Journal of Geophysical Research: Biogeosciences

Disturbance impacts on mercury dynamics in northern Gulf of Mexico sediments

Authors


Abstract

[1] We evaluated the impacts of physical disturbance on sediment Hg biogeochemistry in the northern Gulf of Mexico by exploring changes in Hg abundance and speciation in cores collected between July 2005 and July 2006, a time period that included the passages of Hurricanes Katrina and Rita in this region. Comparisons across space and time reveal large changes in sediment characteristics, both with respect to Hg biogeochemistry and other measures of sediment composition, that indicate substantial disturbance. While the degree of disturbance varied between stations, at sites that clearly received new surface deposits considerable increases in OC and total Hg were observed, along with shifts in major element composition. At the station closest to Hurricane Katrina's track (station A′2), 210Pb levels are consistent with the episodic deposition of >10 cm of sediments. These surface sediments (0–10 cm) at A′2 had the highest %MeHg of all stations and all dates, suggesting that the disturbance resulted initially in increased net methylation. While the observed disturbances elsewhere could not in all cases be definitively linked to hurricane activities, the substantial thickness of deposits (>10 cm) at multiple sites is consistent with a major event, and the similarity in the deposits' chemical fingerprint across all impacted sites suggests similar sources or processes. We estimate that the two hurricanes redistributed approximately 5 times the annual Hg input from the Mississippi-Atchafalaya River system and atmospheric deposition. These observations highlight the need to consider the effects of major disturbances on the biogeochemical cycling of Hg in coastal systems.

1. Introduction

[2] Mercury in the form of methylmercury (MeHg) is a neurotoxin that biomagnifies in aquatic food chains [World Health Organization (WHO), 1990]. The conceptual framework for understanding Hg cycling and MeHg production in aquatic systems has advanced considerably over the past decade, and evidence points to sediments as both important repositories of inorganic Hg and locations where substantial Hg methylation takes place [e.g., Gilmour et al., 1992, 1998; Bloom et al., 1999; Mason et al., 1999; Benoit et al., 2003; Hammerschmidt and Fitzgerald, 2004, 2006; Sunderland et al., 2004, 2006; Lambertsson and Nilsson, 2006]. Hg methylation is influenced by a number of interdependent factors [Benoit et al., 2003; Hammerschmidt and Fitzgerald, 2004], with net MeHg production readily shifting in response to physical [Tseng et al., 2001; Heyes et al., 2004; Kim et al., 2004, 2006; Sunderland et al., 2004] and chemical [Benoit et al., 2003] perturbations. While coastal sediments can act as long-term Hg sinks, they are also prone to major episodic disturbances, such as hurricanes and other storms [Allison et al., 2000, 2005; Goñi et al., 2006, 2007; Walsh et al., 2006; Dail et al., 2007] that could alter Hg cycling and effective burial. Despite the potential impact of major perturbations on the Hg cycle, there is limited information about how large-scale sediment disturbances might alter Hg cycling and speciation.

[3] Studies of seasonal Hg cycling in freshwater [e.g., Gilmour et al., 1992, 1998] and coastal [e.g., King et al., 1999, 2000; Mason and Lawrence, 1999; Hammerschmidt and Fitzgerald, 2004, 2006; Heyes et al., 2006] ecosystems demonstrate that most MeHg is produced in anaerobic sediments by sulfate reducing bacteria (SRB). In marine systems, mercury methylation rates in sediments are strongly influenced by several interdependent factors, chiefly organic carbon (OC) levels, sulfide chemistry, inorganic Hg (Hg(II)) speciation, and temperature, that affect either SRB activity, the availability of Hg(II) for uptake by SRB, or both [Benoit et al., 2003; Hammerschmidt and Fitzgerald, 2004; Lambertsson and Nilsson, 2006]. OC plays a central and particularly complex role in Hg speciation: OC strongly binds Hg(II) and thereby limits methylation; however, OC also acts as substrate for SRB metabolism, which can lead to higher methylation rates due to enhanced SRB activity [Compeau and Bartha, 1985; Choi and Bartha, 1994], or to decreased methylation rates due to high sulfide levels and the formation of Hg-sulfide complexes that are unavailable for uptake by SRB [Benoit et al., 2003; Hammerschmidt and Fitzgerald, 2004]. Sulfate reducers also participate in demethylation [Oremland et al., 1991; Marvin-DiPasquale and Oremland, 1998], further complicating interpretations of OC-Hg dynamics.

[4] Considering many of these factors' sensitivity to or influence on sediment redox conditions, it is reasonable to predict that physical disturbance will influence MeHg levels in sediments. A few studies have explored the links between Hg biogeochemistry and continuous sediment disturbance, such as the influence of fluid mud flow [Tseng et al., 2001], large tides [Sunderland et al., 2004], natural [Heyes et al., 2004] and simulated tidal resuspension [Kim et al., 2004, 2006]. These studies demonstrated that regular mixing and resuspension events can enhance methylation. Hurricanes episodically disrupt aquatic ecosystems by causing large-magnitude sediment resuspension events [Allison et al., 2005; Goñi et al., 2006, 2007; Turner et al., 2006; Walsh et al., 2006; Dail et al., 2007]. Currents and waves generated from other seasonal major storms can also trigger disturbance and relocation of large volumes of sediments [Allison et al., 2000; Corbett et al., 2007]. These physical disturbances undoubtedly have substantial short-term impacts on sediment redox conditions and should, by extension, impact the cycling of redox sensitive elements such as Hg. Therefore, understanding the impacts of storm-induced sediment disturbance on Hg biogeochemistry may be an important part of accurately assessing human and ecological health risks due to MeHg exposure.

