra L.) growing in SP and an impacted depositional area on Tims Branch (TB), 3 km downstream. Historical evidence holds that effective containment of metal and radionuclide contaminated sediments at SP was compromised on more than one occasion, through breaches in the enclosing wooden spillway (18). Following the breach, erosion of SP during episodic storm events is a significant ongoing process (19). Previous studies at SP have shown contrasting bioavailability of contaminants, in favor of greater Ni and negligible U bioavailability (20, 21). Evidence also suggests transfer of Ni into plant tissue, both at SP (21) and TB (22, 23). The breaches in the wooden spillway enclosing SP represent highly significant contamination events, whose effects may be detectable within the annual rings of woody vegetation downstream. The aim of this study was to investigate whether trees from various locations, known to have been impacted by metal and or radionuclide contamination, contained a geochemical signature within their annual rings that was indicative of changes in exposure. We also sought to provide quantitative estimates for metal abundances preserved within annual rings of trees collected from the source term (SP) compared to those of impacted areas further downstream (TB) and a reference uncontaminated area. Further, we hoped to evaluate whether metal distributions within, as compared to between, annual rings of enriched samples might provide information about the physiological influences upon dendrochemical analysis for retrospective biomonitoring. 10.1021/es0261628 CCC: $25.00
2003 American Chemical Society Published on Web 04/01/2003
FIGURE 1. Location of Steed Pond (SP), Tims Branch (TB), and Boggy Gut (BG) on the Department of Energy’s Savannah River Site, SC.
Experimental Section Study Site. The Steed Pond-Tims Branch (SP-TB) system is located in the northwest corner of the Savannah River Site (SRS), a 777 km2 U.S. Department of Energy facility in the upper coastal plain of South Carolina (24) (Figure 1). Steed Pond (SP) is an abandoned farm pond which functioned as a de facto settling basin for waste discharges arising from production of aluminum-clad U targets used in the production of nuclear materials for weapons. Discharges occurred between the mid-1950s and 1985; 61% between 1966 and 1968 (18). Ninety-seven percent of the total gross R-activity released by the entire SRS was discharged directly in to the SP-TB stream system. It is estimated that approximately 44 000 (( 26 000) kg of natural and depleted U was released directly into the system (24). This was accompanied by similar amounts of Ni, and lesser amounts of Al, Cu, Zn, Pb, and Cr. Associated wastes such as nitric, phosphoric, and sulfuric acids were also discharged (25). Strong co-association between inorganic contaminants has also been noted in geochemical studies of the source term, especially between U and Ni (20). Steed Pond originally covered 5.7 ha, but was reduced to 4.5 ha after partial failure and repair of the enclosing wooden spillway in the 1960s. Subsequent spillway breach incidents occurred in 1984 after which it was never repaired. This breach lowered the water level by 4.5 m. Beaver activity resulted in a partial refilling of the pond, although in the early 1990s the animals were removed and the beaver dam was dismantled; activities which coincided with the end of a drought in the region (18). The total soil concentrations (i.e., HF + HNO3 extractable) of U and Ni in SP and TB (Table 1) are well in excess of baseline levels for the Savannah River Site (SRS) (2 mg kg-1 for U; 10 mg kg-1 for Ni) (25); with U and Ni (mg kg-1) concentrations of 1065 (( 54) and 526 (( 18) at SP, and 289 (( 166) and 130 (( 90) at TB. Prior to the dam failure, overflight data estimated that, of the original releases, approximately 75% of the U contamination remained within SP. Following the failure of the spillway, vegetation colonized the majority of the exposed sediments with the exception of
several localized areas of elevated Ni and U (26), which were prone to erosion. Previous studies showed that there were no significantly elevated U concentrations in TB sediments before dam failure (27). Sediment transport studies conducted after the spillway breach by Batson et al. (19) indicated that sediment erosion during typical storm events (on average two per month) increased U export to between 1500 and 2800% of base flow. Sample Collection. Samples were collected from individual Salix nigra trees at a height of 36 in. above the base of the tree, using a Teflon-coated increment corer (20 cm length, 0.5 cm i.d.). Core samples were collected from three study sites. The first site was an uncontaminated site known as Boggy Gut (BG) located on the SRS boundary, approximately 11 km northeast of Steed Pond (SP) (Figure 1). This is a riparian ecosystem located on the Upper Three Runs stream system, above the M-area outfall. Soil analyses show the system to be uncontaminated (Table 1), compared to the