Many processes occurring over different time scales affected the Hg dynamics over the course of the study. For example, event-related, tide-related, and low-level random concentration variability were all apparent in the time series (Figs. 3 and 4). The greatest excursions of modeled concentration were associated with spikes of high turbidity that coincided with strong northerly wind events (fall and winter), primarily affecting the main-channel site (Fig. 3). During the strongest events, which occurred on October 16 and January 22, modeled UTHg concentrations jumped ~30 ng L−1 over baseline values. Elevated northerly winds have been previously documented as important agents of sediment resuspension and transport in the Browns Island area (Ganju et al. 2005). A significant increase in modeled on-island (positive) UTHg flux accompanied these strong fall and winter sediment resuspension events (Fig. 5).
The timing of the strong winds relative to tidal stage appeared to affect the magnitude of these event-driven fluxes. For instance, the high northerly winds of October 16 coincided with a rising tide, producing a 0.1-g increase of the on-island flux (Fig. 5). In contrast, the January 22 wind event of a similar magnitude coincided with a falling tide, resulting in a smaller on-island flux increase (approximately 0.05 g).
Outside these episodic wind-driven sediment resuspension events, the majority of the modeled UTHg concentration variation was associated with the systematic rise and fall of the tides (Figs. 3 and 4). Both the main- and side-channel sites exhibited tide-associated peaks in modeled UTHg concentration, with maxima tending to occur near low water (Figs. 3 and 4). Peak magnitudes varied according to location, season, and timing within the spring–neap cycle. For example, peak magnitudes were greater in the main channel than in the side channel (Figs. 3 and 4), with the greatest systematic concentration variation seen during the spring deployment period in the main channel. This may be because higher water velocities of the main channel are more effective at entraining mercury-laden material from the channel bottom. Neap periods of the lunar cycle generally corresponded to lower water velocities and diminished modeled UTHg concentration variability.
The timing of the winds relative to the tidal cycle influenced not only major event-driven fluxes but also smaller tidally associated fluxes. During the fall deployment, for example, the wind peaked rather consistently in the afternoon, coincident with a rising tide. The co-occurrence of these processes—sediment resuspension by wind and on-island tidal flow—resulted in a rather consistent flux of UTHg onto the island (Figs. 3, 4, and 5). During the spring deployment period, the afternoon winds generally corresponded to a falling tide, resulting in a consistent flux of mercury off the island. During the winter deployment, which lacked this consistent daily alignment of wind and tide, comparatively little net flux occurred outside the major wind resuspension event of January 22.
The source of mercury in the water column varied through time, depending on hydrologic and meteorological forcing and the effects of local geomorphic features. The main channel mouth exchanges water with the open estuary across a broad, shallow tidal bar, the side channel exchanges water directly with a deep (>15 m) river channel (Fig. 1). The similarity in contemporaneous baseline concentrations between the two channels (Figs. 3 and 4) suggests that both channels are, under baseline conditions, responding to a common source of influent water and UTHg, rather than more localized, channel-specific sources. The likely source of the baseline mercury is the sediment-laden waters of the Sacramento and San Joaquin rivers, which flow into the head of the estuary east of Browns Island. The higher spring and winter baseline UTHg concentrations are attributable to elevated turbidity associated with higher winter flows in both rivers (California Department of Water Resources 2010).
During strong wind events, local UTHg sources may predominate. The strong northerly winds of October 2005 and January 2006, for example, caused large spikes in measured turbidity and therefore modeled UTHg concentration—but only in the north-facing main channel, not the south-facing side channel. The bar outside the main channel seems to have served as a local source of resuspended sediment and, presumably, accompanying Hg.
In contrast, the source of the filter-passing Hg fraction seems to be the wetlands themselves. During the spring deployment, elevated FTHg concentration values were observed throughout the ebb of the sampled tidal cycle, coincident with periods of elevated FDOM (i.e., elevated DOC). Because DOC can solubilize Hg (Amirbahman et al. 2002), it is likely that the high-DOC waters draining from the marsh had facilitated the repartitioning of sedimentary-phase THg into the colloidal or dissolved phases, thus accounting for the elevated concentrations of filter-passing mercury. FTHg was also correlated with FMeHg (r = 0.79; Table 1), consistent with the idea that solubilization by DOC promotes transport and methylation.
Peaks in observed FDOM (i.e., DOC) values also accompanied the strongest of the small, systematic off-wetland UTHg fluxes that occurred during large spring tides (Fig. 5). This pattern is consistent with the notion that the DOC-rich waters draining the marsh may have promoted solubilization and mobilization of previously deposited particulate Hg. Resuspension of finer sedimentary particles or particles high in organic matter content, both of which have been associated with higher THg content in the SFE (Conaway et al. 2003), could also have contributed to the spring-tide off-marsh exports.