[5] The northern Gulf of Mexico (nGOM) along the Louisiana (LA) coast supports a highly productive fishery [National Oceanic and Atmospheric Administration (NOAA), 2007], and is an area that regularly experiences tropical storms, hurricanes, and strong winter storms [NOAA, 2003]. While there have been a few studies on Hg in nGOM sediments [e.g., Neff, 2002; Trefry et al., 2007], little is known about the factors controlling Hg distribution and cycling or the impacts of major physical disturbances such as hurricanes or other storms on Hg biogeochemistry in the region. In July 2005 we began a study of Hg biogeochemistry in nGOM sediments. In the summer and fall of 2005, two powerful hurricanes, Hurricane Katrina (23–30 August 2005) and Hurricane Rita (18–26 September 2005), struck this region (Figure 1), causing widespread destruction to the Gulf States [Knabb et al., 2006a, 2006b]. The proximity of the hurricanes' paths to our study area presented an unplanned opportunity to explore the impacts of physical disturbance on Hg biogeochemistry, both in the immediate aftermath of the hurricanes and over the subsequent year.

Figure 1.

Sediment sampling locations for the July 2005, October 2005, March 2006, and July 2006 cruises in northern Gulf of Mexico. Stations D3, C6B, and A′2 were occupied during all four cruises. Stations J4, K4, E2A, and G3 were sampled during July 2005 and July 2006.

[6] We hypothesized that the 2005 hurricanes caused substantial sediment disturbance and mixing/redistribution of sediment Hg, and that this disturbance resulted in enhanced net MeHg production. To test these hypotheses, we evaluated changes in Hg abundance, Hg speciation, and ancillary chemical characteristics in sediment cores collected before (July 2005) and after (October 2005) Hurricanes Katrina and Rita, as well as in subsequent months (March 2006, July 2006). We combined our results with observations from other groups [Walsh et al., 2006; Goñi et al., 2007] to assess the extent to which the hurricane disturbances could have impacted Hg distribution in the region and the potential importance of these disturbances relative to the system's annual Hg budget.

2. Methods

2.1. Site Description

[7] The nGOM (Figure 1) along the LA coast accounts for 70% of the overall GOM commercial fishery landings and ∼10% of the overall U.S. catch [NOAA, 2007]. The Mississippi River (MR) and its main distributary, the Atchafalaya River (AR), are the largest sources of fresh water, sediment, nutrients, and terrestrial organic matter to the nGOM [Swarzenski et al., 2008]. The Mississippi-Atchafalaya River system (MARS) drains 47% of the conterminous United States [Corbett et al., 2007], delivering 1 × 109 kg of nitrogen [Goolsby et al., 2001] to the nGOM, and 3 × 1010 kg organic carbon, more than 60% of which is present as particulate organic carbon [Trefry and Metz, 1994]. The riverine nutrient inputs play a major role in enhancing marine primary production, and over the past several decades have resulted in coastal eutrophication and hypoxic bottom waters in the nGOM, with dissolved oxygen at levels below 2 mg l−1 observed annually from spring through late summer over large areas up to 21,000 km2 [Rabalais et al., 2002, 2007; Turner et al., 2007]. The Mississippi and Atchafalaya Rivers also deliver 2.2 × 1011 kg a−1 of suspended sediments [Goñi et al., 2007], which accumulate as nGOM bottom sediments at rates that range over several orders of magnitude across the LA shelf [Gordon and Goñi, 2004; Corbett et al., 2006].

2.2. Field Sampling

[8] In July 2005 and 2006, box cores were collected along a ∼400 km east–west transect aboard the R/V Pelican at sites that span a range of bottom water hypoxia frequencies and sediment conditions (e.g., organic content, sedimentation; Figure 1). After the hurricanes, a rapid response effort was organized, and a sediment sampling cruise was undertaken from 14–16 October 2005 aboard the R/V Cape Hatteras. Cruises in March 2006 and July 2006 (R/V Pelican) allowed us to further observe changes in sediment chemistry, Hg biogeochemistry, and MeHg formation compared to the baseline data obtained in July 2005. The October 2005 and March 2006 cruises focused on stations in the east of the study area. Three sites (Stations D3, C6B, and A′2) were occupied during all four cruises. Data from four other sites (J4, K4, E2A, and G3) that were sampled only during July 2005 and July 2006 are also included to assess the effect of physical disturbance on Hg chemistry over a larger area.

[9] Box cores were subsampled using 6.5 cm i.d. polycarbonate tubes and immediately refrigerated. Within a few hours of collection, pore water dissolved oxygen levels were measured on board the ship using a microelectrode that was lowered into the sediments incrementally at millimeter resolution. At a subset of stations, 2–3 cores were sectioned into 1 cm or 2 cm increments in a N2 glove box aboard the ship and then either centrifuged (July 2005, March 2006, July 2006) or vacuum-filtered (October 2005). Pooled supernatant was filtered through 0.2 μm cellulose acetate filters into acid-cleaned glass vials, acidified to 0.5% HCl, and immediately frozen. In October 2005, March 2006, and July 2006, an aliquot of pore water was added to sulfide antioxidant buffer, and sulfide was measured using an Orion Sulfide Electrode and Double Junction Reference Electrode [Brouwer and Murphy, 1994], with a minimum detection limit of ∼0.1 μM. Additional cores were subsectioned aboard the ship into acid-washed plastic or glass containers and immediately frozen for analysis of solid phase total Hg (THg) and MeHg, acid volatile sulfide (AVS), total organic carbon (OC), and other ancillary parameters.

