6

In-Lake Processes: Metals

This chapter continues the committee’s analysis of status and trends in water quality in the Couer d’Alene (CDA) Lake by focusing on the heavy metals of concern emanating from the Superfund site, particularly lead (Pb), zinc (Zn), and cadmium (Cd). In addition to revealing trends on metals concentrations in the Lake over the past 30 years, the analyses consider aspects of the time series data that are indicative of various processes occurring in the water column. Understanding the status of metals concentrations in the Lake today is an important context for interpreting trends into the future.

IN-LAKE PROCESSES RELEVANT TO METALS

Status and trends of metals in lakes are controlled by interactions among a variety of complex processes, including hydrology, internal hydrodynamics, thermal stratification, vertical movement of particles, biological productivity, and biogeochemical reactions in sediments and the water column (Davison, 1993; Nriagu et al., 1995). These processes control the distribution of metals between dissolved and particulate forms; the degree of biological uptake and release of metals by phytoplankton; the distribution of dissolved and particulate metals between surface and bottom waters; and metal(loid) trapping, rerelease, and fate in the sediments. In turn, the concentrations of metals in various compartments influence biological and ecological processes at multiple levels of biological organization and, ultimately, the kinds and productivity of microbes, plants, and animals that occupy lakes.

Figure 6-1A shows many of these processes along with the forms that metals can take in lake water, from dissolved to particulate-bound. Depending upon water and sediment photobiogeochemistry, the particulate metals can occur in multiple forms, such as in association with different ligands (e.g., iron and manganese oxides), with living organisms, or with organic detritus (Figure 6-1B). As mentioned in Chapter 5, zinc is an essential metal for life and is taken up more readily than lead by phytoplankton and other organisms (see, e.g., Chen et al., 2000). As a result, during a phytoplankton bloom zinc can be depleted from the water column by biouptake (conversion to phytoplankton and detrital particulate forms) more so than lead, and zinc is rereleased more readily in deeper waters as these organisms settle to the bottom of the lake and die. Data and studies directly addressing such processes in CDA Lake are limited. For example, only limited data are available on dissolved metal speciation (e.g., Smith et al., 2015), colloidal metal (Langman et al., 2020), or the forms of particulate metals (e.g., Balistrieri et al., 2003)—all of which affect metal uptake by phytoplankton. Rather, most metal data from the water column are determined as either “total” metal or metal that has passed through a 0.45-μm filter (and classified as “dissolved;” Chess et al., 2012). Therefore, one must infer from time series and studies in other lakes how different

processes influence trends in metal concentrations in CDA Lake (Balistrieri et al., 2002). A more comprehensive understanding of at least some processes in Figure 6-1 will be critical to evaluating metal trends and their drivers into the future.

The available metals data in CDA Lake are sufficient to consider whether a metal exists predominantly in the dissolved or particulate phase. This tendency is metal-specific and a function of particle type and concentration, pH, ionic strength, redox conditions, and temperature (Balistrieri, 1998). The relative affinity of metals for iron oxide–dominated particles in an oxidized water column follows the ranking Pb >> arsenic (As) >> Zn ~ Cd (see Chapter 7; Balistrieri et al., 2002). Thus, a given concentration of particulate zinc equilibrates with higher concentrations of dissolved zinc than occurs for lead at the same pH.

The committee found that statistically strong, significant relationships occur among dissolved and total lead and zinc where data are adequate to establish the relationship in CDA Lake (Table 6-1). Because these correlations are ubiquitous spatially and among metals, they suggest that equilibration or exchange between dissolved and particulate metals strongly influences dissolved metal concentrations in CDA Lake.

It is important to understand that “total” metal in the water column does not represent all of the metal associated with particulate material because of methodological constraints.¹ Thus, actual percentages of dissolved to total metal are lower than those found in Table 6-1. Another way to get at the true distribution of metals in the Lake is to consider, as geochemists do, the distribution between total particulate metal and dissolved metal using an operational partitioning coefficient. This is the ratio between total particulate metal concentrations in suspended

___________________

¹ As explained in Chapter 8, there are three ways of measuring metal: dissolved (filtered through a 0.45-μm filter), total (unfiltered and held at pH 2), and total particulate (metal concentration from a completely decomposed sample). CDA Lake data are reported as “dissolved” or “total” (whatever is desorbed when the unfiltered water sample is stored at pH 2).

