Scientific Investigations Report 2008–5168
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2008–5168
Selected limnological variables common to the 1991–92 and 2004–06 studies were compared to determine changes associated with the interaction of physical, chemical, and biological processes in Coeur d’Alene Lake. The report on the 1991–92 study (Woods and Beckwith, 1997) presents detailed discussions of historical effects on the water quality in Coeur d’Alene Lake and the results of an applied empirical nutrient load/lake response model (Walker, 1987) to evaluate the lake response to nutrient inputs (Woods and Beckwith, 1997).
Where statistical comparisons are made among sampling stations and (or) study periods, various non-parametric statistical techniques were used in the NCSS statistical software package (Hintze, 2006). The techniques used for various data types are:
In nearly all cases, data from calendar years 1991 and 1992 were grouped into one study period referred to as “1991–92”. Likewise, data from water years 2004, 2005, and 2006 were grouped into one study period referred to as “2004–06”. The data were grouped and compared as study periods rather than individual years to compare overall differences in lake water quality between the two studies.
In all cases, single- and double-tailed hypothesis tests were employed at a significance level of α=0.05. A double-tailed hypothesis test was first used to determine if a significant difference was measured between median values in the sampled populations. If a significant difference was measured, the single-tailed hypothesis test was used to determine which sampled population had a greater median value. Where used in this report, the term “significant” denotes that the comparison was statistically significant at a significance level of α=0.05.
The fate and transport of water and associated water-quality constituents following their delivery into a lake are determined by the interactions of a myriad of physical, chemical, and biological processes operating in the lake over a wide range of spatial and temporal scales. To evaluate the net effects of those processes in a lake, hydrologic and constituent budgets can be developed by quantifying water and constituent mass entering and exiting the lake. The quantified mass inflow or outflow commonly is referred to as a load, which is defined as the quantity of a constituent passing a riverine cross section per unit of time. Load is calculated as the product of constituent concentration and discharge.
Although the calculation of load is simple mathematically, the method of deriving concentration and discharge can produce substantial differences in load calculations, especially for periods longer than 1 day. Constituent concentration and discharge vary substantially during a year and are nonsynchronous in their temporal variation. Concentrations of dissolved and particulate constituents respond differently to discharge changes. Therefore, constituent loads may vary on a daily, monthly, and (or) annual basis. Various basic to complex methods have been used to process multiple-date data sets of concentration and discharge into estimates of seasonal and annual loads. One basic method is to multiply the mean values of concentration and discharge over the period of interest to derive the load. Additional temporal resolution can be gained by linear interpolation of measured values of concentration and discharge to examine daily load variability over the period of interest. More complex methods use linear regression, either simple or multiple, to relate load or concentration to discharge and other explanatory variables such as seasonality and time trend.
Constituent loads typically are quite variable within and among years because discharge changes often are the dominant influence on loads. This is especially true for sediment-associated constituents where the sediment supply is not limited. In this situation, the smallest loads of sediment-associated constituents measured during a year typically are associated with minimum discharges due to decreased water velocities that produce less erosion and transport. Alternatively, increased water velocities during maximum discharge during a year can result in erosion and transport and large loads of sediment-associated constituents. The pattern generally is similar for dissolved constituents, but differences in loads between high and low discharge are decreased because the transport of dissolved constituents is less dependent on water velocity. These within-year patterns of load variability for sediment-associated and dissolved constituents in a lake are important considerations for determining the effects on the fate and transport of nutrient loads. The transport of seasonal loads also applies to annual loads; large constituent loads generally are transported during high discharge years than during low discharge years. ">Constituent loads typically are quite variable within and among years because discharge changes often are the dominant influence on loads. This is especially true for sediment-associated constituents where the sediment supply is not limited. In this situation, the smallest loads of sediment-associated constituents measured during a year typically are associated with minimum discharges due to decreased water velocities that produce less erosion and transport. Alternatively, increased water velocities during maximum discharge during a year can result in erosion and transport and large loads of sediment-associated constituents. The pattern generally is similar for dissolved constituents, but differences in loads between high and low discharge are decreased because the transport of dissolved constituents is less dependent on water velocity. These within-year patterns of load variability for sediment-associated and dissolved constituents in a lake are important considerations for determining the effects on the fate and transport of nutrient loads. The transport of seasonal loads also applies to annual loads; large constituent loads generally are transported during high discharge years than during low discharge years.
Table 4 presents streamflow and nutrient loads from the Coeur d’Alene and St. Joe River inflows and the Spokane River outflow from Coeur d’Alene Lake determined using LOADEST, as well as nutrient load estimates from ungaged sources. Robust comparisons among years and load sources are limited by the difference in study periods (calendar year and water year) and the difference in analytical reporting limits between the two studies.
Streamflow for the 5 years varied substantially in response to differences in annual runoff from the watershed and operation of Post Falls Dam. The annual hydrologic conditions for those 5 years in relation to mean hydrologic conditions were compared to the long-term annual streamflow of the USGS gaging station on the Spokane River near Post Falls (USGS gaging station 12419000). The mean annual streamflow at Spokane River near Post Falls during 1912–2006 was 175 m3/s (U.S. Geological Survey, 2007). In comparison, the annual mean streamflows for calendar years 1991 and 1992 and water years 2004, 2005, and 2006 were, 199, 99, 131, 134, and 186 m3/s, respectively. Therefore, the percentages of long-term annual mean streamflow among the 5 years were 113, 56, 74, 76, and 106, respectively. Because streamflow is a key factor in the calculation of constituent load, the different hydrologic conditions among the 5 years were responsible for most differences in annual loads delivered into and discharged out of Coeur d’Alene Lake.
The residual for each nutrient load (nitrogen or phosphorus) was computed as the sum of total outflow loads minus the sum of total inflows and is presented in table 4. The residual also is presented as a percentage of the total inflow load. Although the residual can be considered roughly the amount of nutrient load retained in the lake, the estimate contains errors associated with measured and unmeasured budget components. Unquantifiable uncertainties exist in these residual estimates because nutrients in Coeur d’Alene Lake are affected to an unknown degree by physical, chemical, and biological processes.
Total nitrogen inflow and outflow loads decreased between the 1991–92 and 2004–06 study periods. The St. Joe River contributed the largest percentage of inflow total nitrogen load in 1991 (47 percent), and the Coeur d’Alene River contributed slightly higher total nitrogen loads than the St. Joe River in other years studied. Combined, the St. Joe and Coeur d’Alene Rivers contributed an average of 82 percent of the total inflow nitrogen load in the 1991–92 study, but an average of only 40 percent of the total inflow nitrogen load in the 2004–06 study. Estimated inputs from ungaged tributaries and other surface-water inflows constitute a larger percentage of the inflow total nitrogen load in the 2004–06 study (36–46 percent) compared with the 1991–92 study (11–13 percent). Reductions in total nitrogen in the St. Joe and Coeur d’Alene Rivers were expected due to reductions in total nitrogen concentrations and loads in municipal wastewater effluent. For example, effluent monitoring reports from the Coeur d’Alene municipal wastewater treatment plant, which is typical of other wastewater treatment plants on the St. Joe and Coeur d’Alene Rivers, show an increase in total nitrogen removal efficiency from 16 to 56 percent between the 1991–92 and 2004–06 study periods due to improvements in treatment processes. Nutrients in ungaged tributaries and other surface-water inflows were expected to be from untreated wastewater, agriculture, and natural sources. Nutrient concentrations in these inflows were assumed to be relatively constant between the 1991–92 and 2004–06 studies and, therefore, constituted a larger percentage of the overall inflow total nitrogen load in the latter study. Additional measurements to quantify loads from ungaged tributaries, as well as other ungaged inflows and outflows including nearshore wastewater, precipitation, ground water, and lake storage changes, would improve load estimates and verify assumptions. Load estimates presented in table 4 for ground-water outflow and lake storage change in the 1991–92 study differ slightly from estimates published in Woods and Beckwith (1997) due to a small error in the calculations used to generate those estimates. The Spokane River is the dominant outflow route of total nitrogen from Coeur d’Alene Lake (85-95 percent of total outflow).