2.3. Laboratory

[10] Total Hg in July 2005 cores was determined by thermal decomposition, amalgamation, and atomic absorption spectrophotometry (USEPA Method 7473; Milestone DMA-80), calibrated using dissolved Hg(II) standards and MESS-3 standard reference material (91 ± 9 ng g−1, National Research Council of Canada). For all other dates, sediment THg concentrations were determined by extraction of THg from sediment using sulfuric/nitric acid (3:7 v/v) [Kim et al., 2004; Heyes et al., 2006], and detection by Inductively Coupled Plasma Mass Spectrometry (ICP-MS, ELAN® DRC II, PerkinElmer, Shelton, CT) coupled with a flow injection analysis system (FIAS; PerkinElmer, Shelton, CT). The FIAS reduces Hg2+ in the aliquot using SnCl2 to a cold vapor of Hg0 which is subsequently detected by ICP-MS. An enriched isotope spike (200Hg) was added to all samples prior to digestion and was used as a procedural internal standard. The median of the relative standard deviations (RSD) of all the replicate samples (n = 61) was 2.8%, and the operational method detection limit was 0.8 pmol g−1, based on three times the standard deviation of the digestion blanks (n = 15). Digestion and analysis of MESS-3 yielded recoveries of 110 ± 7% (mean ± 1 standard deviation, n = 17). Total Hg in pore water was analyzed by cold vapor atomic fluorescence spectrometry (CVAFS) after reduction to Hg0 using SnCl2 [Gill and Fitzgerald, 1987; U.S. Environmental Protection Agency (USEPA), 2002]. The method detection limit was 0.02 ng l−1 (0.1 pM) based on three times the standard deviation of blanks (n = 9).

[11] MeHg in sediment was measured by CVAFS (July 2005) or ICP-MS (October 2005, March 2006, July 2006) after distillation of MeHg from sediment, followed by ethylation and trapping of methyl-ethyl-Hg on Tenax® traps [Bloom, 1989; Horvat et al., 1993; Mason et al., 1999; Heyes et al., 2004]. The median of RSD of the replicate samples (n = 49) was 22%. A comparable RSD (14–25%) was found for MeHg measured in replicate cores collected from the same box core (n = 12). Distillation blanks all fell below instrument detection limits. Analyses of a standard reference material, IAEA-405 sediment (MeHg = 5.49 ± 0.53 ng g−1), gave recoveries of 95 ± 13% (n = 24). In addition, a subset of nGOM sediment samples were spiked with 200 pg MeHg to monitor distillation recovery (96%, n = 36).

[12] Total organic carbon was determined by combustion, GC separation, and thermal conductivity detection (Perkin Elmer 2400, PerkinElmer, Shelton, CT) on freeze-dried sediment samples that had been acidified with 1 M HCl to remove inorganic carbon. AVS was analyzed by distilling frozen sediment samples in deoxygenated 1M HCl at room temperature under N2 gas flow, trapping evolved H2S in a zinc acetate buffer, and detecting sulfide colorimetrically [USEPA, 1996]. Concentrations of major and trace elements (i.e., Al, Si, K, Fe, Zn, Br, Rb, Sr, Cl, Ca, Mn, and Pb) in the sediments were analyzed by polarized energy-dispersive X-ray fluorescence (XRF) spectroscopy using a Spectro XEPOS® (Spectro Analytical, Kleve, Germany). NIST SRM 1944 (New York/New Jersey Waterway Sediment) was used to determine the accuracy of the measurements, and we obtained 95 ± 1% comparing against the certified elements (Al, Mn, Fe, Ni, Zn, and Pb). Procedures for 210Pb, δ13COC, and OC analysis for the long core at A′2 (October 2005) are described in detail by Swarzenski et al. [2006, 2007]. Briefly, sediments for 210Pb analysis were dried, powdered, and counted in a calibrated high-purity Ge well-type gamma detector with a precision typically <5%. Sediments for δ13COC and OC were dried, homogenized, acidified to remove inorganic carbon, and detected using Micromass Optima continuous-flow mass spectrometer and Shimadzu OC 5000A analyzer, respectively.