TABLE 6-1 Percent of Total Metals in the Dissolved Form in Photic Waters in the CDA River, and in C4, C1, and C5 in CDA Lake for 2015–2020

NOTES: The entries above for the Lake are the slope of the statistically significant correlation between dissolved metal as a function of total metal for the years 2015–2020 at a given location. The entries for the CDA River are the ratio of flow-normalized flux of dissolved lead to the flow-normalized flux of total lead over these years.

material² compared to dissolved concentrations in the water column. For the purposes of illustration, the committee calculated very general partitioning coefficients by comparing the zinc and lead concentrations on decomposed particles filtered from the water column of the Lake near Harrison in June 2005 by Kuwabara et al. (2006) to mean dissolved lead concentrations between June and September 2005. The calculated partition coefficient for lead was 6,873 (mg Pb/kg divided by mg Pb/L). Reversing the ratio showed that dissolved lead concentrations were 0.014 percent of total particulate lead concentrations in that example. The calculated partitioning coefficient for zinc was 86, such that dissolved zinc concentrations were 1.1 percent of total particulate zinc concentrations. These operational coefficients illustrate that extremely large reservoirs of metal are present in the particulate form (e.g., in the bed sediments) that could replenish dissolved concentrations in the water column if conditions were conducive to metal release from the particles.

***

The following analysis of metals in CDA Lake starts with zinc, then cadmium, and finally lead. For each of these metals, there are observations about how the metal behaves, in particular its relationship to season, depth, location, and lake inflows, followed by an analysis of trends in metal concentration at each station where there are sufficient data. The trend analysis is explored in two ways. One is purely graphical—it is locally weighted scatterplot smoothing. The second is the use of a formal statistical hypothesis test for trend, the Seasonal Kendall test. Both of these methods are described in detail in Appendix B.

ANALYSIS OF TRENDS IN DISSOLVED ZINC

Of the metals examined by the committee, zinc cycling in the water column of CDA Lake is perhaps the most complicated. For each of the lake locations considered below, the raw data are shown, both in graphical form from the past five years and in tabular form for the entire dataset. This is followed by a graph of monthly averaged dissolved zinc concentration data from the photic and bottom zones to reveal seasonal patterns, and finally by a formal trends analysis of dissolved zinc concentration in photic and bottom waters over the past 26 years.

Dissolved Zinc Concentrations at C4

The C4 data shown in Figure 6-2 provide a good example of the patterns of dissolved zinc concentration in CDA Lake from 2015 to 2021. The maximum zinc concentration is always observed in the first month of the

___________________

² Total particulate metals in suspended material (total particulate metal) are the concentrations per mass of particulate determined in a completely decomposed sample (e.g., Balistrieri et al., 2002; Kuwabara et al., 2006). It is challenging to collect enough suspended material for such measurements, so total suspended particulate metal and partitioning coefficients from the water column are not commonly determined.

calendar year that samples were collected (after each winter data gap). The high concentrations in March reflect zinc inputs from the watershed and, historically, were correlated with discharge from the CDA River (CH2M Hill and URS Corp., 2001). A second consistent aspect of zinc at C4 is the decline of concentrations in the photic zone between the first sampling of the year and September.

Table 6-2 shows the concentration of dissolved zinc in the photic zone of C4, by month and year from 2004 to 2020. Declines over time are evident in the months in which there was extensive sampling. For example, dissolved zinc concentrations in the euphotic zone in 2004 were 15–34 μg/L higher than in 2020 during the same month. Table 6-2 shows that photic zone zinc concentrations are beginning, in the most recent years, to approach the Lake Management Plan target of 36 μg/L during summer stratification. However, exceedances of the 36 μg/L target still occur when the water column is mixed every spring and every fall.