Total nitrogen was retained in the lake in all years studied except water year 2006, when the residual was positive, indicating a higher outflow than inflow of total nitrogen load. However, the residual is less than the overall error in the total nitrogen budget for 2006, and inflow and outflow loads of total nitrogen most likely were equal. Residuals for all other years were larger than the overall corresponding errors and ranged from 17 to 47 percent of the total inflow load.
Total phosphorus inflow and outflow loads fluctuated with streamflow in both study periods, but loads relative to streamflow were higher in the 2004–06 study than the 1991–92 study because total phosphorus concentrations in gaged inflows were statistically higher in the latter study. The St. Joe River contributed the largest percentage of inflow total phosphorus load (38–50 percent) in all years studied. The Coeur d’Alene River was the second largest source, contributing 28–39 percent of the inflow total phosphorus load. Combined, the percent contributions to inflow load from the St. Joe and Coeur d’Alene Rivers were relatively consistent across both study periods (72 percent in the 1991–92 study and 84 percent in the 2004–06 study). Ungaged tributaries and other surface-water inflow comprised the third largest source, contributing 8–13 percent of the inflow total phosphorus load. As with total nitrogen, total phosphorus loads in outflow of the Spokane River (94–98 percent) dominated other outflow routes.
Total phosphorus was retained in the lake in all years studied, and the residuals, expressed as a percentage of inflow, were slightly higher in the 2004–06 study than the 1991–92 study. Residuals for all years were larger than the overall corresponding errors and ranged from 36 to 67 percent of total inflow load. Total phosphorus was retained in the lake in all years studied, and the residuals, expressed as a percentage of inflow, were slightly higher in the 2004–06 study than the 1991–92 study. Residuals for all years were larger than the overall corresponding errors and ranged from 36 to 67 percent of total inflow load.
Table 5 presents selected trace-metal loads from the Coeur d’Alene and St. Joe River inflows and the Spokane River outflow determined using LOADEST. Robust comparisons among years and load sources were hampered by differences in study periods (calendar year and water year), in analytical reporting limits between the two studies, and in sampling locations on the St. Joe and Spokane Rivers, which were not normalized for the trace-metal load budget. Total cadmium loads could not be calculated for the St. Joe River using LOADEST because total cadmium concentrations frequently were less than the method reporting limit.
The largest loads of the three constituents presented (total cadmium, total lead, and total zinc) occurred in 1991, the year of the highest streamflows (table 4). Samples collected in the 2004–06 study at gaging stations on the St. Joe and Spokane Rivers indicated reductions in loads because the trace metal concentrations were within the range of differences in analytical reporting limits used during the two studies. The reporting limit was lowered and more uncensored data points were available for the LOADEST calculations in the 2004–06 study than in the 1991–92 study. Trace metal concentrations for the Coeur d’Alene River exceeded analytical reporting limits used in both studies; therefore, loads at that station were least affected by those differences. The Coeur d’Alene River was, by far, the largest source of trace metal loads to the lake. Combining both study periods, the Coeur d’Alene River contributed an average of 100 percent of total cadmium and 97 percent of total lead and total zinc gaged inflow loads.
An important comparison can be made based on the relative magnitude of constituent loads delivered into the lake as opposed to those discharged out of the lake. The sum of loads from the Coeur d’Alene and St. Joe Rivers was compared to the load at the Spokane River (lake outlet), for total cadmium, total lead, and total zinc (table 5). The retention of the three constituents in the lake, expressed as the residual as a percentage of inflow, was slightly higher in the 2004–06 study period than in the 1991–92 study period. However, not all inflows to the lake could be quantified, and reported residuals may be lower than actual values. The percentage of total cadmium retained ranged from 36 to 70. A larger percentage of total lead was retained by the lake, ranging from 83 to 96. These high residuals reflect the propensity of lead to be associated with sediment particles and to be prone to sedimentation in the lake. Zinc is the most soluble of the three trace metals, and the percentage retention in the lake ranged from 33 to 52. Relative to loads, higher errors were present in lead and zinc load estimates for the St. Joe River compared to the Coeur d’Alene and Spokane Rivers.
The physical limnological processes of thermal stratification and convective circulation are strongly correlated with the vertical distribution of water-quality properties and constituents. Thermal structure in some lakes may be established, in part, by riverine inflows. However, the major source of heat for most lakes is the short- and long-wave radiation that impinges on the surface of a lake (Wetzel, 1983). Wind energy distributes the surface heat into the water column until density differences impede deeper mixing. In lakes deep enough to resist full-depth convective circulation, solar heating and wind mixing during the summer vertically segregate the water column into three zones: epilimnion, metalimnion, and hypolimnion. The upper zone, the epilimnion, is the stratum in which most biological production occurs because light generally is sufficient to drive photosynthetic production by phytoplankton. The metalimnion is the stratum of maximum temperature change; density differences may be sufficient to impede settling of detrital material into the lower stratum, the hypolimnion. A thermocline is present within the metalimnion if the rate of temperature change exceeds 1oC/m. The hypolimnion, which overlies the lakebed sediments, typically is more thermally stable than the epilimnion or metalimnion. During thermal stratification, the hypolimnion is isolated from atmospheric exchange and may develop a dissolved oxygen deficit if biological and chemical oxygen demands exceed the oxygen mass available within the hypolimnion at the onset of thermal stratification. During spring and autumn, solar radiation input is less than during summer, and windy conditions often are more prevalent. This combination facilitates convective circulation, the process whereby a weakly stratified water column undergoes vertical mixing when wind energy is sufficient to overcome the thermal gradient. A lake is termed dimictic if it undergoes convective circulation throughout its water column in spring and autumn. Such mixing is an important mechanism for the vertical movement of water-quality constituents such as dissolved oxygen, nutrients, and trace metals.
Heating of Coeur d’Alene Lake during the 1991–92 and 2004–06 studies produced annual maximum water temperatures within 1.2oC among the 5 years compared. The maximum water temperature measured during 1991 was 23.4oC at station 6 during early August and, during 1992, 23.2oC at station 6 during late June (Woods and Beckwith, 1997; appendix A). Maximum water temperatures for 2004, 2005, and 2006 were 22.6oC (station 6, mid-July), 22.2oC (station 6, mid-July), and 22.4oC (station 4, mid-August), respectively (appendix B).
These similarities in maximum water temperatures did not result in similar thermocline depths among the stations for the 5 years. The median thermocline depths for stations 1, 3, 4, 5, and 6 during the two study periods ranged from 5.5 m at station 6 in 2004–06 to 13 m at station 1 in 1991–92 (fig. 3). At each station, median thermocline depths varied by as little as 1.5 m (station 6) and by as much as 3.5 m (station 5). These differences reflect numerous influences on a location’s propensity to stratify thermally; important influences include depth, exposure to wind, proximity to riverine inflows, and exposure to internal waves at the thermocline (Woods, 2004).
The development and breakdown of thermal stratification have important consequences for the fate and transport of constituents in lakes. At the deep stations in Coeur d’Alene Lake (stations 1, 3, and 4), thermal stratification (depicted by the presence of a thermocline) developed by mid-July and persisted until early October 1991. In 1992, the thermocline developed by June and persisted until mid-October (Woods and Beckwith, 1997, fig. 9). During 2004, thermoclines developed by mid-June and persisted into mid-October, except at station 4 where the thermocline disappeared prior to the October sampling trip (appendix B). The thermoclines at stations 3 and 4 developed in mid-June 2005, following an earlier development in late May at station 1; thermoclines at these three stations disappeared by early October (appendix B). During 2006, thermoclines developed by mid-June and persisted into mid- October, except at station 4 which was not sampled during the October sampling trip (appendix B). The thermocline at station 4 was detected in late August.
Water column transparency was characterized as Secchi-disc transparency depth and euphotic zone depth, both of which are shown in figure 3 as median values for each station for the two study periods. The smallest medians for both variables were measured at station 6, the shallowest of the five stations and the most exposed to turbid riverine inflows from the St. Joe River. The three deep stations had similar median values although the largest values for Secchi-disc transparency depth were recorded at station 1, the farthest from riverine inflows. The variation of euphotic zone depth among the five stations was similar to that of Secchi-disc transparency depth. Some lack of correspondence between the two variables at a station likely is due to the subjective nature of measuring Secchi-disc transparency depth.