background Ni and U concentrations for the SRS of 10 and 2 mg kg-1, respectively (24). Samples were collected from SP, from an individual growing approximately 5 m from the breached wooden spillway (SP1), which was found to contain elevated concentrations of Ni within the leaves of 75.5 ((3.6) mg kg-1 (21). Rhizosphere Ni and U concentrations at this point were 18 and 133 times higher, respectively, than those at BG (Table 1). In other areas of the pond, concentrations of Ni and U reached 1224 ((32) and 2480 ((230) mg kg-1, respectively (20). Samples were also collected from TB, regarded as the impacted depositional site for the study, located 3 km southeast of the source term. Soil contaminant concentrations at TB are by comparison intermediate between SP and BG, with Ni and U present in streamside soils at levels 9 and 36 times higher than those at BG (Table 1). Tree-core-sampling excursions took place in midmorning to account for diurnal fluctuations in sap flow. Six individual trees growing within 10 m of the stream were sampled at TB (TB1-6), because of evidence of a more heterogeneous spatial distribution of soil contaminants (23). Multiple samples were collected to identify an individual that contained evidence of metal enrichment. Because of constraints placed upon scheduling and beamtime availability, SXRF analyses concentrated on one individual from each site, and specifically from SP and TB, samples were from individuals known to be impacted, following confirmation by inductively coupled plasma-optical emission spectroscopy (ICP-OES) analysis of triplicate samples. Five vertically adjacent replicate core samples were collected from each tree; three for determination of bulk elemental composition by ICP-OES, one for identification of annual ring positions, and one for analysis by SXRF. All coring equipment was rinsed with deionized water between samples. Immediately after collection, samples were sealed in plastic bags (which were marked with careful note of orientation) and stored on ice for transportation back to the laboratory. Cores were dried in an air circulation oven at 60 °C for 48 h. Bulk Elemental Analysis. Tree-core samples were ground separately in a Wiley mill to pass through a 1-mm sieve and homogenized, and three 0.25-g subsamples of each core were digested in 0.25-g aliquots with 10 mL of 5 M HNO3 (tracemetal-free grade) + H2O2 using a CEM 2000 microwave digestion oven and Teflon PFA vessels. An HNO3 blank and sample of SRM1515 (Apple Leaves standard reference material, National Institute of Standards) were also included in each batch of microwave digestions as quality control samples. Microbeam Synchrotron X-ray Fluorescence Analysis. Dried tree cores were sliced into 2-cm transverse sections, retaining the original orientation with respect to the bark and center of the tree, and sliced longitudinally with a surgical steel blade. Sections of approximately 1 mm thick were VOL. 37, NO. 9, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1767
TABLE 1. Concentration of Contaminants of Concern Detected within Total Extracts (HF + HNO3) of Soil from the Rhizosphere of Sampled Trees at Boggy Gut, Steed Pond, and Tims Brancha
a
site
Ni
Cr
Cu
Zn
Pb
U
Boggy Gut (BG) Steed Pond (SP)
14.8 (0.2) 526 (18)
8.6 (1.0) 78 (3.9)
2.3 (0.2) 45 (1.9)
2.5 (1.3) 85 (5.2)
27 (2.5) 52 (2.3)
8.46 (0.2) 1062 (54)
1 2 3 4 5 6
138 (46) 132 (9.2) 348 (50) 125 (10) 148 (16) 119 (30)
32 (12) 27 (1.9) 47 (8.0) 35 (2.0) 28 (2.4) 33 (7.5)
Tims Branch (TB) 21 (0.6) 11 (1.8) 26 (0.6) 16 (1.4) 18 (1.0) 13 (1.1)
57 (9.4) 49 (4.4) 60 (1.2) 47 (23) 42 (0.4) 48 (6.4)
29 (5.7) 27 (1.5) 40 (2.2) 23 (1.5) 26 (1.1) 25 (0.9)
322 (79) 299 (69) 666 (19) 268 (52) 363 (32) 295 (36)
Results are reported in mg kg-1 DW and as means ( SD, where n ) 3.
mounted uncovered on Kapton tape (trace-metal free), and attached to cardboard 35-mm slide mounts. SXRF microbeam analyses were conducted at the X26A Beamline at the National Synchrotron Light Source, Brookhaven National Laboratory (28-30). The incident X-ray beam was tuned to 17.5 keV using a Si(111) channel-cut monochromator. For larger-scale dendroanalysis of core sections, a monochromatic beam collimated to 350 × 350 µm with tantalum shutters was used. For smaller scale mapping the beam was focused to 10 × 10 µm using rhodium-coated Kirkpatrick-Baez focusing optics (31). Energy-dispersive XRF compositional data were collected using a Canberra SL30165 Si (Li) detector. Both the sample and detector sit in air. Detection limits for most of the elements analyzed here vary between 0.1 and 10 mg kg-1 under the conditions of data collection. All one-dimensional line scans were carried out using a step size of 350-500 µm and a count time of 90 s pixel-1. Two-dimensional composition maps used count times of 35 s and 6 s for larger scale and fine scale interrogation, respectively. In practice, the SXRF analyses of trace elements in these samples are restricted to the energy interval 3-17.5 keV. The sensitivity is poor at low energy (
