Estuary Interface Pilot Study Ted Daum, Rainer Hoenicke, and Lauren Gravitz San Francisco Estuary Institute, Oakland, CA Introduction Objectives   Sampling Plan   Flows   Results and Analyses Conclusions Acknowledgments References

Introduction

After the first three years of pollutant characterization throughout the Estuary, it became evident that sampling stations at the Estuary margins generally exhibited higher concentrations of trace elements and trace organic pollutants in water and sediment than those in the deeper parts of the Bay. It was not clear which factors were primarily responsible for this phenomenon, and in order to determine what role pollutant inputs from adjacent watersheds are playing, sampling at the interface between the Bay and upland had to be conducted. Initially, one station at the upper end of the tidal prism of Coyote Creek was selected, and in 1997 the sampling was expanded to the mouth of the Guadalupe River, also known as Alviso Slough.

Objectives

The overall goals of the Estuary Interface Pilot Study (EIP) have remained the same as in 1996:

1. Link pollutant patterns found in the Estuary with those in adjacent watersheds to test if runoff and sediment taken at the lower end of Coyote Creek and the Guadalupe River differ from each other and from water and sediment in the South Bay, including the Local Effects Monitoring stations maintained by the San Jose-Santa Clara Wastewater Treatment Plant and the Sunnyvale Treatment Plant.

2. Explore what kinds of ancillary water quality parameters and watershed characteristics should be measured or described to explain some of the patterns found, improve sampling design, and fine-tune testing methodology.

Specific questions for the second year of sampling included:

1. Is the concentration gradient for certain pollutants that was observed in 1996 for Coyote Creek also applicable for the Guadalupe River?

2. Are there pronounced differences in the pollutant profiles between the two interface stations?

3. Are there pronounced differences between high- and low-flow periods between the interface stations and those in the Estuary?

4. Which factors may influence the findings?

This article describes a two-year data set which should not be interpreted as a definitive assessment of Coyote Creek or Guadalupe River watershed contributions to the Estuary. However, the data will be used in designing a new monitoring component of the RMP that is scheduled to take effect some time in 2001, and that meets the new objective of determining loading pathways of contaminants to the Estuary.

Sampling Plan

In 1997, a second sampling station was selected in the lower reach of the Guadalupe River known as Alviso Slough (BW15). The South Bay Yacht Club graciously provided access to their dock for sampling purposes, and their assistance is gratefully acknowledged. The Coyote Creek sampling station at Standish Dam (BW10) was also occupied in 1997. That station is located very close to Dixon Landing Road and Highway 880 where the city boundaries of Fremont, Milpitas, and San Jose converge (Figure 6.1). Both locations are within the tidal prisms. During the wet season, runoff amounts are large enough to dominate the pollutant signal, while during the dry season, water sampled at both stations was a brackish mix of both freshwater runoff and Bay water. Both sites were selected for their accessibility, location in the brackish transitional zone, and the fact that sediment deposition and accumulation was likely to occur. The winter of 1996/97 was very unusual in that rains produced extremely high runoff during December and January, while very little precipitation occurred during subsequent months.

The same parameters in water and sediment were measured here as in the Estuary at approximately the same times (late February/early March, late April, and early August). The sampling methodology for water was similar to that employed by the RMP. Sediment was sampled from the creek bank at low tide using a Dykon^®-coated scoop (see Appendix A: Methods). Any surface diatom layer was removed before collecting the top five centimeters of an area approximately the same size as the van Veen grab used at the Estuary stations. Each sample was then homogenized. The homogenate was divided into aliquots for analysis of trace elements, trace organics, conventional sediment parameters such as grain size, total nitrogen, and total organic carbon (TOC), and for archiving. Parameters analyzed in water included trace elements and trace organic contaminants, ammonia, chlorophylla, dissolved organic carbon, hardness, nitrates, nitrites, pH, phaeophytin, phosphate, silicates, and total suspended solids (TSS). Parameters analyzed at the bottom as well as the top of the water column included conductivity, dissolved oxygen (DO), salinity, and temperature.