The foregoing comparison of water temperatures and water column transparency among the study periods and the five stations indicated no substantive changes in water column heating and convective circulation because heating and circulation are largely driven by physical limnological processes, which have not changed at Coeur d’Alene Lake. Anthropogenically induced changes in sediment delivery to the lake and (or) increased phytoplankton productivity have the potential to affect water column transparency. No such effects were detected with these data. A longitudinal gradient exists from south (low transparency) to north (high transparency) as a consequence of turbid riverine inflows entering the lake from the south.
The concentration of dissolved oxygen in water is affected by temperature, barometric pressure, oxygen production by photosynthesis, oxygen consumption by respiration and decomposition, exchange across the air-water interface, and mixing (Hem, 1985). The temporal and spatial distributions of dissolved-oxygen concentrations in lakes are important indices of water quality. However, some effects described earlier in this report concealed important processes that can be better evaluated with the variable percent saturation of dissolved-oxygen concentration, defined here as the ratio (expressed as a percentage) of measured dissolved-oxygen concentration to that which would exist under saturated conditions at the same temperature.
During the 1991–92 study, the overall range in dissolved-oxygen concentrations over depth and time among the pelagic stations was 0 to 13.6 mg/L; both extremes were measured at station 6 (Woods and Beckwith, 1997; appendix A). The overall range in dissolved oxygen concentrations during 2004–06 was 0.2 to 13.7 mg/L with the minimum measured at station 6 and the maximum at station 1 (appendix B). Over the two study periods, the highest concentrations at each station were measured in the winter in association with minimum water temperatures confirming the inverse relation between dissolved-oxygen concentration and temperature. Minimum concentrations were measured in the hypolimnia of each station during late summer or autumn as prolonged thermal stratification restricted mixing of the oxygenated upper water column with the hypolimnion.
Ranges for the percent saturation of dissolved-oxygen concentrations for the 5 stations during the study periods are plotted in figure 4. Wider ranges and smaller minimum values were recorded at stations 5 and 6. Station 6 attained anoxic or near-anoxic conditions in each year; station 5 percent saturations were as low as 24 in the 2004–06 study. Similar minimum percent saturations, ranging from 54 to 65, were recorded at the three deep stations (1, 3, and 4), which have not shown hypolimnetic anoxia.
Saturations greater than 100 percent were measured at all stations in 4 or more of the 5 years. These saturations were associated with the euphotic zone during the summer months when photosynthetic-oxygen production exceeded oxygen consumption processes. The minimum percent saturations were in the hypolimnia of the stations during late summer or autumn.
A statistical analysis of dissolved-oxygen concentrations and percent saturations among the two study periods and five stations (tested separately) indicated no significant change between study periods. At all stations, median dissolved-oxygen concentrations were significantly lower in the hypolimnion than in the euphotic zone. Median concentrations were significantly lower at stations 5 and 6 than at all other stations for both study periods. A significant longitudinal gradient in the loss of dissolved oxygen is apparent from north to south. However, that gradient is mostly a consequence of the shallow depths and higher biological productivity at stations 5 and 6.
Phosphorus is one of several essential nutrients in the metabolism of phytoplankton and aquatic plants and was measured in both studies. Eutrophication research has focused heavily on phosphorus, the nutrient often having the smallest supply-to-demand ratio for aquatic plant growth (Ryding and Rast, 1989). Phosphorus concentrations for both studies were reported as total phosphorus and dissolved orthophosphate (dissolved phosphorus also was reported for the 2004–06 study). Total phosphorus represents phosphorus in solution (dissolved or colloidal) and phosphorus contained in or attached to biotic and abiotic particulate material. Dissolved phosphorus and orthophosphate are determined from the filtrate that passes through a filter with a nominal pore size of 0.45 µm.
The comparison of phosphorus focused on total phosphorus because analytical method reporting limits for dissolved orthophosphate differed between the two studies, resulting in numerous non-detected orthophosphate values in the 2004–06 study. Table 6 presents changes in method reporting limits and number of non-detected values in the 1991–92 and 2004–06 studies for relevant constituents, including dissolved orthophosphate. During the 1991–92 study, the method reporting limit for dissolved orthophosphate was 1 µg/L; mean concentrations over the 2 years were slightly more than 1 µg/L (Woods and Beckwith, 1997), but median concentrations at all pelagic stations except station 6 were less than 1 µg/L. The more rigorous statistical basis for method reporting limits (Childress and others, 1999) applied during the 2004–06 study resulted in a reporting limit of 6 µg/L. About 78 percent of dissolved orthophosphate analyses in the 2004-06 study were non-detected values, and lake-wide median orthophosphate concentrations at all pelagic stations in the 2004–06 study were non-detected values, less than 6 µg/L. Therefore, statistical comparisons were made only for total phosphorus. The comparisons were made using the Kaplan-Meier logrank and Peto-Prentice score tests, incorporating the multiple reporting limits for total phosphorus.
Total phosphorus concentrations among the pelagic stations ranged from less than 1 to 192 µg/L during 1991–92 (Woods and Beckwith, 1997; appendix A). The largest concentration measured was under anoxic conditions within the hypolimnion of station 6 during September 1991. Total phosphorus concentrations for 2004–06 ranged from 2 (estimated) to 168 µg/L with the largest concentration from the anoxic hypolimnion of station 6 during August 2006 (appendix B). Median concentrations of total phosphorus within the euphotic zone and hypolimnion of each station for both study periods are shown in figure 5. A significant longitudinal gradient is evident from south to north, with higher concentrations associated with the two shallower, southern stations. When grouped by study, total phosphorus concentrations were significantly higher in the euphotic zone and hypolimnion in the 2004–06 study than in the 1991–92 study at all pelagic stations, which corresponds with the increase in total phosphorus loads and concentrations in inflows from the Coeur d’Alene and St. Joe Rivers.
The differences in median concentrations of total phosphorus between euphotic zone and hypolimnion at each station (fig. 5) reveal that water quality differed at the three deeper stations than the two shallower stations. During each study period, statistically similar concentrations for the two depth zones were detected at the deeper stations (1, 3, and 4). In most cases the difference was only 1 µg/L or less. In contrast, the concentration differences between the euphotic zone and hypolimnion of stations 5 and 6 ranged from 2.5 to 3 µg/L and from 5.5 to 7 µg/L, respectively. The hypolimnetic phosphorus concentrations at stations 5 and 6 significantly exceeded those of their euphotic zones in all cases.
Higher median concentrations of total phosphorus within the euphotic zones and hypolimnia of stations 5 and 6, compared to the concentrations detected at the deeper pelagic stations, reflect several processes. Seasonal anoxia at station 6 releases dissolved constituents such as phosphorus, nitrogen, iron, and manganese from lakebed sediments and detrital material within the water column. Additionally, both stations are within the inflow plume of the St. Joe River, which is a primary source of phosphorus for the lake.
Total phosphorus was not detected in any of the blank samples submitted during the 2004–06 study; therefore, decontamination procedures were considered adequate to remove this constituent from sampling equipment. Mean relative standard deviation, a measure of variability induced during sample processing and analysis, was 12 percent for total phosphorus, based on replicate sample data collected during the 2004–06 study. Most of the high relative standard deviations were for samples with total phosphorus concentrations less than 10 µg/L. At these concentrations, small differences between concentrations in replicate and routine samples can inflate the relative standard deviation. Total phosphorus concentrations in 38 percent of the replicate samples matched exactly with corresponding routine samples. Selected analytical data for quality control samples collected during the 2004–06 study are provided in appendix C.
Nitrogen, like phosphorus, is essential to the metabolism of aquatic biota. In Coeur d’Alene Lake, the supply-to-demand ratio of nitrogen is small (but typically not as small as that of phosphorus); thus nitrogen may limit the growth of aquatic plants. The nitrogen cycle in aquatic ecosystems is complex because nitrogen can assume many redox states and chemical forms, and because most processes involving nitrogen are biologically mediated. In aquatic ecosystems, nitrogen commonly exists as dissolved molecular nitrogen, nitrogen-containing organic compounds, ammonia, nitrite, and nitrate. Nitrogen concentrations for the two studies generally were reported as total nitrogen and dissolved inorganic nitrogen. Total nitrogen represents the sum of ammonia, organic nitrogen, nitrite, and nitrate in solution (dissolved or colloidal) and contained in or attached to biotic and abiotic particulate material. Dissolved inorganic nitrogen represents the sum of ammonia, nitrite, and nitrate and is determined from the filtrate that passes through a filter with a nominal pore size of 0.45 µm.