3. Results

3.1. July 2005

[13] Sediment characterizations focused primarily on the surficial 10 cm, the zone most important for contemporary Hg dynamics in sediments. Figure 2 summarizes the concentrations of sediment Hg data and OC from Stations A′2, C6B, D3, J4, K4, E2A, and G3. The July 2005 data serve as initial conditions for the purpose of evaluating changes over the study period. Sediments sampled at these seven sites contained low OC (mean = 0.53%; range = 0.10–1.09%). Despite the low OC levels, pore water O2 decreased to below the detection limits within the top 1 cm of the sediment (data not shown), indicating that reducing conditions and limited bioturbation are found at these sites, in agreement with past observations of limited bioturbation in the area [Rabalais and Turner, 2001]. Total Hg levels at the July 2005 stations averaged 132 ± 50 pmol g−1. The percentage of Hg present as MeHg (%MeHg) ranged from 0.2 to 1.3% (mean = 0.62% ±0.28%), similar to observations in Chesapeake Bay (CB) (∼1% [Mason et al., 1999]), Lavaca Bay (0.6% [Bloom et al., 1999]), Bay of Fundy (0.6% [Sunderland et al., 2004]), and Long Island Sound (LIS) (0.7% [Hammerschmidt and Fitzgerald, 2004]). The vast majority of Hg in sediments was solid-complexed: pore water THg concentrations ranged from 2.6 to 62 pM, yielding log KD values that ranged from 3.1 to 5.0 l kg−1 (log KD = log [Hgsolid]/[Hgdissolved]) for THg. These values are comparable to log KD observed in LIS (3.2–4.8 l kg−1 for Hg(II) [Hammerschmidt and Fitzgerald, 2004]) and the Chesapeake Bay and associated shelf (3.5–5.5 l kg−1 for THg [Hollweg et al., 2007]). However, considering the lower OC content in nGOM relative to other coastal systems (e.g., mean OC was 2.5% and 2.8% in LIS and CB/shelf, respectively, assuming ∼40% of total organic matter is OC), the nGOM sediments have a higher KD per unit OC.

Figure 2.

Sediment profiles of THg, OC, %MeHg, and log KD at A′2, C6B, D3, E2A, G3, J4, and K4 across multiple sampling events.

3.2. October 2005

[14] A comparison of July 2005 and October 2005 data indicates substantial changes in sediment composition at A′2, the station nearest to Hurricane Katrina's path, consistent with physical disturbance of sediments. THg and OC at A′2 increased by an average of 86% and 270%, respectively, between July 2005 and October 2005. Increases in Fe, Mn, and Al concentrations mirror the changes in OC and Hg, and Si levels shifted in the opposite direction (Figure 3). These shifts in major element composition suggest that the surface sediments at A′2 are entirely different material, as opposed to mere mixing of previous sediments with the addition of new OC and Hg. In a longer core collected at A′2 in October 2005, OC levels decreased sharply at 12cm, from >2% OC at the surface to prehurricane levels (<0.8%) in deeper sediments (Figure 4). Radioisotope data from A′2 provide further evidence of disturbance (Figure 4). Excess 210Pb (t1/2 = 22.3 yr) was detected at relatively constant levels over the surficial 10 cm at A′2, with an abrupt decrease in activity below 10 cm. The discontinuity in the 210Pb profile coincides with the shift in OC at A′2, and a shift in δ13COC, with relatively depleted δ13COC below 10 cm. Despite the recent disturbance, the O2 penetration depth in the sediment was again <1 cm at A′2 in October 2005, indicating that anoxic conditions in the sediment were reestablished within weeks of deposition. Dissolved sulfide was undetectable (<0.1 μM) at A′2 (October 2005); sulfide was also below detection limits at all other stations/dates where it was measured. Goñi et al. [2007], who studied hurricane-related sediment disturbance in this region in October 2005 (discussed in more detail below), observed net deposition at stations close to A′2.

Figure 3.

Sediment profiles of selected trace elements (Al, Si, Fe, and Mn) at A′2, C6B, D3, E2A, G3, J4, and K4 across multiple sampling events.

Figure 4.

A′2 sediment profiles in October 2005, shortly after Hurricanes Katrina and Rita: %OC, OC:N, stable isotopic compositions of OC (δ13C), and excess 210Pb.

[15] At Station C6B, changes in THg and OC were evident in the surface 3–4 cm of the core, with greater concentrations of THg and OC in July 2005 than October 2005. Similar shifts occurred for Al, Fe, and Mn (Figure 3). The C6B profile changes are consistent with the scouring of the top few centimeters of the sediment column at this site. At Station D3, the October 2005 and July 2005 sediment depth profiles for THg, OC, and major elements (Figures 2 and 3) were similar over the top 10 cm, as would be expected if little disturbance had occurred between sampling dates. This result agrees with the limited deposits found in this area by Goñi et al. [2007].

[16] Similar to the relatively constant THg levels we observed at D3 between sampling dates, other studies have found that Hg concentrations at a specific site are fairly constant over time. For example, Hollweg et al. [2007] found that total Hg varied by <30% for mid-Atlantic Bight stations that were resampled during 5 cruises over a 2-year sampling campaign. Mason et al. [1998] tested the heterogeneity of sediments within Lavaca Bay, a contaminated estuary in Texas, and found that spatial variability, estimated by sampling within a 25 m radius, was similar to that found for the mid-Atlantic Bight. Additionally, Bloom et al. [1999] sampled a site in Lavaca Bay over a year and found that while there was a measurable change in the %MeHg seasonally, the total Hg concentration was consistent (RSD = 25%, n = 14) and showed no seasonal trend. These results reinforce the conclusion that the changes at A′2 and the surface at C6B resulted from physical disturbance and are not due purely to sediment heterogeneity at the sampling site.

3.3. March 2006 and July 2006

[17] Sediment disturbances continued over subsequent months at A′2 and C6B (Figure 2). Between October 2005 and March 2006, the surficial 10 cm of sediments at A′2 underwent a major shift, returning to OC and THg levels similar to those before the hurricanes. Al, Fe, Mn, and Si underwent comparable shifts (Figure 3). OC and THg levels remained similar at A′2 in March 2006 and July 2006, as did Al, Fe, Mn, and Si levels (Figure 3). Surface sediment chemistry at C6B shifted in the opposite direction, with material containing higher concentrations of THg and OC accumulating between October 2005 and March 2006. Higher Al, Fe, and Mn levels and lower Si levels accompanied the changes in THg and OC. Sediment composition at C6B varied little between March 2006 and July 2006. In contrast, THg, OC, and major element composition at D3 remained relatively constant across all four sampling events, apparently because of the cohesive hard Beaumont clay basement that defined the surface 10 cm of sediments on all four sampling dates and the apparent lack of new deposition in this area.