The conceptual model for zinc cycling in stratified deep water lakes involves biogenic stripping (scavenging) of zinc from photic waters and recycling of zinc into bottom waters (Balistrieri et al., 2002). Figure 6-3 shows the seasonal cycling of zinc at site C4. Each year in late spring (mid-April through June), zinc in the photic zone declines rapidly at C4. This period includes the falling limb of the hydrograph (declining inputs), the beginning of stratification, and the clearing of the water column (settling of particles). The decline in photic zone zinc is partly due to trapping—settling of particles from the water column to the bottom of the Lake. Indeed, based on the analysis presented in Chapter 3 (Figure 3-21), about 53 percent of the total zinc coming into the Lake is ultimately trapped.³ A reason for the decline in photic zone zinc is that phytoplankton begin growing as waters warm, and clear and zinc is an essential element taken up by phytoplankton (Chen et al., 2000). If phytoplankton uptake exceeds zinc inputs during the growing period, the net result is lower concentrations of dissolved zinc in the photic zone via conversion of dissolved zinc to particulate zinc (living and detrital organic particles). Zinc is then rereleased below the thermocline as settling organic particles decompose or are consumed (see, e.g., Lee and Fisher, 1992; Sigg, 1985). The timing of the decline of dissolved zinc in photic waters and increase in bottom waters at C4 shown in Figure 6-3 is consistent with this model.

The bottom waters at C4 follow a seasonal pattern as well, but it is different than the pattern for the photic zone (Figure 6-3). Bottom water concentrations are more constant over the course of the year and tend to decline

___________________

³ Percent trapped is the percentage difference in flow-normalized flux between the mouth of the CDA River and the Spokane River below the Lake outlet. (470 − 219 = 251, then divided by 470 = 53%).

TABLE 6-2 Concentrations of Dissolved Zinc in the C4 Photic Zone by Month, in ug/L

NOTE: A similar table is not shown for C1. The colors follow a scale running from red for the highest values, to brown, then yellow, then green, and finally aqua for the lowest values. SOURCE: Data provided by IDEQ.

in the fall when the Lake destratifies. Zinc concentrations in the photic zone increase at that time through mixing with the enriched bottom water. Averaged over the year, the differences in concentration between bottom water and photic zone water as of 2020 are 12.8 μg/L, with bottom waters having higher concentrations of dissolved zinc (as much as 25 μg/L higher in bottom waters than in surface waters toward the end of the period of stratification). Figure 6-3 shows only the last six years of the record so that the interpretation is not overly confounded by the influence of the interannual trends present in the data.

A pH sag (pH < 7) also begins in deeper waters in CDA Lake in late spring, and pH < 7 is common in many years in bottom waters at C4. Surface water pH in most years is > 7.0 in the photic zone, but drops to as low as pH 6.1 on the bottom as the year progresses. Reduced pH is caused by organic particle decomposition where waters are isolated from oxygen inputs and carbonic acid is produced (Davison, 1993). When pH drops below 7, zinc release from settling particles could be enhanced (see Chapter 7), as could release of zinc from bottom sediments into bottom waters (Kuwabara et al., 2000). The relative importance to bottom waters of zinc release from particle settling compared to zinc release from sediments cannot be determined from existing data. The important point is that zinc release from particulate materials (either suspended or in surface sediments) into the water column in CDA Lake does not require anoxia. Release can occur under oxic conditions, especially if pH declines (also see Chapter 7).

Formal Trend Analysis of Dissolved Zinc

The trend analysis conducted by the committee shows declines in dissolved zinc concentration at C4 over time in both surface and bottom water: by 41 percent over the past 17 years, and by 33 percent over the past ten years (Figure 6-4). The implication is that the Lake is responding to the remediation (reduction) of zinc inputs discussed in Chapter 3. The rate of decline of flow-normalized dissolved zinc flux entering the Lake over the same 17-year

period is −35 percent. The close correspondence between the rate of decline of the Lake concentrations and the river inputs suggests that the water column of the Lake is responding rapidly to the declining inputs of zinc from the CDA River. Balistrieri et al. (2002) suggested that benthic flux would dominate zinc concentration later in the year, during periods of low river discharge, while influx from the watershed would control zinc concentrations in the Lake during periods of elevated river discharge. The time series data are not an adequate basis upon which to define these processes quantitatively. But if there are seasonal differences in fluxes, they do not detectably slow the overall response of the Lake to the longer-term decline in inputs.

Dissolved Zinc Concentrations at C1

Data for dissolved zinc in the photic zone at C1 are available for 2004 to 2020, but only data from 2015 to 2020 are shown in Figure 6-5. As seen in Figure 6-5, monthly geometric mean concentrations of dissolved zinc in the photic zone at C1 were highest in February through April, but the spring peak is dampened at C1 compared to C4, by 10–20 μg/L (compare Figures 6-2 and 6-5). The amplitude of the seasonal cycle is somewhat smaller at C1 compared to C4, but the temporal dynamics of the cycle are similar.