Comparisons involving nitrogen focused on dissolved-inorganic nitrogen concentrations because analytical method reporting limits for total-nitrogen concentrations differed greatly between the two studies. The method reporting limit for total nitrogen during 1991–92 was 205 µg/L (200 µg/L for total ammonia plus organic nitrogen and 5 µg/L for dissolved nitrite plus nitrate [table 6]); as such, nearly all total nitrogen concentrations in Coeur d’Alene Lake were reported as less than 205 µg/L for that period. Improved analytical methods prior to the 2004–06 study lowered the reporting limit to 60 µg/L. This change effectively negated the ability to make comparisons of total nitrogen between the two studies.
Method reporting limits for dissolved nitrite plus nitrate and dissolved ammonia also differed between the two studies as shown in table 6; however, the incidence of concentrations less than the reporting limit was much less than for total nitrogen. During the 1991–92 study, method reporting limits were 5 and 2 µg/L as nitrogen for dissolved nitrite plus nitrate and dissolved ammonia, respectively. During the 2004–06 study, a more rigorous statistical basis for method reporting limits (Childress and others, 1999) resulted in reporting limits of 16 and 10 µg/L as nitrogen, respectively, for dissolved nitrite plus nitrate and dissolved ammonia.
Dissolved inorganic nitrogen concentrations among the pelagic stations ranged from less than 7 to 332 µg/L (320 µg/L of which was ammonia) during 1991–92 (Woods and Beckwith, 1997). The largest concentration was measured under anoxic conditions within the hypolimnion of station 6 during September 1991. The dissolved inorganic nitrogen concentrations for 2004–06 ranged from less than 21 to about 265 µg/L (249 µg/L of which was ammonia), with the largest concentration from the anoxic hypolimnion of station 6 during late August 2004.
Median concentrations of dissolved inorganic nitrogen within the euphotic zone and hypolimnion of each station are shown in figure 6. Due to an increase in method reporting limits for the 2004–06 study, median concentrations in the euphotic zone for 2004–06 were less than 26 µg/L at all five stations. Unlike median total phosphorus concentrations in the euphotic zone (fig. 5), no distinct longitudinal gradient for median concentrations of dissolved inorganic nitrogen was evident over the five stations in the euphotic zone (fig. 6). However, such a gradient did exist for hypolimnetic concentrations (fig. 6). The three deep stations had significantly higher median concentrations of dissolved inorganic nitrogen than did stations 5 and 6, with station 4 having the highest median concentrations. Median hypolimnetic concentrations for the three deep stations ranged from 80 µg/L (station 1 in 2004–06) to 106 µg/L (station 4 in 1991–92). In contrast, the range for median hypolimnetic concentrations at stations 5 and 6 was from 30 µg/L (station 6 in 1991–92) to 58 µg/L (station 5 in 2004–06).
The significant differences in median concentrations of dissolved inorganic nitrogen between the euphotic zone and hypolimnion of the three deep stations (fig. 6) were attributable to several limnological processes. The greater depth of stations 1, 3, and 4 decreased the frequency of convective circulation and allowed development of water-quality stratification during summer. Phytoplanktonic assimilation of dissolved inorganic nitrogen decreased summertime concentrations within the euphotic zone (Woods, 2004); this process also occurred at stations 5 and 6. Additionally, the highest loads and concentrations of dissolved inorganic nitrogen were delivered to the lake during winter months, when inflow plumes travel through the lake as underflow. As lower-concentration inflows in other seasons flushed through the lake as overflow or interflow, some higher concentration contributions may have been trapped throughout the year in the deeper areas of the lake. The process of benthic flux likely served to enrich hypolimnetic concentrations at the deeper stations as well as station 5.
No significant temporal changes were detected in hypolimnetic dissolved inorganic nitrogen between study periods at any pelagic station. A statistical comparison in dissolved inorganic nitrogen in the euphotic zone between study periods was hampered by the change in reporting limits and number of reported non-detected values in the 2004–06 study. In the 1991–92 and 2004–06 studies, a significant difference existed between euphotic zone and hypolimnetic concentrations at all five stations, but particularly at the three deep stations. As previously discussed, a robust comparison was difficult for all nitrogen fractions between the two studies because of differences in method reporting limits.
Dissolved ammonia, nitrite, and nitrate were not detected in any of the blank samples submitted during the 2004–06 study; therefore, decontamination procedures were deemed adequate to remove these analytes from sampling equipment. Mean relative standard deviations were 8 percent for dissolved ammonia and 1 percent for dissolved nitrite plus nitrate, based on replicate sample data collected during the 2004–06 study. As with total phosphorus, most high relative standard deviations were for samples with low concentrations. Seventy-one percent of ammonia analyses and 50 percent of nitrite plus nitrate analyses in replicate samples matched corresponding routine samples.
The limiting nutrient concept states that the ultimate yield of a crop will be limited by the essential nutrient most scarce relative to the specific needs of the crop (Ryding and Rast, 1989). This concept, in concert with the stoichiometry of the photosynthesis equation, has led to the widespread use of nitrogen-to-phosphorus ratios (N:P). These ratios commonly are used in eutrophication studies to evaluate if nitrogen or phosphorus was the nutrient most likely to limit phytoplankton growth. The atomic ratio of nitrogen to phosphorus, 16N:1P, in the photosynthesis equation corresponds to a mass ratio of 7.2N:1P. Typically, N:P values are calculated using the biologically available form of these two nutrients, dissolved inorganic nitrogen and dissolved orthophosphate. If N:P (by weight) is less than 7.2, then nitrogen may be limiting, whereas if N:P exceeds 7.2, then phosphorus may be limiting (Ryding and Rast, 1989).
The mean euphotic zone values of N:P for Coeur d’Alene Lake previously published for the 1991–92 study (Woods and Beckwith, 1997) were 34.4 for 1991 and 22.7 for 1992 and indicated a strong tendency towards phosphorus limitation of phytoplankton growth. The lower value for 1992 was a consequence of lower concentrations of dissolved inorganic nitrogen because dissolved orthophosphate concentrations were comparable during both years.
The calculation of N:P for 2004–06 dissolved inorganic nitrogen data was complicated by a reporting limit of 6 µg/L for dissolved orthophosphate and median concentrations of dissolved inorganic nitrogen less than 26 µg/L for 2 years in the study. These complications resulted in few samples with dissolved orthophosphate and dissolved inorganic nitrogen greater than their respective reporting limits. Values of both constituents for the calculation of N:P ratios were detected in only 36 percent of samples collected during the 1991–92 study and 19 percent of samples collected during the 2004–06 study. However, for a general comparison between study periods, the median N:P ratio values for each station when dissolved orthophosphorus and dissolved inorganic nitrogen were detected were calculated and compared. Median detected N:P ratios in the 1991–92 study ranged from 16 (station 6) to 26 (station 5). Median detected N:P ratios in the 2004–06 study ranged from 7 (station 6) to 13 (station 3). Whereas the lake was clearly phosphorus-limited during 1991–92, the lake may have been much closer to the boundary value of 7.2 that separates nitrogen from phosphorus limitation during 2004–06. However, the data are insufficient to draw reliable conclusions regarding limiting nutrients.
Chlorophyll-a is the primary photosynthetic pigment of phytoplankton and, as such, is used as an estimator of phytoplankton biomass and an indicator of lake productivity. As discussed previously, the comparison of chlorophyll-a concentrations between the two studies must recognize that different analytical methods were used. Concentrations of chlorophyll-a from samples collected in 1991-92 were adjusted using the NWQL regression equation to estimate the results that would have been expected using the fluorometric method that was applied in 2004-06. Adjusted concentrations from 1991–92 were used in statistical comparisons among stations and study periods.
Adjusted chlorophyll-a concentrations in Coeur d’Alene Lake during 1991–92 ranged from less than 0.1 to 3.1 µg/L; the highest concentrations were measured at stations 5 and 6 in both years (Woods and Beckwith, 1997; appendix A). Chlorophyll-a concentrations during 2004–06 (analyzed fluorometrically) ranged from less than 0.1 to 17.9 µg/L; the highest concentrations were measured at station 6 (appendix B). However, differences in median concentrations among stations were not statistically significant, and therefore, no spatial pattern was detected.