Flows

Because there are no currently operating stream gauges near the mouth of Coyote Creek, its flow was calculated by combining flow data from the U.S. Gegological Survey (USGS) Stream Gauge Station 58 on Coyote Creek at Edenvale and Station 83 on Upper Penetencia Creek, a major tributary to Coyote Creek. This combined value is the best available estimate for Coyote Creek discharge into the South Bay in lieu of a stream gauge closer to the mouth of the creek. A USGS stream gauge currently operates on the Guadalupe River at San Jose approximately 11 km from the Guadalupe River Station, with no major tributaries between it and the station. Values from this stream gauge station were used for stream flow calculations. Rainfall data for both Standish Dam (BW10) and Guadalupe River (BW15) were taken from the San Jose rain gauge, which is the rainfall data source used in the Santa Clara Valley Non-Point Source Pollution Program (SCVNSS, 1991).

Stream and rain gauge locations are found in Figure 6.1. Flows peaked at an estimated 3,500 cubic feet per second (cfs) at Guadalupe River during January floods. The estimated flow on Coyote Creek during the January flood was approximately 5,300 cfs. Because of the high runoffs, Anderson Reservoir filled to capacity by January 23, began discharging over its spillway to Coyote Creek, and continued to do so throughout the remainder of the month. By February 1 flows had receded to where the reservoir was again below the spillway. Flows during the 1997 sampling were higher than those of 1996 (see Table 6.1). There are four reservoirs which empty directly or indirectly to the Guadalupe River: Calero, Almaden, Guadalupe, and Lexington. Calero Reservoir did not discharge over its spillway during the 1996/97 wet season. Almaden Reservoir spillway discharge occurred from January 1 through February 16, and from May 12 through May 30. Guadalupe Reservoir spillway discharge occurred from January 1 through January 8, and again from January 22 through January 27. Lexington Reservoir spillway discharge occurred from January 3 through January 8, from January 10 through January 31, and from April 1 through April 9. Stream and rain gauge hydrographs for Coyote/Penetencia creeks and Guadalupe River are shown in Figure 6.2.

Results and Analyses

All available data from this Pilot Study have been included in the data tables (see Appendix C: Data Tables). Total silver concentrations are not available. Total lead and dissolved silver concentrations are available for the wet season sampling period only. Dissolved lead concentrations are not available for the dry season. No values for silver concentrations in sediment for the sampling event in late summer are reported due to blank contamination. Statistical analyses were performed using SAS (SAS Institute, 1990).

With a second year of data available for the Standish Dam sampling station, as well as a year's worth of data from the Guadalupe River station, additional analyses were performed. Potential seasonal differences between the EIP stations were examined, as were comparisons between years for the Standish Dam station. The mean values of the combined EIP stations (1996-97 Standish Dam and 1997 Guadalupe River) were pooled and compared with those of the other San Francisco Bay reaches during the same time period. Bay reaches are defined in the Sediment Introduction. They are, in addition to the EIP stations, the Southern Sloughs, South Bay, Central Bay, Northern Estuary, and Rivers. Significant difference of means was determined using one-way ANOVA and Tukey-Kramer Honestly Significant Difference (HSD; p = 0.05). Stream flow and rainfall data were examined in relation to contaminant concentrations measured in the EIP. Normalizing factors for contaminant concentrations in sediment were determined and used to account for possible variations in concentrations. It should be noted that the observed sediment concentrations are heavily influenced by flow conditions prior to sampling. During low-flow periods, sediment accumulates in the flat, low-energy reaches of the two creeks and is dominated by small particles in the clay-sized fraction. With the advent of the rainy season, flow velocities increase, thereby scouring the creek beds and banks and carrying smaller-sized particles into the Bay. At the time of the wet-season sampling events in mid-winter, the sediment accumulated during the low-flow periods had likely already been mobilized (see Figure 6.2) for EIP sampling in relation to rainfall and stream flow hydrographs.