Dissolved zinc at C1 is about 8 percent less than at C4 (averaged over 2015–2020, for all months, and averaging photic and bottom waters). Concentrations in the C1 photic zone are similar to those in the C1 bottom waters in February–April, reflecting a vertically mixed lake (Figure 6-6). Concentrations of dissolved zinc in the C1

photic zone begin to decline, and bottom water concentrations begin to increase, in May, such that by September bottom water concentrations exceed photic zone concentrations by 20 μg/L. This difference narrows as the Lake mixes in October through December.

Concentrations of dissolved zinc at C1 in surface waters were similar to C4 in 2015–2020 in summer/fall. Indeed, the differences in zinc concentrations between the two sites in May through September are much smaller (average is only 1.3 μg/L) than the differences over the entire year (average is 3.8 μg/L). Thus, after inputs of zinc from the CDA River begin to slow in late spring, photic zone concentrations of zinc appear to become nearly uniform across the northern Lake, perhaps as a result of widespread mixing. Photic zone concentrations of dissolved zinc at C1 met the Lake Management Plan target of 36 μg/L from June through October from 2017 to 2020 (data not shown), falling into the window between the Lake Management Plan target and the benchmark for ecological disturbance (Chapter 9).

Formal Trend Analysis of Dissolved Zinc

The committee’s formal trend analysis for dissolved zinc concentrations at C1 shows highly significant declines over time in both surface and bottom water: by 38 percent over the past 17 years and by 30 percent over the past ten years (2.2–3.0 percent decline per year respectively; Figure 6-7). The average difference between mean concentrations in the photic zone and bottom water as of 2020 was about 12.3 μg/L, which is similar to the difference at C4 (12.8 μg/L).

In summary, the temporal and depth patterns of dissolved zinc concentrations at C1 and C4 have tracked each other for the past 17 years, with concentrations averaging about 8 percent lower at C1 than at C4.

Dissolved Zinc Concentrations at C5

Seasonal dynamics of dissolved zinc in the southern Lake (site C5) show important differences from the northern Lake (C4 and C1). The three panels of Figure 6-8 use data from 2015 to 2020 to illustrate (1) the difference between the annual cycles of zinc in the photic zone and bottom water at C5, (2) the differences between zinc in the photic zone at C4 and C5, and (3) the differences between zinc in the bottom waters of C4 and C5.

Thus, while zinc concentration in surface water at C4 peaks in January before starting to decline, at C5 the period of elevated discharge has low dissolved zinc concentrations, consistent with the proximity of C5 to unenriched inputs (Zn < 5 μg/L) from the St. Joe River. (The concentration of dissolved zinc in the St. Joe River averages about two orders of magnitude lower than in the CDA River.) Dissolved zinc in the C5 photic zone never exceeded 12 μg/L in April in any year during the period of record, in contrast, with average levels above 60 μg/L during April at C4 in many years (compare Tables 6-2 and 6-3). Low dissolved zinc concentrations (< 11 μg/L)

TABLE 6-3 Concentrations of Dissolved Zinc (in μg/L) by Month at C5

NOTE: The colors follow a scale running from red for the highest values, to brown, then yellow, then green, and finally aqua for the lowest values. SOURCE: Data courtesy of the CDA Tribe.

at C5 extend later into the highest flow years for the St. Joe River (2008, 2011, 2017, and 2018) than during drier years (e.g., 2010, 2015, and 2016). Through a combination of dilution and hydrodynamic advection from south to north, elevated discharges from the St. Joe River reduce zinc concentrations at C5.

An increase in dissolved zinc at C5 begins, on average, in May for surface waters and in April for bottom waters, as the influence of the St. Joe River wanes, although the actual month that concentrations begin to increase varies with the water year (Table 6-3). Concentrations increase progressively through August in both the surface and the bottom waters (Figure 6-8A). Dissolved zinc concentrations in bottom waters exceed those in surface waters by 30–40 μg/L by the end of the summer. The decline in zinc concentrations observed in the photic zone through

the summer at C4 is not evident at C5, and the difference between surface and bottom waters is greater at C5 than at C1 and C4. The perception that C5 has lower zinc concentrations than C4 reflects differences between the two sites in the spring season; these differences recede later in the year as processes internal to the Lake become more influential than inputs.