Median concentrations of chlorophyll-a within the euphotic zone of the five stations are plotted for each year in figure 7. Median concentrations increased significantly at all stations in the 2004–06 study compared to the 1991–92 study. Adjusted median concentrations for 1991–92 ranged from 0.9 µg/L (stations 1, 3, and 4) to 1.2 µg/L (station 6). For 2004–06, median concentrations ranged from 1.5 µg/L (station 5) to 2.7 µg/L (station 6). For comparison, the unadjusted median values for the 1991–92 study also are plotted on figure 7. Although adjusted chlorophyll-a concentrations for the 1991–92 study increased compared to the unadjusted concentrations, they were still significantly less than concentrations during 2004–06 at every station.
Blank samples were not analyzed for chlorophyll-a during the 2004–06 study. One replicate sample was analyzed for chlorophyll-a; relative standard deviation was 5 percent in comparison with the corresponding routine sample.
A nutrient load study conducted in 1975 as part of the National Eutrophication Survey led to the determination that Coeur d’Alene Lake was mesotrophic, or moderately productive (U.S. Environmental Protection Agency, 1977), which prompted continued observation of the lake trophic status. The 1991–92 study determined that the lake had become oligotrophic, or less productive, in terms of lake-wide mean chlorophyll-a and total phosphorus concentrations and mesotrophic in terms of Secchi-disc transparency depth, likely due to the documented reduction in nutrient loads throughout the basin from 1975 to 1991 (Woods and Beckwith, 1997). Although lake-wide mean chlorophyll-a and total phosphorus concentrations during the 2004–06 study have increased, the trophic status classification remained the same as determined by the 1991–92 study based on these variables. Lake-wide median values for the three limnological variables used to classify the trophic status are presented for both studies and compared to trophic classification data from Wetzel (1983) in table 7. Wetzel (1983) classifies a lake as oligotrophic if average or median total phosphorus concentrations are in the range of 3.0–17.7 µg/L, average or median chlorophyll-a concentrations are in the range of 0.3–4.5 µg/L, and average or median Secchi-disc transparency depths are in the range of 5.4–28.3 m. Because the lake-wide median Secchi-disc transparency depth for Coeur d’Alene Lake (4.6 m) was less than this range, the lake is classified as mesotrophic based on this variable. The mesotrophic classification range for median Secchi-disc transparency depth is 1.5-8.1 m. Wetzel (1983) provides detailed discussion of the boundaries that separate trophic classifications.
The comparison of trace metal concentrations between the two studies was restricted to total concentrations because dissolved concentrations were not analyzed in the 1991–92 study. Of the 126 samples collected from the euphotic zone and hypolimnion during the 1991–92 study, median total-recoverable concentrations were less than 1, 3.3, and 98.6 μg/L, respectively, for cadmium, lead, and zinc (Woods and Beckwith, 1997). About 93 percent of cadmium, 24 percent of lead, and 13 percent of zinc concentrations were less than their analytical reporting limits of 1, 1, and 10 µg/L, respectively (table 6). Most trace-metal concentrations from station 6 were reported as “less-than” or censored data.
Zinc concentrations were the focus of a comparison of median concentrations in euphotic zone and hypolimnion samples from the five stations during the two studies (fig. 8). Concentrations in 2004–06 significantly decreased from those measured during 1991–92 for the euphotic zone and hypolimnion and at all stations except station 6. At every pelagic station except station 6, paired zinc concentrations were significantly higher in the hypolimnion than in the euphotic zone. During 1991–92, median zinc concentrations ranged from less than 10 (station 6) to 100 µg/L (station 4) in the euphotic zone and from less than 10 (station 6) to 132 µg/L (station 4) in the hypolimnion. During 2004–06, median zinc concentrations ranged from 2 (station 6) to 55 µg/L (station 4) in the euphotic zone and from 2 (station 6) to 78 µg/L (station 4) in the hypolimnion.
A distinct spatial difference in total zinc concentrations is evident as shown in figure 8. The lowest median values are from station 6 which was far from the influence of the trace-metal-rich inflow of the Coeur d’Alene River. Although station 5 also would appear to be out of the influence of the Coeur d’Alene River, total zinc concentrations shown in figure 8 indicate otherwise. The three deep stations are directly influenced by the inflow plume of the Coeur d’Alene River and exhibited significantly higher total zinc concentrations than the two shallower stations. Concentrations at station 4 are slightly but statistically higher than all other stations.
At all pelagic stations except station 6, total cadmium and total lead concentrations decreased significantly from 1991–92 to 2004–06. However, as stated previously, robust temporal comparisons for cadmium and lead are hampered by a substantial decrease in reporting limit in the 2004–06 study. Median cadmium and lead concentrations were significantly higher in the hypolimnion than in the euphotic zone at every station except station 6, where no significant stratification was detected regarding cadmium. As with zinc, the three deep stations exhibited significantly higher total cadmium and total lead concentrations than the two shallow stations; the highest total cadmium and total lead concentrations were measured at station 4.
Total cadmium was not detected in any of the blank samples submitted during the 2004–06 study; however, total lead and total zinc were detected in 24 and 19 percent, respectively, of the blank samples. Field notes recorded during one of the total lead and total zinc detections (April 2005) state that pieces of a rubber seal in the grab sampling device were in the blank sample water during sample processing. The sampling device was soon replaced. Trace metals concentrations in samples collected around this time may be biased high. Mean relative standard deviations in replicate sample data were 2 percent for total cadmium, 3 percent for total lead, and 5 percent for total zinc. Therefore, variability induced during sample processing and analysis was low for these constituents.
The 1991-92 study included two littoral sampling events and the 2004-06 study included eight sampling events. To compare pelagic and littoral water quality between the two studies, only samples collected during the same weeks were compared. For pelagic and littoral comparisons, the latter study period is termed “2004–05” because comparison data were not collected during the same weeks in water year 2006. Mutual constituents sampled between littoral and pelagic stations during the two studies included total phosphorus, chlorophyll-a, and zinc. Comparisons were made using two methods:
Comparison 1 indicated that when data sets were compared as a whole, median total phosphorus concentrations were significantly lower in the littoral zone than in the pelagic euphotic zone for weeks sampled in 2004–05. Median total zinc concentrations were significantly lower in the littoral zone than in the pelagic euphotic zone for weeks sampled in 1991–92. In all other cases, total phosphorus, chlorophyll-a, and total zinc concentrations were statistically similar for dates sampled in both study periods. Zinc concentrations for the littoral zone and pelagic euphotic zone in the 2004–05 study decreased significantly from the 1991–92 study.
Comparison 2 indicated that total phosphorus, chlorophyll-a, and total zinc concentrations are statistically similar between the pelagic stations and their closest littoral stations on the same weeks sampled. Table 8 presents the median values and ranges for each constituent for each pelagic station and its closest paired littoral stations. A true comparison at station 6 could be made only for total zinc concentrations because chlorophyll-a and total phosphorus were not collected at any of the closest littoral stations. However, chlorophyll-a and total phosphorus concentrations were significantly higher and total zinc concentrations were significantly lower at station 6 than at NS1 (near station 5), the closest littoral station where all three constituents were collected. This observation along with observations previously stated in this report regarding concentration gradients for nutrients and metals, indicate that water-quality conditions at station 5 (and corresponding littoral stations) are slightly different than at station 6.
Because statistically significant differences were determined in some analyte concentrations among pelagic stations, comparison 2 provides a more valid determination than comparison 1 of the difference between littoral and pelagic water quality because use of this method pairs each pelagic station with the respective littoral area. Additionally, any error introduced by temporal variation was decreased because pelagic samples were compared only with littoral samples collected during the same week.
Focused limnological studies of Coeur d’Alene Lake were conducted between the 1991–92 and 2004–06 studies and included many variables that were sampled in the 1991–92 and 2004–06 studies. These post-1992 studies benefited the design of the 2004–06 study and added spatial and temporal information for many variables. The limnological basis for sampling the lake during the 2004–06 study (table 2) recognized the important role played by physical limnological processes in determining how the lake may be affected in the future by watershed-remediation activities associated with the Bunker Hill Superfund Site OU3 ROD and Lake Management Plan.