Water Metals

Figure 6.3 shows the concentrations of dissolved trace metals in water. The wet season transitional sampling period (April, Cruise 14), showed the highest concentrations of most dissolved metals in the EIP stations, with the exception of mercury from the wet-season sampling (January, Cruise 13). Selenium was consistently higher at the Guadalupe River station for all samples; zinc was higher, and nickel slightly higher, at Standish Dam in the spring sampling. Concentrations of all dissolved trace metals except selenium were higher at Standish Dam in the 1997 sampling year than in 1996. The mean of the pooled reaches was significantly higher for the EIP stations than at any of the other Estuary reaches for selenium and zinc (one-way ANOVA, p = < 0.0001).

Figure 6.4 shows the concentrations of total trace metals in water. A similar seasonal pattern was found in the total trace metal concentrations as was found in the dissolved fraction. The wet-season transitional sampling period showed the highest concentrations in both EIP stations with the exception of selenium in both stations, and chromium, copper, and nickel at the Standish Dam station. The Guadalupe River station showed consistently higher concentrations of all total trace metals in the wet season transitional sampling period. Concentrations of all total trace metals were higher at Standish Dam in the 1997 sampling year than in 1996. The pooled mean was significantly higher for the EIP stations for selenium (one-way ANOVA, p = < 0.0001) than at any of the other Estuary reaches. The pooled mean values at the EIP stations for arsenic, mercury, nickel, and zinc were not significantly different from those for the Southern Sloughs, but were significantly higher than those of the other Estuary reaches (one-way ANOVA, p = < 0.0001).

Water Organics

Figure 6.5 shows the concentrations of dissolved trace organics in water. Dissolved PAHs and chlorpyrifos were higher in the spring samples at both EIP stations, and for chlordanes the Guadalupe River station concentrations were higher. DDT and dieldrin were higher at the Standish Dam station in the summer. Concentrations of dissolved PCBs, diazinon, and chlordanes were higher in the 1996 sampling year, while dieldrin, chlorpyrifos, and PAHs were higher in the 1997 sampling year at the EIP stations. The mean values of the pooled reaches were significantly higher for the EIP stations than the Estuary reaches for DDTs and chlordanes (one-way ANOVA, p = < 0.0001). The pooled mean value at the EIP stations for dieldrin was not statistically different from that of the River reach, but was significantly higher than those of the other Estuary reaches (one-way ANOVA, p = < 0.0002). The pooled mean values at the EIP stations for chlorpyrifos and PAHs were not significantly different from those for the Southern Sloughs, but were significantly higher than those of the other Estuary reaches (one-way ANOVA, p = < 0.0001).

Figure 6.6 shows the concentrations of total trace organics in water. DDTs, chlordanes, and dieldrin dominated in one or both of the EIP stations. Concentrations of dieldrin and PAHs were higher in the 1996 sampling year at the EIP stations. The mean values of the pooled reaches were significantly higher for the EIP stations than the Estuary reaches for DDTs and chlordanes (one-way ANOVA, p = < 0.0001). The pooled mean value for PAHs was not significantly different between the EIP stations and the Southern Sloughs, but was significantly higher than mean values at the other Estuary reaches (one-way ANOVA, p = < 0.0001).

Sediment Metals

All raw trace metal concentrations at the EIP stations, except selenium, were higher in the 1997 sampling year compared to 1996. The mean raw concentrations of trace metals at the EIP stations were not significantly different than those of the other reaches, with the exception of cadmium. The mean raw concentrations of cadmium at EIP and Southern Slough stations were not significantly different from each other but as a group these were significantly higher than concentrations at the other reaches (one-way ANOVA, p = < 0.0001). The EIP, Southern Sloughs, and South Bay stations were not significantly different from each other, but as a group exhibited significantly higher metals concentrations than the other Bay reaches (one-way ANOVA, p = < 0.0002).