There are several possible reasons for the summer increase in dissolved zinc concentrations in both bottom and photic waters at C5. First, declining river discharges could allow internal processes like lake hydrodynamics to move zinc from north to south (see Chapter 4). The occurrence of contaminated sediments in the southern Lake almost to the same degree as in the northern Lake (Horowitz et al., 1995) is evidence that north-to-south advection of particles can occur. Second, decomposition of phytoplankton in bottom waters during summer can increase zinc concentrations in three possible ways. The decomposition itself can lead to release of zinc from organic particles, it can cause production of enough carbonic acid to lower pH below 7.0, and it can lower dissolved oxygen levels. In Chapter 7, the committee presents evidence of the second mechanism at work; mean monthly pH at 15 m depth goes from greater than 7 in March–April to 6.3–6.9 between April/May and November. A pH reduction to less than 7.0 could facilitate zinc desorption from settling particles and from surficial bottom sediments.

Increases in zinc in the euphotic zone in summer suggest that incoming dissolved zinc is greater than what might be stripped from the water column by phytoplankton. The coincident timing of different processes and uncertainties about hydrodynamics make it difficult to differentiate their relative roles in the increase of dissolved zinc concentrations in both photic and bottom waters at C5 through the summer/fall.

Formal Trend Analysis of Dissolved Zinc

Statistical trends over the 14 years of record (considering all months in each year through the full water column) show a decline of dissolved zinc at C5 of −30 percent over the entire period (Figure 6-9), but the vast majority of that decline takes place in the first few years. Over the period 2010–2020, the downward trend is very small (less than 10 percent) and is particularly difficult to evaluate because a substantial number of surface water observations are below the limit of detection during times of high St. Joe River discharge. The deeper water samples are mostly

above the limit of detection and appear to have no significant trend over the decade 2010–2020. The differences between median surface water and bottom water concentrations are much more pronounced at C5 (20–30 μg/L) than they are at C1 or C4 (about 12 μg/L). This suggests increased importance of internal processes as a source of zinc at C5 compared to C1 and C4, including those processes already described (benthic flux from the sediments, mixing with bottom waters from the north, and biogenic recycling).

Table 6-3 shows dissolved zinc concentrations by year and month for 2007–2020 for both the euphotic zone and the bottom waters. It shows that the bottom waters in the summer and early fall have much higher concentrations than the euphotic zone water, but that all seasons and depths show declines in dissolved zinc over the 14-year period of record. Two of the years in this period of record—2008 and 2017—had particularly high discharge on the CDA and St. Joe Rivers (as measured by annual 30-day maximum discharge). In both of these years, Table 6-3 shows lower dissolved zinc concentrations for May through November as compared to most of the surrounding years. This is an indicator that high discharge events in the St. Joe River have a moderating influence on dissolved zinc at C5 in surface and deep waters.

The substantial differences between dissolved zinc concentrations in bottom waters and euphotic zone at C5 are most evident in the evaluation of exceedances of benchmarks in recent years (2015–2020). The 36 μg/L Lake Management Plan target was met in all but two months in euphotic waters at C5 between 2015 and 2020. However, the target was not met in bottom waters in 69 percent (35 of 51) of the months when data were available, particularly April through October. In particular, exceedances continued to occur in bottom waters from May through October from 2015 to 2020 (Table 6-3).

High discharge from the St. Joe River clearly drives water column zinc concentrations down in spring at C5, but this effect is mitigated as discharge recedes, such that zinc concentrations increase throughout the summer. Although one can postulate that the source of this high dissolved zinc at C5 is internal to the lake, it is not possible to quantitatively differentiate the contributions from advection of zinc-rich water from the north; influx of zinc into bottom waters from sediments following low pH or low dissolved oxygen conditions; or biogenic recycling and release of zinc by decomposition of organic material.