Two recent reports (URS Greiner, Inc., and CH2M-Hill, Inc., 2001 and Woods, 2004), noted the importance of physical limnological processes in the fate and transport of constituents and, by extension, water-quality conditions in Coeur d’Alene Lake. Those processes include, but are not limited to, lake hydraulic residence time, wind-generated circulation, and in-lake routing of inflow plumes.
Surface-water runoff from a drainage basin commonly is the primary source of inflow to most lakes; ground-water inflow and direct precipitation on the lake surface usually are of much smaller magnitude. The rate at which water enters and leaves a lake affects the amount of turbulence within the lake water column, both in the horizontal and vertical dimensions. Limnologists have long used the ratio of inflow rate to lake volume, commonly termed flushing rate (Ryding and Rast, 1989), to quantify this physical limnological process. As flushing rate increases, the hydrologic influence on the lake from the drainage basin also increases because the lake volume is frequently replaced. The inverse of flushing rate is retention time, which represents the time theoretically needed to fill a lake if the lake were empty. If lake volume is divided by lake outflow rate instead of inflow rate, then the time needed to empty the lake, the hydraulic residence time, is obtained. On an annual basis, retention time and hydraulic residence time often are comparable. Hydraulic residence time was selected for analysis in this report because outflow rates for Coeur d’Alene Lake have been measured for many years; measurements of the sum of inflow rates for all significant inflow sources are rare for most lakes. Although retention time and hydraulic residence time are theoretical concepts, the processes that they incorporate are important for understanding fate and transport of constituents in lakes. In years of greater than normal inflow and outflow, increased water column turbulence and advective transport of constituents can be expected. Conversely, years of less than normal inflow and outflow produce less water column mixing and may increase the constituents trapped in the lake.
The hydraulic residence time for Coeur d’Alene Lake is 0.50 year based on a normal full-pool volume of 2.8 km3 divided by the mean annual outflow rate of 5.6 km3/yr. Outflow statistics for the lake were derived for a 93-year period of record (1913–2006) for the USGS gaging station on Spokane River near Post Falls, Idaho (U.S. Geological Survey, 2007). During that period of record, however, annual mean outflow volume varied widely. For the minimum outflow rate of 1.9 km3/yr, the hydraulic residence time increased to 1.5 years; conversely, for the maximum outflow rate of 10.5 km3/yr, hydraulic residence time decreased to 0.27 year. This range of hydraulic residence times and historical outflow rates indicate that the lake volume could be replaced in as few as 98 days or as many as 548 days. This range also represents an index of the physical limnological process of water column turbulence and the presumed relation with hydraulic residence time. Retention of constituents delivered from the drainage basin to Coeur d’Alene Lake is expected to decrease as hydraulic residence time decreases. In addition to inter-annual variability, outflow from the lake varies intra-annually in response to climate conditions in the drainage basin. During the 1913–2006 period of record, the smallest monthly mean outflow of 0.07 km3/mo was in August, whereas the largest monthly mean outflow of 1.3 km3/mo was in May. Based on these outflow rates, constituents inflowing to the lake are more likely to be retained in August when water column turbulence and water outflow are minimal.
Although insight was gained into the relation of inflow or outflow magnitude on the generation of turbulence and advective transport within lakes, the concepts of retention and hydraulic residence times remain theoretical because lakes rarely are filled or emptied. The two concepts are best suited for general comparisons among lakes representing wide ranges of retention and hydraulic residence times.
The physical, chemical, and biological responses of a lake to the delivery of water and associated constituents from the drainage basin are closely tied to shape, exposure to wind, and depth of the lake. A long, narrow lake shape, like that of Coeur d’Alene Lake, is more prone to channelized flow along its major axis. Shape and surface area of a lake are determinants of wind exposure. Increased wind exposure enhances the development of large-scale turbulent processes such as surface waves and internal seiches, which can displace large masses of water in the horizontal and vertical dimensions, important mechanisms for water column mixing. Deep lakes, such as Coeur d’Alene Lake, are less prone to turbulent, full-depth mixing by wind energy, so lakebed sediments are less likely to be periodically re-suspended except perhaps by internal waves near the thermocline. The low frequency of turbulent, full-depth mixing in deep lakes also restricts the exchange of dissolved, colloidal, and particulate constituents between the upper and lower water columns.
The movement of riverine inflows within a lake can be quite complex because of lake characteristics such as shape, depth, and temporal and spatial differences in density between riverine and lake water. Overflow, interflow, and underflow are three generalized methods of inflow plume routing (Fischer and others, 1979). Overflow occurs if the inflow plume is warmer (less dense) than the lake; river water floats on the lake surface. Interflow occurs when the inflow plume is colder than the lake upper water column but is warmer than the lower water column; interflow is routed to the lake depth where the temperature, or density, of the inflow plume and lake are equal. Underflow occurs when the inflow plume is colder than, or near the temperature of, the lake lower water column. Turbulence at the interface of the inflow plume and the lake mixes the two water masses until thermal equilibrium is reached. The spatial extent of inflow plume routing is highly dependent on the magnitude of riverine discharge. Riverine inflows generated by snowmelt runoff and floods can penetrate farther into the receiving lake because the large inflow volumes produced by such events increase advective transport.
The fate and transport of constituents such as nutrients, trace metals, and sediment within and occasionally through Coeur d’Alene Lake are highly dependent on inflow plume routing of the two primary inflow sources, the Coeur d’Alene and St. Joe Rivers. Inflow plume routing was evaluated using data from the 1991–92 and 2004–06 limnological studies as well as data from several other studies of the lake. For the 1991–92 study, available data included water temperatures for the two rivers (Harenberg and others, 1992; 1993) and numerous full-depth profiles of water temperatures conducted for most months at pelagic stations (Woods and Beckwith, 1997). An inflow plume routing evaluation also was evaluated for water year 1999 using similar water-temperature data; however, pelagic water column profiles were conducted only during June through October at pelagic stations 1, 3, 4, and 5 (URS Greiner, Inc., and CH2M Hill, Inc., 2001).
Forty-four comparisons of inflow and lake temperatures reported for 1991–92 and 1999 (URS Greiner, Inc., and CH2M Hill, Inc., 2001) indicated that overflow was the most common mode of inflow plume routing, occurring in about 60 percent of the comparisons (URS Greiner, Inc., and CH2M Hill, Inc., 2001). Interflow or underflow each occurred in about 20 percent of the comparisons. Overflow occurred in all months except October, November, and December; during those 3 months, underflow was the most likely mode of inflow plume routing. Interflow tended to occur during spring or autumn when the lake was likely transitioning into or out of thermal stratification. Inflow volume also was evaluated as part of the 44 comparisons because of its affect on the spatial extent of inflow plume routing. At small inflows, the influence of the plume on the lake was diminished by rapid mixing and equilibration of riverine and lake temperatures; the opposite was true for large inflows. Underflows tended to be associated only with small inflows, typical for October through December. Underflows occurred during that 3-month period because the Coeur d’Alene and St. Joe Rivers cooled more rapidly than the lake, which has a much greater capacity to store and retain heat captured during the summer months. Overflows occurred over a wide range of inflows because the Coeur d’Alene and St. Joe Rivers each have lengthy backwater-affected reaches that capture a large amount of solar radiation.
During June 1999, USGS scientists using specialized water-quality instrumentation and water column sampling methods tracked discharge and chemistry of the Coeur d’Alene and St. Joe River inflow plumes into and through Coeur d’Alene Lake. The short-term study sought to answer two questions: (1) Can riverine inflows and associated chemical nature be clearly identified within the lake? (2) Do sediments, nutrients, and trace metals carried by riverine inflows travel far enough into the lake to be discharged out of the lake into the Spokane River? The field work was conducted during June 2–3, 1999, at pelagic stations 1–5 in addition to three stations at the mouths of the Coeur d’Alene and St. Joe Rivers and at the lake outlet to the Spokane River. Study results clearly identified the riverine inflows as a combination of overflow and interflow within the upper 5–13 m of the lake, from station 4 and northward to the lake outlet (URS Greiner, Inc., and CH2M Hill, Inc., 2001). Much of the lake south of station 4 is shallow enough to allow full-depth mixing of the two riverine inflows. Light transmission, conductivity, and concentrations of lead, zinc, and nitrogen differed substantially between the riverine inflows and lake water. Lead concentrations delivered by the Coeur d’Alene River were higher than concentrations in lake water. Zinc concentrations, delivered almost exclusively by the Coeur d’Alene River, were lower than concentrations in lake water. Light transmission, conductivity, and nitrogen concentrations in riverine water also were lower than concentrations in lake water. The chemical nature of water exiting the lake to the Spokane River during the June 1999 experiment was more closely related to riverine inflows than to lake water; however, this is not expected to be true throughout the year.