The implication of these collective observations is that on-island UTHg flux at Browns Island seems to be largely particle-associated and event-driven—dependent on the magnitude of the winds, their timing with respect to the tides, and the orientation of channel inlets relative to wind direction. Net off-island UTHg flux seems to be associated primarily with the export of filter-passing mercury and is strongest during periods of large spring tides. High current velocities also encourage Hg transport—the systematic concentration fluctuations that accompanied the tides (Fig. 3) were of greatest magnitude during times of peak water velocity, perhaps because of entrained fine, organic-rich sediment. The positive–negative net flux imbalance is slight, so even relatively minor forcing changes—such as the orientation of tidal channels or whether afternoon breezes align with a rising or falling tide—may shift the wetland from sink to source or vice versa.
Wetlands, particularly estuarine wetlands, are most often thought to be sedimentary sinks and thus presumptive sinks for total Hg (Ganju et al. 2005; Selvendiran et al. 2008; Shanley et al. 2008). Net on-wetland flux of sediment has been previously documented at Browns Island (Ganju et al. 2005), and we therefore expected to observe net on-wetland fluxes of Hg as well. That we found no appreciable net on-wetland flux of Hg is contrary to our hypothesis that an on-island flux of sediment would have a proportional flux of associated Hg which is retained by the wetland. It appears instead that for Browns Island, and perhaps other estuarine wetlands as well, episodic deposition-related acquisitions of particle-associated mercury were offset by persistent tidally driven exports of filter-passing mercury. The large concentration excursions (spikes) driven by major wind events are much larger than the smaller systematic concentration variations associated with astronomical tides and tidal currents (Figs. 3 and 4). But at Browns Island, over the timescale of a spring–neap cycle, the fluxes driven by these opposing forces—large, pulsed on-island fluxes and smaller but more persistent off-island fluxes—are roughly equivalent.
A better understanding of the numerous physical processes that influence mercury fluxes helps inform and guide wetland restoration efforts in SFE and elsewhere. Our observations at Browns Island, for example, indicate that tidal exchange volume, water velocities, channel orientation, and proximity to a sediment source are all important determinants of Hg flux. Also, as pointed out previously (Bergamaschi et al. 2011), a sill that limits wetland drainage can restrict export of wetland-derived material. Although Browns Island, an established tidal wetland, did not appear to be a significant sink of Hg for the period of the study, it is possible that newly restored wetlands would not be in sediment equilibrium and would thus act as sediment (and presumably Hg) sinks until equilibrium is achieved.
The results of this study also confirm the importance of using continuous, longer-term measurements to help elucidate mercury dynamics. Measurements over a single tidal cycle or a small number of cycles can significantly bias flux estimates. For example, if Browns Island cumulative fluxes for the winter deployment are calculated by simple extrapolation from the single tide over which discrete samples were collected (Figs. 3 and 4), the apparent cumulative 3-week flux would be >0.5 g UTHg off the island; the estimated annual flux would be >15 g off-island. In contrast, calculations based on our continuous measurements over the entire spring–neap cycle yield a 3-week flux estimate near zero (Fig. 5), suggesting that annual fluxes are minimal.
Still, there is no assurance that the deployments in this study adequately represent the full annual cycle either. To better elucidate fluxes of mercury in tidal systems, continuous proxy measurements should be undertaken over as long a time period as possible. Continuous, high-resolution measurements will help capture episodic, short-term events or processes that might otherwise be missed by lower-resolution sampling. Longer-term deployments will help improve estimates of fluxes and detect trends or other effects of biogeochemical processes that may be evident only over longer periods.
One key result of this study is a demonstration that estimating mercury fluxes in tidal estuaries or other dynamic environments is a practical undertaking. The parameters used to develop the concentration model used in this study—turbidity and FDOM—are relatively easily and commonly measured. Importantly, they are also mechanistically related to the underlying drivers of Hg fluxes in this tidal wetland (Domagalski 2001; Ravichandran 2004). Other, more complex systems may require measurement of additional parameters to adequately model UTHg concentrations, but for many settings, it seems likely that a relatively small number of direct measurements may be extended over space or time by using proxy measurements. When doing so, it is axiomatic that the proxy measurements be mechanistically related to Hg concentration through the processes under study rather than simply empirical. Further, calibrations required to develop proxy:concentration relationships are likely site-specific, and perhaps even season- or storm-specific; the underlying relationships among the directly measured and proxy parameters must be established specifically for the particular circumstances and goals of any given study, and verified over the life of the study.
The utility of continuous measurements of this type can be expanded considerably through the simultaneous deployment of relatively inexpensive complementary sensors. For example, particle concentration and size distribution can be estimated from acoustic backscatter and attenuation data obtained from the same equipment used to measure water velocities (Gartner 2004). Also, continuous multiwavelength fluorescence, light absorbance, and optical attenuation can also be measured. Together, these measurements would provide a more complete picture of particle and dissolved constituent properties (Spencer et al. 2007; Pellerin et al. 2009; Bergamaschi et al. 2011), and thus better links to processes governing mercury cycling. These relatively inexpensive, multi-pronged approaches provide a previously unimaginable ability to relate biogeochemical processes to physical forcings, sedimentological events, hydrologic flow paths, and geomorphological features, thus aiding the development of more accurate predictive models to help guide environmental management, mercury mitigation, and wetland restoration efforts.