Sediment Contaminant Normalization

A common practice used to improve the sensitivity of comparing trace element and organic contaminant concentrations in sediments is to normalize them to some sediment constituent which is unaffected by human activities such as contaminant input (Luoma, 1990; Hanson, 1993; Daskalakis and O'Connor, 1995). Some of the constituents commonly used include aluminum, iron, TOC, and grain size (Daskalakis and O'Connor, 1995). Statistically speaking, these are independent variables, i.e. their concentrations are independent of the variable being examined, the dependent variable. The dependent variable in this case is defined as the organic or trace element contaminant whose concentration is dependent on the concentrations of the independent variable(s) found in the sediment.

Since all four of the independent variables mentioned above were analyzed in this pilot study, the first step in the normalization process was to determine how they were correlated with each other. If there was a high correlation between the four independent variables, it would then be possible to reduce the group to a single representative normalizing analyte. This was indeed the case. A Pearson product-moment pair-wise correlation was used to determine if there was a significant linear relationship between the variables. Aluminum was significantly correlated with iron (r = 0.89; p = 0.0001), with clay (r = 0.74; p=0.0001; a surrogate for grain size), and with TOC (r = 0.66; p = 0.0001). It was, therefore, chosen as the normalizing variable.

The next step was to calculate correlation coefficients for aluminum and the trace elements and organic contaminants. This was done using the Pearson product-moment pair-wise correlation. Chromium, copper, nickel, lead, selenium, and zinc had a correlation value r of at least 0.60 (p = 0.0001). Mercury was more closely correlated with TOC than with aluminum (r = 0.57; p = 0.0001). Silver, cadmium, and total chlordanes, PAHs, PCBs, and DDTs had lower correlations (r < 0.50) with both aluminum and TOC. Interestingly, TOC was not significantly correlated with any of the organic contaminants in the data set, which is contrary to what would be expected. All available 1997 RMP observations (n = 50 102, depending on the analyte) were used in the above correlation calculations. The r values are found in Table 6.2.

Now it is possible to calculate an adjusted (i.e., normalized) value which takes into account these independent variables. This value is commonly expressed as a ratio of the contaminant concentration for which there was a significant correlation divided by the concentration of Al (or in the case of Hg with TOC) at each site. Figure 6.7 shows normalized values for the 1997 data compared with the corresponding raw values.

Figure 6.8 shows the concentrations of all available trace metals in sediment. Raw value concentrations at the Standish Dam station were higher in the summer than in the winter for all trace metals. However, when normalized for aluminum, all trace metals at this station except chromium were higher in the winter in relation to summer. Looking at raw value concentrations, there were no predominant seasonal differences at the Guadalupe River station except for mercury, which was higher in the winter. However, when normalized for aluminum, trace metals were higher in the winter sampling period for chromium, copper, nickel, lead, selenium, and zinc. These results are not surprising, since high flows prior to wet-season sampling had removed the finer particle sizes that generally contain greater contaminant mass per unit weight than larger sediment particles. Therefore, the relative concentrations of these contaminants, after normalizing for aluminum, were greater in the winter sampling period with its greater hydrographic activity.

Sediment Organics

Figure 6.9 shows the concentrations of trace organics in sediment. Concentrations of PCBs, DDTs, and chlordanes were higher in wet-season samples compared to those collected during the dry season, but total PAHs were higher in the dry-season samples. Because dieldrin was below the detection limit at most of the stations, it was not included in these analyses. As in 1996, the 1997 sampling year showed that EIP stations had the highest concentrations of DDTs and chlordanes. There was no significant difference in concentrations of PAHs or PCBs at Standish Dam in the 1996 sampling season compared with 1997. Chlordanes were higher, and DDTs slightly higher in 1997. There was no difference in the mean concentration of PAHs at the Estuary interface stations compared with the other Bay reaches. The mean concentrations of total chlordanes was significantly higher at the EIP stations than the other Bay reaches (one-way ANOVA, p = < 0.0001). The mean concentrations of total DDTs at the EIP stations were significantly higher than concentrations at the other reaches (one-way ANOVA, p=<0.0001).