Numerous riverine inflow temperatures and water column profiles of water temperature were available for water years 2004 and 2005 during the 2004–06 study to evaluate inflow plume routing. Table 9 lists 15 riverine inflow temperatures measured within two weeks of a limnological sampling trip in water years 2004 and 2005, similar to a table presented in URS Greiner, Inc., and CH2M Hill, Inc. (2001). One measurement is shown for which there was no sampling data. The water temperatures and discharges of the Coeur d’Alene and St. Joe Rivers were compared to the range of water column temperatures at pelagic station 3, the deepest area in the lake. Of the 15 comparisons, 7 indicated overflow, 6 indicated interflow, and 2 indicated underflow (table 9). Underflows occurred in December of both years because inflowing rivers cooled faster than the lake, which had a much larger heat storage capacity. Interflows occurred in early April, May, and June and in early September 2004 and in late March and mid-May 2005. The predominance of interflows during spring months was partly due to more rapid warming of the inflowing rivers than the lake, which was emerging from winter cooling. Overflows were recorded during October of both years, as late summer heating kept riverine temperatures similar to lake temperatures. For summer months, overflow was recorded only in July 2004 and during June, July, and August 2005. Cold (2.0–2.3°C) riverine water that entered the lake in February 2005 during a period of inverse stratification (lake water at 3.4°C with maximum density, underlying colder, less dense water) caused an overflow condition.
The primary focus of the 1991–92 study was on nutrients and lake productivity; collection of trace metal samples from the lake water column was less frequent and at fewer depths. However, lakebed sediments also were sampled during this period for characterization of concentration, partitioning, and environmental availability of selected trace metals (Horowitz and others, 1993; 1995a; 1995b; 2001). These additional sediment studies estimated that 75 million metric tons of metals-enriched sediments cover about 85 percent of the bottom of Coeur d’Alene Lake. These sediments range in thickness from 17 to more than 119 cm. Horowitz and others (1993) reported mean concentrations of arsenic (151 mg/kg), cadmium (62 mg/kg), lead (1,900 mg/kg), mercury (1.8 mg/kg), and zinc (3,600 mg/kg) in lakebed surface sediments in Coeur d’Alene Lake. Naturally occurring background concentrations in lakebed sediments were reported for arsenic (5 mg/kg), cadmium (3 mg/kg), lead (24 mg/kg), mercury (0.05 mg/kg), and zinc (110 mg/kg) (Horowitz and others, 1993). Metals-enriched sediments in Coeur d’Alene Lake generally are extremely fine-grained and are susceptible to remobilization by river- and wind-induced lake currents.
Although eutrophication by nutrients and contamination of lakebed sediments by trace metals may have appeared to be unrelated water-quality problems for Coeur d’Alene Lake, the potential development of water column anoxia near the lakebed could release nutrients and trace metals from lakebed sediments into the overlying water column. When anoxia develops, a well-known consequence of eutrophication, reductive dissolution of metal oxides, such as iron and manganese, can occur in lakebed sediments and the hypolimnion (Brezonik, 1994). At dissolved-oxygen concentrations less than 1 mg/L, iron-phosphate complexes may become unstable and dissolve and release phosphorus from lakebed sediments into the hypolimnion (Jones and Bowser, 1978; Baccini, 1985). Horowitz and others (1993) reported that most lakebed trace metals in Coeur d’Alene Lake were associated with iron oxides. Under anoxia, the reductive dissolution of iron oxides could release trace metals into the water column of the lake. The potential development of anoxia in Coeur d’Alene Lake prompted a study by the USGS in 1987. Hypolimnetic dissolved oxygen concentrations as low as 4 mg/L were detected during late summer in the deep northern basin of the lake (Woods, 1989).
Based on published literature related to anoxic releases of constituents from lakebed sediments, the 1991–92 study did not anticipate releases of nutrients and trace metals from the lakebed sediments because hypolimnetic dissolved oxygen concentrations exceeded 1 mg/L. The study’s focus was oriented toward understanding the lake assimilative capacity for nutrients to maintain adequate concentrations of hypolimnetic dissolved oxygen. However, additional studies done in Coeur d’Alene Lake by N.S. Simon (U.S. Geological Survey, written commun., 1993), Kuwabara and others (2000; 2003; 2006), and Winowiecki (2002) after the 1991–92 study detected the presence of a benthic flux, the movement of constituents into and out of lakebed sediments, even in the presence of hypolimnetic dissolved oxygen concentrations greater than 1 mg/L.
Benthic fluxes define the transport of dissolved constituents across the sediment-water interface. In the absence of advection, this transport primarily is a function of concentrations in the overlying water and in porewater just below the interface (the chemical gradient) and the molecular or eddy diffusion coefficients for the constituents of interest. Benthic fluxes have both direction and magnitude. The direction indicates whether the sediment supplies or removes dissolved constituents to or from the overlying water. Sediments act as a source when dissolved-constituent concentrations in the porewater are greater than in the overlying water (positive flux) or a sink when dissolved-constituent concentrations in the overlying water are greater than in the porewater (negative flux). The magnitude of benthic fluxes depends on the steepness of the chemical gradient and the transport mechanism (molecular compared to eddy diffusion); the magnitude can be used to determine the relative importance of sediments as a source or sink for dissolved constituents in a lake.
Balistrieri (1998) evaluated the benthic flux studies of N.S. Simon (U.S. Geological Survey, written commun., 1993) and Winowiecki (2002) conducted in September 1992. Constituent concentrations in water overlying sediments and in porewater were used by Balistrieri (1998) to calculated benthic fluxes using Fick’s First Law, which assumes that benthic fluxes were controlled by molecular diffusion across the sediment-water interface. Porewater was collected by N.S. Simon (U.S. Geological Survey, written commun., 1993) and Winowiecki (2002) using diffusion-controlled equilibrator samplers, also known as peepers or dialyzers, which were inserted into the lakebed sediments by divers. Balistrieri (1998) noted that oxidation of the porewater samples likely occurred during collection and handling of the samples. This oxidation probably affected the concentrations of certain metals in the porewater (iron and metals that adsorb to iron-oxide phases), and thereby, influenced the benthic flux calculations.
The benthic flux study of Coeur d’Alene Lake conducted in August 1999 and discussed by Kuwabara and others (2000) used an in situ benthic flux chamber that isolated a volume of water overlying the sediment and periodically sampled constituent concentrations in the chamber. Fluxes were calculated as a function of time on the basis of changes in dissolved-constituent concentrations during deployment of the chamber. Using this method, no assumption is required about the mechanism of transport (molecular or eddy diffusion) to determine fluxes.
The results of benthic flux measurements from the chamber method are summarized in table 10 for dissolved cadmium, iron, lead, manganese, and zinc. Each value is reported as micrograms per square centimeter of lakebed surface per year (µg/cm2/yr) moving out of the lakebed sediments. The in situ benthic flux chamber method is considered more representative of conditions in Coeur d’Alene Lake, compared to the peeper method by Winowiecki (2002), because the in situ benthic flux chamber method trapped an existing parcel of lakebed sediment, associated porewater, and overlying lake water, and measured benthic flux without major alteration of pre-existing conditions. Dissolved oxygen concentrations in the lower hypolimnion of the lake near the experiment stations were consistently high, 8–10 mg/L, during August 1999, and oxygen consumption rates in the chamber measured during the experiment were small, 6.0–9.5 millimoles of oxygen per square meter per day. Water in the chambers did not become anoxic at any point during the experiment (Kuwabara and others, 2000). Therefore, oxygen consumption in the chamber during the experiment did not substantially alter the chemical environment at the sediment-water interface from ambient conditions.