[18] Sites J4, K4, G3 and E2A were only sampled in July 2005 and July 2006. Comparisons of sediments from J4 and K4, both of which are located within 50 km of Hurricane Rita's path, indicate that substantial sediment disturbance occurred during this time period. At J4 concentrations of both THg and OC nearly doubled between July 2005 and July 2006 (Figure 2), and major shifts in Fe, Mn, Al, and Si could be observed (Figure 3). Comparable changes occurred at K4. Stations E2A and G3 exhibited relatively small changes between July 2005 and July 2006 in THg, OC, and elemental compositions (Figures 2 and 3). At E2A and G3, some depth increments (e.g., 6–10 cm at G3) do appear consistently different between July 2005 and July 2006 across several indicators. However, the overall differences in elemental composition (depth-averaged RSD = 4–14% for individual elements) were small compared to stations that clearly experienced disturbance, and comparable to differences observed at D3 between different sampling trips (RSD = 3–15%). The spatial variability in Hg and OC shifts that we observed (i.e., more change at J4 and K4, and limited change at E2A and G3) are consistent with the spatial pattern of hurricane-induced deposits seen by Goñi et al. [2007].

[19] Seasonal variability and competing biogeochemical processes could mask disturbance-related alterations to Hg speciation, making their detection less straightforward than detecting changes in major element composition or total Hg. For example, %MeHg at A′2, C6B, and D3 was generally higher in July 2006 than March 2006 (Figure 2), likely because fresh OC inputs from spring and summer algal blooms and higher temperatures led to increased methylation. Nonetheless, pronounced changes in Hg speciation induced by sediment disturbance were evident at some stations. For example, a ∼50% increase in %MeHg accompanied the perturbations that occurred at A′2 between July 2005 and October 2005. At J4, %MeHg over the top 5 cm was on average 2 times lower in July 2006 than July 2005, and KD was 1–2 orders of magnitude greater in J4 (July 2006) cores than J4 (July 2005) cores (Figure 2). At other stations/dates, well-defined changes were less evident.

4. Discussion

4.1. Characterizing Disturbance-Related Surface Sediment Deposits

[20] The comparisons of October 2005, March 2006, and July 2006 with July 2005 data reveal large changes in sediment composition at multiple stations. These changes, the extent of which varied considerably between stations, were evident either shortly after the hurricanes' passages or in subsequent months, and in most cases exceeded what could be reasonably explained by spatial heterogeneity. Station A′2 stands as a clear example of a hurricane-impacted site, with excess 210Pb values in October 2005 supporting the interpretation that the altered sediment chemistry was due to recent deposition. Different dynamics shaped changes at C6B, where hurricane impacts apparently scoured surface sediments. The longer term posthurricane data (March 2006 and July 2006) from both Stations A′2 and C6B indicate continued movement of OC- and Hg-enriched sediments.

[21] To further characterize the mobile and stationary sediments, we conducted a principal component analysis (PCA) using elemental concentrations (not including Hg and OC) at A′2, C6B, D3, E2A, J4, K4, E2A, and G3 for July 2005, October 2005, March 2006, and July 2006. The analysis identified two components that explain 85% of the variance in the chemical composition, with the first component (PC1) accounting for 73% of the variance. PC1 has a strong positive loading on Al, suggestive of a higher abundance of clays, along with strong positive loadings of Fe, Mn, Cu, Zn, Ni, and Pb, which are known to have high affinity for fine and clay particles due to these particles' large surface areas and higher number of metal-binding sites. PC1 has strong negative loadings on Si and Ca, which are typically more enriched in larger particles. Stations that experienced increases in OC and THg (A′2 (October 2005), C6B (March 2006), C6B (July 2006), K4 (July 2006), and J4 (July 2006)) had corresponding distinct shifts from low to high PC1 scores (Figure 5). The return to prehurricane OC and THg levels at A′2 (March 2006) and A′2 (July 2006) are also reflected in a return to prehurricane PC1 scores. Low PC1 scores defined all July 2005 samples, and PC1 scores at D3 remained low on all dates (Figure 5). In this sense, an elevated PC1 score may serve as an aggregate chemical “fingerprint” of disturbance-related sediment deposits. The changes in PC1 scores at G3 and E2A were mostly modest, with PC1 scores in July 2006 generally lower than sites that clearly received new surface sediment deposits, except for one depth at E2A and two at G3 (Figure 5).

Figure 5.

THg, OC, and PC1 scores. The first letter in the symbol denotes the station, and the second letter denotes sampling date (A, A′2; C, C6B; D, D3; E, E2A; G, G3; J, J4; K, K4; a, July 2005; b, October 2005; c, March 2006; d, July 2006). The dashed rectangle encloses disturbance-impacted stations (A′2 (October 2005), C6B (March 2006), C6B (July 2006), J4 (July 2006), and K4 (July 2006)).