***

Table 6-4 summarizes the zinc trend results for the three sites. In all cases, the trends in concentration were shown to be highly significantly (p < 0.01) downward for the period of record at all sites. For C5, the slopes of these downward trends could not be reliably defined over the past ten years because of the large number of photic zone values that were less than the limit of detection of 5 μg/L. Over the most recent decade, dissolved zinc has declined by about 30 percent at C1 and C4—locations that represent areas directly influenced by zinc inflows from the CDA River. Trends in the past decade are more difficult to quantify at C5 because many of the surface water samples had values that were less than the limit of detection of 5 μg/L. C5 is not in the direct flow path of the CDA River and is subject to internal zinc inputs that may slow the response of the Lake to changing inputs.

TABLE 6-4 Trends for Dissolved Zinc at Sites C1, C4, and C5

NOTES: See Appendix B for methods of calculation. Analysis uses both surface water and bottom water samples. The * for the 10-year period at C5 indicates substantial uncertainty about the amount of change, due to the relatively large number of euphotic zone values reported as less than the limit of detection.

ANALYSIS OF TRENDS IN DISSOLVED CADMIUM

The spatial and temporal patterns of dissolved cadmium in CDA Lake are very similar to those for zinc, and much of the process description presented on zinc will not be repeated here. Attempts were made to evaluate trends in dissolved cadmium at all five sites, but at sites C6 and SJ1 the data were all reported as less than the limit of detection (which was typically 0.1 μg/L), so no analysis was conducted for those two sites. At site C5, data below the detection limit were very common (25–50 percent of the data depending on depth and time period). Trend analyses are certainly possible for such datasets, but the complexity of determining the most appropriate statistical methods for doing so was judged to be beyond what was needed for this report. Nonetheless, in general, the bottom water samples at C5 show an indication of a downward trend over the 14 years evaluated. The records for 2004–2020 at sites C1 and C4 had no more than 10 percent of their samples below detection limits, so they were evaluated as described in Appendix B.

Dissolved Cadmium Concentrations at C4

The entire record for cadmium in the photic zone of C4 is shown in Table 6-5 as a heat map that shows the rapid declines through the late spring and summer as well as the rapid decline in the last few years of the record. Figure 6-10 shows the final six years of the photic zone data from C4. The seasonal pattern in the cadmium data is quite strong and similar to that for dissolved zinc, with the highest values occurring in the high-flow season of early spring. Indeed, the highest value in each of the years shown is always the first value in the year and the minimum is around September. The ratio of the maximum value for the year to the minimum is typically about 3:1. This seasonal pattern is related to the timing of maximum discharge for the year, with concentrations falling through the summer months as biological uptake and settling occurs.

Figure 6-11 shows the formal analysis of overall trends in dissolved cadmium at C4. The concentrations of cadmium in the photic zone average about 85 percent of the magnitude of those near the bottom. The trend in the

record shows a very sharp discontinuity, with virtually no trend from 2004 through about 2014, and then a rather steep downward trend after 2014. The trend over the full 17-year period is a change of about −49 percent. The change over the most recent ten years is also −49 percent, with most of that trend focused in the final five years of the record. The trends are all statistically highly significant (p < 0.01).

The temporal pattern of the trend in dissolved cadmium at C4 is similar to the temporal pattern of the trend in total cadmium flow-normalized flux for the CDA River near Harrison. That record shows almost no change from 2004 through 2013, followed by a steep decline to 2020 (see Figure 3-16B). At C4, the total change in dissolved

cadmium concentration for the 17 years from 2004 to 2020 was a decline of 31 percent, while the total change from 2015 to 2020 was a decline of 20 percent. The flow-normalized flux of dissolved cadmium for the CDA River shows a similar pattern, but the declines are not as steep in percentage terms as those for the flow-normalized flux of total cadmium for the CDA River (see Chapter 3). Understanding the linkage between the river inputs and ambient concentrations in the Lake is an important topic that should be pursued in the future. Clearly, the Lake cadmium trends are highly responsive to changes in the inputs, and it is reasonable to project that the Lake cadmium concentrations will continue their steep decline if the river fluxes to the Lake continue to decline as they have over the decade 2010–2020.

Dissolved Cadmium Concentrations at C1

The trends and seasonal patterns in dissolved cadmium at site C1 (Figure 6-12) are similar to those at C4. The overall trend for 2004–2020 is a decline of 41 percent, with most of the trend happening in the final decade. For both trend periods (2004–2020 and 2011–2020) the declines are highly significant (p < 0.01). The ratio of photic zone concentrations to bottom water concentrations averaged about 0.85 but was slightly larger around 2010 (a value of about 0.93).