Based on the in situ benthic flux chamber results listed in table 10, dissolved cadmium, iron, lead, manganese, and zinc had positive fluxes out of the lakebed sediments. The smallest flux was for lead, 1.9 µg/cm2/yr; the largest was for manganese, 3,700 µg/cm2/yr. Standard error was high (greater than 50 percent) relative to average flux values reported for iron, lead. and manganese (Kuwabara and others, 2000).
The benthic flux of dissolved nutrients and DOC also were measured by the in situ benthic flux chamber method (table 11). The benthic fluxes for orthophosphate, nitrite plus nitrate, ammonia, the sum of nitrite plus nitrate and ammonia (dissolved inorganic nitrogen), and DOC were all positive, indicating movement out of the lakebed sediments. The smallest flux was for orthophosphate, 7.2 µg/cm2/yr; the largest was for DOC, 2,700 µg/cm2/yr. Kuwabara and others (2000) determined that molecular diffusion was likely the dominant transport mechanism for constituents across the sediment-water interface in Coeur d’Alene Lake.
The flux of dissolved constituents across the sediment-water interface is a result of the coupling of physical, chemical, and biological processes (Santschi and others, 1990). Biologically mediated chemical reactions can mobilize dissolved constituents from solid phases within the lakebed sediments, thereby allowing transport by molecular and eddy diffusion. Oxidation of organic matter in the upper sediments of aquatic environments can affect the partitioning of certain constituents between sediment and porewater and produces nutrients such as ammonia and orthophosphate that are needed for phytoplanktonic production. Studies of freshwater and marine sediments indicate that the oxidation of organic matter proceeds using a thermodynamically predictable sequence of oxidants: oxygen, nitrate, manganese oxyhydroxides, ferric oxyhydroxides, and sulfate (Froelich and others, 1979; Berner, 1980; Pedersen and Losher, 1988; and Luther and others, 1998). These reactions are reflected in the composition of porewater as a function of depth. With increasing depth, oxygen and nitrate disappear followed by the appearance of nitrite, ammonia, and dissolved manganese and iron, then sulfate disappears. Oxygen is the primary oxidant of organic matter in the oxic zone. Suboxic conditions occur when oxygen concentrations are very low and nitrate, manganese oxyhydroxides, and ferric oxyhydroxides are used as oxidants. The location of this zone in the upper sediments of Coeur d’Alene Lake is of particular interest with respect to metals because the reduction of manganese and ferric oxyhydroxides can result in the release of associated trace metals (cadmium, lead, and zinc) into the dissolved phase. The absence of oxygen and oxidation of organic matter by sulfate characterize anoxic conditions. Sulfate reduction results in the production of sulfide. This sulfide appears either in porewater or is precipitated as a metal sulfide phase, if there are sufficient concentrations of dissolved metals, primarily iron. The depth scale where these reactions occur can be large (meters) or small (millimeters to centimeters) depending on the supply of organic matter, bottom-water anoxia, and sedimentation rates. If these reactions are compressed into the upper few centimeters just below the sediment-water interface, then concentrations of oxygen, nitrate, and sulfate should be lower in porewater just below the interface, relative to concentrations in the overlying water because of organic matter diagenesis. Thus, the direction of benthic fluxes for these species (oxygen, nitrate, and sulfate) should be into the sediment. Porewater profiles of sulfate within the upper 30 cm of Coeur d’Alene Lake sediments indicate that the transition from oxic, through suboxic, to anoxic conditions exists, depending on location, either within the upper 1 cm or the upper 2–5 cm just below the interface (Balistrieri, 1998). Kuwabara and others (2000) measured oxygen benthic fluxes that ranged from -6.0 to -9.5 millimoles of oxygen per square meter per day, consistent with oxygen consumption by the sediments. Analytical methods employed by Kuwabara and others (2000) did not separate nitrate and nitrite concentrations, so no benthic fluxes were determined for nitrate alone.
In contrast, oxidation of organic matter produces nitrite, ammonia, orthophosphate, and dissolved manganese and iron. The concentrations of these constituents should be higher in porewater relative to the overlying water. Thus, the direction of benthic fluxes for these species should be out of the sediments if no other reactions (adsorption of phosphorus to iron oxides) trap them in the sediments. For example, when a thin oxic layer overlies a suboxic or anoxic layer, dissolved iron produced in the suboxic zone can diffuse into the oxic zone to be oxidized and precipitated as solid-phase iron oxyhydroxides. These iron oxides could then adsorb dissolved orthophosphate. This scenario would effectively trap iron and orthophosphate in the sediment and prevent them from diffusing into the overlying water. However, benthic flux data strongly indicate that sediments in Coeur d’Alene Lake act as a source of dissolved iron and orthophosphate, as well as ammonia and manganese, to the overlying water (tables 10 and 11). Overall, however, there is a net sedimentation of these constituents to lake sediments when they are associated with particulate material, as demonstrated in the lake mass balance calculations for metals (table 5).
Oxidation of organic matter can indirectly mobilize or sequester constituents such as cadmium, lead, and zinc. If these trace metals are predominantly supplied to the sediments in particulate form and in association with iron and manganese oxides, then the reduction of iron and manganese oxides during organic matter oxidation (after all available dissolved oxygen and nitrate are exhausted and reduced) results in the release of not only dissolved iron and manganese, but all other trace metals associated with those phases. The dissolved trace metals can be either transported by molecular or eddy diffusion across the sediment-water interface, or if sufficient concentrations of sulfide are present, be precipitated or co-precipitated as authigenic metal-sulfide phases. Although authigenic sulfides may be forming in Coeur d’Alene Lake, as reported by Harrington and others (1998), most benthic flux data indicate that sediments in Coeur d’Alene Lake act as a net source of dissolved cadmium, lead, and zinc to the overlying water (table 10).
The 2004–06 limnological study sought to increase understanding of benthic flux by identifying it as an important water-quality issue for Coeur d’Alene Lake through the studies by Horowitz and others (1993, 1995a, 1995b); Balistrieri (1998); Harrington and others (1998); Kuwabara and others (2000); and Woods (2004). The logistical difficulties involved in installation and retrieval of diffusion-controlled samplers or in situ flux chambers prompted the design of a boat deployed, modified gravity coring device that could retrieve undisturbed samples of the sediment-water interface. With this device, up-to-date information on benthic flux was added in the 2004–06 study. Interpretation of the data generated by the new sampling device comes with some caveats. First, the device was designed to monitor changes in the chemical composition of water immediately above the lakebed sediments over time and to compare those results to concurrently sampled water 1 m above lakebed sediments. Second, the device did not measure gradients in the pore water or at the sediment-water interface; the analytical results for the small volume sampled represent the net effect of all processes operating at the sediment-water interface. Third, no attempt was made to control oxidation of samples during sample processing; as such, the partitioning between dissolved and particulate phases cannot be assessed.
A summary of the analytical results obtained on 22 sampling trips during the 2004–06 study are listed in table 12 for 8 constituents. All samples were obtained at pelagic station 4, near the main channel in situ flux chamber study reported by Kuwabara and others (2000). Table 12 lists constituents with significantly higher median concentrations in sediment-water interface samples than those in lower hypolimnetic samples (collected 1 m above the lakebed). Samples were compared using the paired Wilcoxon Signed-Ranks Test. No significant difference was apparent for paired median concentrations of dissolved inorganic nitrogen. Paired median concentrations of total phosphorus, dissolved organic carbon, total cadmium, total zinc, total iron, total lead, and total manganese were significantly higher in the sediment-water interface than in the lower hypolimnion. Of these seven analytes, the differences between the sediment-water interface and lower hypolimnion were, from smallest to largest: dissolved organic carbon, 1.2 times; total zinc, 1.8 times; total cadmium, 2.7 times; total phosphorus, 3.5 times; total iron, 12 times; total lead, 19 times; and total manganese, 27 times.
U.S. Department of the Interior | U.S. Geological Survey
URL: http://pubs.usgs.gov/sir/2008/5168
Page Contact Information: Publications Team
Page Last Modified: Thursday, 10-Jan-2013 18:55:25 EST