[22] Two possible explanations for the observed changes in the surface sediment bulk chemistry (OC, THg, elemental composition) in nGOM were: (1) remobilization and redistribution of sediments that were already present in the Gulf, having being deposited in the prior months, years, or decades; and (2) inputs of new, terrestrially derived OC and Hg from the MARS in response to increased precipitation and runoff, for example, during the hurricanes. Either process, or a combination of the two, could substantially impact Hg chemistry in the nGOM by altering redox conditions in large volumes of sediment, mixing previously buried legacy Hg into surficial bioactive sediment layers, or providing new substrate (relatively labile Hg and OC) for enhanced methylation.

[23] Goñi et al. [2007] and Walsh et al. [2006] used X-radiographs and radioisotope evidence to document the extensive sediment reworking from Hurricanes Katrina and Rita in a 3600 km2 area that coincides with our study area. In cores from net-depositional zones, X-radiographs revealed layers of fine deposits atop coarse sediments after the hurricanes [Walsh et al., 2006; Goñi et al., 2007]. Radioisotope measurements in the fine sediment layers provided evidence of recent disturbance. Unsupported 210Pb levels were relatively constant over depths up to 38 cm in some cores, and 7Be (t1/2 = 55.3 d) and 234Th (t1/2 = 24.1 d) were detected to depths of 20 cm [Walsh et al., 2006], indicating that material's recent deposition. These deposits had low 7Be:234Th ratios, which Goñi et al. [2007] interpreted as being consistent with particles originating as resuspended material that scavenged 234Th from the water column, as opposed to new terrestrial inputs enriched in 7Be. Goñi et al. [2007] also found that the hurricane redistributed sediment had C isotopic compositions (δ13COC, range = −25 to −19‰) and OC:N ratios (range = 5–13) that were similar to prehurricane sediments from an earlier study in the region. In addition, they argue that a major pulse input of suspended sediments and OC from the MARS is inconsistent with the observation that the Mississippi and Atchafalaya Rivers were at their lowest discharge period of the year when the hurricanes occurred (discharge during the hurricane period was ∼33% of winter and spring levels [Walsh et al., 2006]). On the basis of the above observations, Goñi et al. [2007] and Walsh et al. [2006] concluded that most of the hurricane-related deposits came from redistribution of material that had already been residing in nGOM bottom sediments, as opposed to new inputs during the hurricanes.

[24] The posthurricane fine surface deposits observed by Goñi et al. [2007] contained high OC (1–2%) relative to the deeper layers. In this respect, those deposits are similar to the surface deposits we observed at A′2 (October 2005), C6B (March 2006), C6B (July 2006), K4 (July 2006), and J4 (July 2006). In addition, our surficial (0–10 cm) A′2 (October 2005) data for δ13COC (−22.9 ± 0.4‰) and OC:N (13.9 ± 0.9) compare favorably with those measured by Goñi et al. [2007]. The combined evidence (timing, radioisotopes, change in OC, δ13COC, OC:N) suggests that the fine, OC- and Hg-enriched surface deposits found at A′2 (October 2005) were hurricane-related and similar to those observed by Goñi et al. [2007] at multiple sites in net hurricane deposition zones. While a conclusion on the impact of the hurricanes cannot be drawn with the same degree of confidence at all stations, a link between the observed changes in sediment chemistry and the hurricanes stands as one of the more plausible explanations, considering: (1) the overlap between the Goñi et al. [2007] study area and our study area; (2) their observation of widespread hurricane sediment disturbance and deposits that commonly exceeded 10 cm; (3) in the case of J4 and K4, their proximity to Hurricane Rita's path; and (4) the similarities in sediment chemistry between the hurricane-related deposits at A′2 (October 2005) and the deposits at these other stations. In the case of C6B, where sediments with relatively high OC and relatively high Hg (and high PC1 score) accumulated between October 2005 and March 2006, these changes are consistent with the migration of hurricane deposits away from temporary, unstable accumulation sites, such as in the vicinity of A′2, to regions that may have been scoured during the hurricanes. This explanation agrees with observations elsewhere in the nGOM that gravity-driven sediment transport across the Louisiana shelf can be triggered by moderate to large waves and currents during winter storms and tropical cyclones [e.g., Wright et al., 2001; Walsh et al., 2006; Dail et al., 2007]. The sediment disturbances at J4 and K4 also cannot be definitively linked to the hurricane's passage because of the time elapsed between sampling events and the possibility of other triggers for sediment redistribution during that year. However, the substantial thickness of the deposits (>10 cm) suggests a major event with deposited sediments having chemical compositions similar to the mobile deposits at A′2 (October 2005) and C6B (March 2006) and C6B (July 2006). Thus, it is reasonable to suggest that the changes could be hurricane related, either from deposition immediately after the hurricanes or redistribution of unstable deposits during subsequent months.

[25] The actual source of the Hg- and OC-enriched sediments remains unclear. Goñi et al. [2007] identified both net erosional and net depositional zones after the hurricanes, and assembled a strong case for resuspended nGOM sediments as the primary source of posthurricane sediment deposits. However, we observed distinct differences in the chemical composition (i.e., elemental composition or PC1; OC levels; THg concentrations) of the disturbance-related surface deposits and July 2005 sediments, including specifically the C6B (July 2005) surface sediments that were eroded (Figure 5). One potential explanation is that physical sorting after resuspension, due to the slower settling velocities of fine sediments, caused a relative enrichment of OC and fine-particle-associated elements in the surface sediments. It is also possible that fine sediments were resuspended by wave action in relatively shallow bays and coastal wetlands and were flushed onto the continental shelf when the storm surge returned seaward. Turner et al. [2006] observed the landward movement of bay and wetland sediments due to the hurricanes; it stands to reason that some of these sediments also moved seaward when floodwaters receded.