PCB Fingerprinting

Analysis of the congener spectrum of PCBs to discern source, fate, and transport patterns has been undertaken in a number of studies (van Bavel, 1997; Johnson et. al., 1998). A congener spectrum of this sort is often called a "fingerprint". PCB fingerprints were generated from samples collected at the EIP stations and representative stations in all reaches of the Bay, for the dissolved and particulate fractions of water and for sediments, for all cruises.

Similar patterns of higher molecular weight congeners dominated in the EIP and South Bay in both the water fractions and the sediments. These patterns were distinctly different from those measured in the rest of the Bay, which consisted of higher percentages of the lower weight congeners, and lower overall concentrations. A concentration gradient can even be seen between the EIP stations, which are higher, and San Jose (C-3-0), the representative South Bay station. This suggests a possible ongoing source load near the EIP stations, and a mixing of the PCB congener signal away from the watersheds. An example of a PCB fingerprint is shown in Figure 6.10.

The concentration gradients found in the water fractions between the EIP stations and South Bay representative station San Jose (C-3-0) were not seen in the sediment samples. On the contrary, the concentrations of PCB congeners at San Jose (C-3-0) were at least as high if not higher, which suggests that this area could be a PCB sink for sediments transported away from the EIP stations. There was, however, a discernible sediment concentration gradient between Coyote Creek (BA10) and the rest of the South Bay stations. Furthermore, the South Bay as a whole exhibits higher PCB concentrations in sediment than do the other Estuary reaches and the EIP stations. Similar gradient patterns are seen in other localized watershed sampling efforts in the Bay. Preliminary data from the San Leandro Bay Project (see Daum and Thompson, 1998) strongly suggest localized inputs of PCBs, as well as PAHs, and some trace metals.

Water Particulates and Sediments

In order to assess whether the Coyote Creek or Guadalupe River watersheds contribute significant sources of trace metal or organic contamination, the concentrations of these contaminants which are on the particulate fraction of the water coming into the Bay must be determined. This was done for three trace metal contaminants of concern: copper, mercury, and nickel. Particulate concentrations for each metal were calculated by subtracting the filtered (dissolved) concentration from the unfiltered (total), with the difference being the particulate fraction. Dividing this value by the total suspended solids (TSS) concentrations gave the normalized water particulate concentration. Each one of these measurements has an associated uncertainty, which can be calculated. The water particulate values from the EIP stations were then compared with the sediment and water particulate values in the Southern Sloughs and South Bay stations to determine if the EIP station concentrations were higher compared to those found in the other stations. Figure 6.11 shows these comparisons for the wet- and dry-season sampling events.

Sampling during the wet season showed roughly the same concentrations of water particulates and sediments for the EIP, Southern Sloughs, and South Bay stations for copper and nickel. The Guadalupe River station showed elevated levels of mercury. Results for the dry weather sampling were striking. Water particulate concentrations were much higher in all three metals and at both Standish Dam and Guadalupe River stations.

Conclusions

The second year of the Estuary Interface Pilot Study showed some definite patterns emerging. Although only two years of data have been analyzed, some conclusions can be made. There appear to be similarities in the concentration gradients in both of the EIP stations for many of the contaminants in both the water and sediment fractions. The particulate fraction of water entering the Estuary from the Guadalupe River in the dry season have concentrations of copper, mercury, and nickel which are greatly elevated compared to the respective sediment concentrations of these metals in the Southern Sloughs and South Bay. Hornberger et al. (1998) found background levels (i.e., not enriched by human activity) in Grizzly Bay and San Pablo Bay for mercury to be 0.06 ppb +/- 0.01; nickel 82­110 ppb; and copper 23­41 ppb. If the natural background levels in the Coyote Creek and Guadalupe River watersheds are comparable to Grizzly and San Pablo bays, then the incoming particulate metals' concentrations are indeed enriched. More study is needed to determine if this is the case.

