Two tidal creeks were sampled over three full tidal cycles in mid July 2010. Despite freshwater input to the greater marsh, both creeks are tide dominated and receive negligible freshwater input (Drake et al., 2009). One creek (Sweeney) has been enriched with liquid fertilizer (w70 mM NO-3 ) that has been added directly to the creek water column since 2004 as part of the long-term nutrient addition study, TIDE (Deegan et al., 2007, Johnson et al., 2009). The fertilization targets the growing season and runs as a direct liquid addition to Sweeneyfrom mid-May to mid-September every year influencing roughly 12.4 ha of marsh area (Drake et al., 2009). The second creek (West) has served as the reference creek for this study. Sites 1 and 2 are 1.9 km apart and are 1.4 and 0.9 km from the nearest major channel, respectively. Each tidal cycle sampled experienced a different tidal range, first neap (2.2 m), followed by mid (3.1 m), and spring (3.6 m) tides (NOAA).

SIGMA and ISCO auto samplers were used to collect water samples for nutrient analysis. The instruments were secured on the marsh platform with intake tubes staked approximately 15 cm above the creek bed. The auto samplers filled two 500 mL sample bottles every hour for twelve hours over three tidal cycles (July 9,15, and 17, 2010) for a total of three sets of twelve replicate samples in each creek. Samples were temporarily stored in coolers on ice to keep them cool and dark, and were then filtered through 0.4 mm Millipore polycarbonate filters within 6 h of collection. Both the filter and the filtrate were retained for biogenic and dissolved silica analysis, respectively.

Profiling acoustic current meters (Nortek Aquadopp Pro and Nortek Aquadopp Pro HR) were deployed in the thalweg of the channel in order to calculate the flux of water through each creek over the tidal cycle. To maximize the depth of the water column that was measured, the instruments were fixed directly to the creek bed and minimal blanking distances were used such that the velocity profiles started at 28 cm and 17 cm above the bed at the fertilized and control creek, respectively. Velocity profile data were collected upward through the water column in 40 10 cm bins at the fertilized site and 33 3 cm in the control creek. Both water depth and velocity profiles were collected at 10 min intervals over the sampling period. Slight variation in placement and collection between the two meters was a result of the two instruments being different models. However, the difference in placement was less than one measuring bin of data and both regions were interpolated based on a log profile and the lowest measurement, therefore, therewas no missing data. The cross sectional area of the channel was measured close to each current meter using a standard elevation survey. Tidal discharge (m3s-1) for each 10 min interval was then calculated by applying the velocity at each depth in the profile over the appropriate channel width based on the cross section data, and integrating over the creek depth.

Dissolved silica (DSi) data were obtained colorimetrically by running the sample filtrate on a SEAL AutoAnalyzer3 using standard colorimetric techniques (Strickland and Parsons, 1968; Grasshoff et al., 1983). Dried sodium hexafluorosilicate (Na2SiF6) was used as the silicate standard (Strickland and Parsons, 1968). This standard was compared with an external standard from HACH company which was also a sodium hexafluorosilicate solution with a concentration of 50 mg L-1 SiO2. We ran the HACH external standard in 10 and 15 mM solutions against our in-house standard and obtained values of 10.3 ` 0.06 and 15.8 ` 0.22 mM, respectively.
Water column biogenic silica (BSi) values were determined using the wet alkaline method, in which the polycarbonate filter was digested in a 1% Na2CO3 solution at 85C for five hours (Conley and Schelske, 2001). Subsamples were taken after 3, 4 and 5 h, and BSi concentration was then calculated by linear extrapolation through the three sample points in order to correct for the disso- lution of mineral silicates (DeMaster, 1981). Though this method results in the dissolution of other amorphous and mineral silicates, it remains the most accurate method of measuring BSi (Liu et al., 2002; Saccone et al., 2007; Cornelis et al., 2011b). All laboratory equipment used in the processing of both the DSi and BSi samples was plastic rather than glass so as not to contaminate the samples. DSi and BSi fluxes were calculated by multiplying the concen- tration at each hour by the average of the discharge values over the same hour. (From: Vieillard et al. 2011. Estuarine Coastal and Shelf Science 95: 415-423).