4.2. Impact of Hurricanes on Regional Hg Budgets

[26] By combining our Hg-focused observations with the sediment and OC redistribution estimates from Goñi et al. [2007] and Walsh et al. [2006], it is possible to develop a regional-scale estimate of the hurricane impact on Hg redistribution and evaluate whether the hurricane-related Hg disturbance could have been an important process in the overall nGOM Hg budget. Goñi et al. [2007] estimated that 1.2 × 1012 kg of sediment and 1.4 × 1010 kg OC were redistributed during Hurricanes Katrina and Rita. Our results show a strong correlation between total Hg and OC in sediments (Figure 6; R2 = 0.72). This relationship, which is based on observations at 14 stations and utilizes data from both before and after the hurricanes, demonstrates that there is a consistent THg:OC relationship over a large area despite disturbance and spatial variations in sediment chemistry. Using this Hg:OC relationship and the Goñi et al. [2007] estimates of OC redistribution, we estimate that on the order of 50000 kg of Hg were redistributed during the hurricanes. In comparison, the Mississippi and Atchafalaya Rivers deliver 7500 kg Hg a−1 to this area of the nGOM, and direct atmospheric deposition loads an additional 2500 kg Hg a−1 [Rice et al., 2009]. Thus the hurricanes may have redistributed ∼5 times the annual input of Hg.

Figure 6.

Log transformed THg and OC, based on observations at 14 stations from the July 2005, October 2005, March 2006, and July 2006 cruises in northern Gulf of Mexico.

4.3. Assessing Impact of Disturbance on Hg Speciation

[27] We hypothesized that the 2005 hurricanes and other physical disturbances would result in enhanced net MeHg production, at least initially, as a result of the substantial sediment disturbance and associated changes in redox conditions, Hg bioavailability, and microbial activity. To explore this hypothesis we used %MeHg as an indicator variable. The change in MeHg concentration with time in a volumcuding transport can be defined as

equation image

where [HgMHg]total represent their respective total pools in the sediment (i.e., solid-complexed and dissolved) and kmethapp and kdemethapp are apparent rate constants, and take into account the speciation of Hg(II)inorg and MeHg (including solid:water partitioning) and the bioavailability of those species for methylation and demethylatoin. At steady state (d[MeHg]total/dt = 0), the expression simplifies to:

equation image

or, after substituting total Hg for Hg(II)inorg (a reasonable substitution considering that MeHg was generally <1%)

equation image

Thus, %MeHg can serve as an aggregate measure of the balance between production and degradation of MeHg [Heyes et al., 2006], and temporal or spatial variations in %MeHg indicate shifts in that balance, taking into account a range of important factors (e.g., Hg speciation, availability of those species for microbial transformation, and microbial activity) that are difficult to independently quantify [Krabbenhoft et al., 2007].

[28] Station A′2 (October 2005) was the most clearly impacted station immediately after the hurricanes. %MeHg at A′2 (October 2005) was higher than at all other stations and dates (ANOVA analyses of log transformed %MeHg comparing A′2 (October 2005) 0–10 cm with other dates: higher than A′2 (July 2005), A′2 (March 2006), and A′2 (July 2006) combined, p = 0.047; higher than all other disturbed stations C6B (March 2006), C6B (July 2006), J4 (July 2006), and K4 (July 2006) combined, p = 0.011; and higher than at all other stations and dates combined, p = 0.015). The elevated %MeHg levels at A′2 (October 2005) are consistent with the hypothesis that disturbance and the associated alteration of redox conditions, microbial activity, or inorganic Hg bioavailability resulted in increased net methylation. The reestablishment of anoxic pore water within 1.5 months indicates substantial microbial respiration. In addition, the mean AVS level at A′2 (October 2005) (mean = 2 μmol g−1; range = 1–2.5 μmol g−1) was the highest of all October 2005 stations (0.58 ± 0.81 μmol g−1). While some of the AVS may have been cotransported with fine sediments, the preservation and/or production of new AVS, which consists mostly of FeS and is rapidly oxidized in oxic conditions [Cooper and Morse, 1996; Kim et al., 2006], indicates that reducing conditions were quickly established. Finally, an inspection of logKD versus OC shows that A′2 (October 2005) samples were distinctly different from all other stations, with relatively low KD considering their OC levels (Figure 7). On the basis of known controls on Hg chemistry, a lower KD (i.e., relatively higher dissolved Hg, which is presumably more available for methylation) and increased OC is a combination that would tend to favor increased methylation [Choi and Bartha, 1994].

Figure 7.

Plots of THg, MeHg, %MeHg, and log KD versus OC for stations A′2, C6B, D3, E2A, G3, J4, and K4 across multiple sampling events. (R2, the coefficient of determination from linear regression analyses, p < 0.001.) The first letter in the symbol denotes the station, and the second letter denotes sampling date (A, A′2; C, C6B; D, D3; E, E2A; G, G3; J, J4; K, K4; a, July 2005; b, October 2005; c, March 2006; d, July 2006).