It is also probable that copper, lead, and nickel are enriched over background levels in creek sediment, after normalizing for aluminum. The elevated mercury concentrations are probably due to the New Almaden Mine which is located in the Guadalupe River watershed. Results from the Santa Clara Valley Urban Runoff Pollution Prevention Program have indicated that suspended stream sediments are enriched compared to suspended sediments in the South Bay for copper, lead, and nickel among others, which might be contributors to the elevated sediment levels noted here. And formerly widespread use of pesticides, including chlordanes, occurred in both watersheds prior to use restrictions and have been found in urban runoff (BASMAA, 1996; SFEI, 1998).

Specific events such as tidal or storm-influenced shifting water masses, with the resulting pulses of TSS loading, can skew the calculated particulate concentrations for metals and organics. This may explain some of the observed concentration peaks as being artifacts of these events, which are especially acute at the EIP stations. The RMP Base Program sampling effort is not frequent enough to incorporate these specific conditions, nor is it meant to be. Event-driven sampling, perhaps examining a limited suite of contaminants, could be undertaken to incorporate these conditions. Conversely, avoiding these conditions in the Base Program sampling may enable the collection of more truly representative ambient or background data.

Acknowledgments

The matching financial contributions from the City of San Jose and the Santa Clara Valley Urban Runoff Pollution Prevention Program are gratefully acknowledged. Interpretive comments from Jay Davis regarding PCBs and Khalil Abu-Saba regarding water particulate trace metals are gratefully acknowledged.

References

BASMAA. 1996. Monitoring Data Analysis. Prepared by Woodward-Clyde Consultants, Oakland, CA.

Daskalakis, K.D. and T.P. O'Connor. 1995. Normalization and elemental sediment contamination in the coastal United States. Environ. Sci. and Technol. 29:470­477.

Daum, T. and B. Thompson. 1998. Sediment investigation in San Leandro Bay. San Francisco Estuary Institute. Oakland, CA.

Hanson, P.J., D.W. Evans, and D.R. Colby. 1993. Assessment of elemental contamination in estuarine and coastal environments based on geochemical and statistical modeling of sediments. Marine Environ. Res. 36:237­266.

Hornberger, M.I., S.N. Luoma, A. van Geen, C. Fuller, and R. Anima. 1998. Historical trends of trace metals in the sediments of San Francisco Bay, California. Marine Chemistry.

Johnson, G.W., W.M. Jarman, C.E. Bacon, J.A. Davis, R. Ehrlich, and R.W. Risebrough. 1998. Unmixing polychlorinated biphenyl source fingerprints in surface waters of San Francisco Bay. Manuscript submitted to Environ. Sci. and Technol.

Luoma, S.N. 1990. Processes affecting metal concentrations in estuarine and coastal marine sediments. In Heavy metals in the marine environment. R.W. Furness and P.S. Rainbow, (eds.). CRC Press, Inc., Boca Raton, FL.

SAS Institute. 1990. SAS/STAT User's Guide, Version 6, Fourth Edition. SAS Institute, Cary, NC.

SCVNSS, 1991. Volume 1: loads assessment report. Santa Clara Valley Nonpoint Source Study. Woodward-Clyde Consultants, Oakland, CA.

SFEI, 1998. RMP watershed pilot study: an information review with emphasis on contaminant loading, sources, and effects. San Francisco Estuary Institute, Oakland, CA.

van Bavel, B. 1997. Multivariate data evaluation in environmental chemistry. SETAC 18^th Annual Meeting, 1997, San Francisco, CA.