[29] Given the kinetics of demethylation, the elevated %MeHg at A′2 (October 2005) cannot be explained as an artifact of the sediment source, i.e., hurricane-deposited sediments transported from a location where %MeHg was originally higher. In order for artifact-%MeHg to be the dominant explanation, response times associated with demethylation would need to be on the order of weeks to months. This is inconsistent with demethylation potential rate constants we measured at A′2 (October 2005), and at other stations and on other dates, using sediment core incubations spiked with enriched Me199Hg, which suggest that sediment MeHg is turned over rapidly, with a characteristic time of hours (B. Liu et al., manuscript in preparation, 2009). Comparable demethylation potential rates have been measured in other systems [Kim et al., 2004; Heyes et al., 2006; Hollweg et al., 2007]. Demethylation of ambient MeHg undoubtedly occurs more slowly than demethylation of freshly spiked Me199Hg, but the ambient %MeHg signal would respond over timescale of days to weeks, as has been argued elsewhere [e.g., Heyes et al., 2006]. Indeed, repeated sampling at a location in Lavaca Bay over a seasonal cycle showed significant changes in %MeHg over periods as short as a month [Bloom et al., 1999].

[30] It is interesting to contrast the changes in %MeHg at J4 between July 2005 and July 2006 with the changes we observed between July 2005 and October 2005 at A′2. Despite the increased OC at J4, %MeHg was a factor of 2 lower in July 2006. At J4 (July 2006), methylation appears to have been limited by lower bioavailability of Hg(II), as indicated by the factor of 20 greater KD. An important difference between the situations at A′2 and J4 is that conditions at J4 (July 2006) had as long as an additional 9 months for sediment chemistry to reach a new steady state, which in addition to a higher KD may have also included lower levels of labile OC to fuel SRB activity and methylation.

[31] OC is among the most important determinants of Hg chemistry. However, there is inconsistency in the literature on the relationship between MeHg production and OC in marine systems. For example, Hammerschmidt and Fitzgerald [2004] observed negative correlations between methylation potential and organic matter levels in sediments from Long Island Sound. On the other hand, Sunderland et al. [2006] found a positive MeHg-OC correlation in the Bay of Fundy. Similarly, Lambertsson and Nilsson [2006] argued that organic matter abundance was the primary factor determining net methylation in the Öre River estuary. Thus, the relative importance of OC as a Hg ligand versus substrate for methylating microbes is system-dependent, perhaps related to OC quality. In nGOM, we found that MeHg concentration was weakly correlated with OC (R2 = 0.38, p < 0.001) across all sites and dates (Figure 7). Because OC also binds MeHg (KD,MeHg ∼ 102–103 l kg−1 [Hammerschmidt and Fitzgerald, 2006]), some of the correlation between MeHg and OC could result from MeHg binding to OC after methylation. %MeHg and OC were not significantly correlated when viewed across all stations and depths (Figure 7). However, when disturbed and undisturbed sediments are grouped separately, we find that %MeHg in disturbed sediments is positively correlated with OC (R2 = 0.26, p < 0.01), albeit weakly and driven mostly by the elevated %MeHg at A′2 (October 2005). The relationship between %MeHg and OC in the other sediments is not significant. Considering the numerous factors that influence Hg methylation, a low univariate R2 for the disturbed sediments sites should not be surprising. A difference in the relationships between %MeHg and OC for the two sediment types may reflect the different roles that OC can play in Hg cycling, perhaps due to differences in the OC quality. At A′2, OC in hurricane fine surface deposits had a different composition than in undisturbed layers (OC:N, δ13C; Figure 4); however, more research is needed to determine if the fine sediments contained, for example, a higher proportion of labile OC that led to enhanced SRB activity and Hg methylation.

5. Summary

[32] Hurricanes Katrina and Rita redistributed an enormous mass of sediment over a large area on the nGOM continental shelf [Walsh et al., 2006; Goñi et al., 2007]. Our results show that these sediment disturbances also redistributed OC- and Hg-enriched fine grain sediments and altered sediment chemistry and Hg levels. The hurricane-related Hg redistribution, which we estimate at ∼50000 kg, was substantial relative to annual Hg inputs from the Mississippi River (7500 kg a−1) and direct atmospheric deposition (2500 kg a−1). In addition, our results suggest that the redistribution impacted Hg speciation at some locations, and potentially led to higher net MeHg production, at least immediately following disturbance (i.e., A′2 (October 2005)). The longer term impact of such a large-magnitude disturbance event on net MeHg production, however, was less clear from our data, and is worthy of further study. The past record shows that nGOM is an area frequently impacted by major storms and hurricanes [Stone et al., 1997; NOAA, 2003; Dail et al., 2007], and their impacts need to be taken into consideration when evaluating Hg dynamics.

Acknowledgments

[33] This research was supported by funding from NSF Chemical Oceanography (OCE-0600624) and NOAA-OHHI (NA04OAR4600207). Additional support was provided to NNR by NOAA (NA03NOS4780038), and to PWS from the USGS Coastal and Marine Geology Program. We are grateful to the crews of the R/V Pelican and R/V Cape Hatteras for assistance with sample collection. We acknowledge two anonymous reviewers for helpful comments on an earlier version of the manuscript. We also thank a number of individuals for logistical support (W. Morrison, R. Martin, A. Sapp; LUMCON); analytical support (G. Bernier, O. Yigiterhan, UConn; N. Lupoli, HSPH), advice on sample collection and processing (C. Hammerschmidt; Wright State); and access to instrumentation (D. Brabander, K. McCarthy, XRF analyses at Wellesley College; J. Jay, pore water Hg analyses at UCLA).

Ancillary