In summary, although only the two northern Lake sites could be evaluated for trends in dissolved cadmium concentrations, the trends are clear and summarized in Table 6-6. Evaluated over the full period, 2004–2020, dissolved cadmium is declining, but virtually all of the decline takes place after about 2014. The data also show that cadmium concentrations in deeper water (> 20 m) are greater than those in surface water, by about 15 percent.

TABLE 6-6 Trends in Dissolved Cadmium at Sites C1 and C4

NOTES: See Appendix B for methods of calculation. The dark blue shading indicates that for all trend periods the observed trends are highly significant (p < 0.01). The analysis combines results from both surface and deep water.

ANALYSIS OF TRENDS IN TOTAL LEAD

The patterns in total lead concentration in the water column of CDA Lake were evaluated at three sites: C1, C4, and C5. This discussion is primarily based on the data from C4 because it is the most responsive to the river inputs of lead to the Lake, but the results at the other two sites are summarized at the end of this section.

Total Lead Concentrations at C4

Figure 6-13 shows the total lead data in surface water from C4. These data show a range of values over this 17-year period (2004–2020) of nearly three orders of magnitude. The data also are suggestive of a non-monotonic trend, rising from 2004 to sometime around 2015 and then downward to 2020. The trend is investigated in detail below. To gain some sense of the seasonality of the data, it is helpful to focus on a smaller number of years. Figure 6-14 shows only the data for 2015–2020.

An orderly pattern in the data is observed throughout the record (illustrated for 2015–2020 in Figure 6-14), starting with the highest values typically being in the late winter or spring, declining quite steeply every month, reaching an annual minimum around August or September, and rising slightly through the fall. The sampling is somewhat irregular in the months of September through March, with a few years missing values in April, May, and June. Only July and August have complete coverage over these 17 years. The irregularity of the sample collection is an impediment to data interpretation, but a number of clear patterns can be discerned in the data.

Table 6-7 is a heat map of the dataset (in the few cases where there are multiple sampling dates in the month, the heat map uses the median for the month). One of the notable things in Table 6-7 is that the months following some of the highest observed concentration values also tend to be high relative to values in that particular month in other years (e.g., 2011–2012; 2017). The final column in Table 6-7 shows the estimated annual flux of total lead for the CDA River. Generally, the highest concentration values in the table correspond to the years of highest estimated annual flux. Figure 6-15 shows the estimated monthly flux of total lead into the Lake, and it shows an obvious similarity to the pattern of concentration values in the Lake that are shown in Figure 6-14.

Figure 6-16 shows the relationship between the monthly mean values of observed surface water concentration of total lead at C4 (shown as a time series in Figure 6-14) in relation to the estimated average daily flux into the Lake for the months for which there are surface water concentration data at C4. The correlation between the

TABLE 6-7 Concentrations of Total Lead in C4 Surface Waters by Month, in ug/L

NOTES: The colors range from red for the highest values, through brown, then green, and then blue for the lowest values. The column to the right of the concentration data is the estimated annual flux values of total lead for the CDA River near Harrison in metric tons (MT)/yr. These flux values were computed using the Weighted Regressions on Time, Discharge, and Season (WRTDS) model described in Chapter 3. SOURCE: Data provided by IDEQ.

natural logarithms of these two variables is 0.88, and estimated slope of the regression of ln(concentration) on ln(monthly flux)is 0.47. This means that concentration increases in proportion to the 0.47 power of flux into the Lake for that month. More complex models of this relationship are certainly possible (ones that might include lag effects) but this simple analysis demonstrates that lead concentrations at C4 are very responsive to the current month’s inputs of total lead. The implication of this is that concentrations of total lead in the Lake should be responsive to future declines in total lead input to the Lake. The fact that estimated monthly flux of lead into the Lake explains a very large part of the variation in Lake lead concentrations is what one would expect given that 75–85 percent of the lead entering the Lake is particulate.

The surface and deep water concentrations were each modeled using a loess fit (as described in Appendix B), illustrated in Figure 6-17. The first observation is that, unlike zinc and cadmium, the lead data show much smaller differences between surface and deep water. At most, the deep water has concentrations averaging about 0.2 μg/L higher than the surface water, and for much of the record the differences are even smaller. The processes that drive the more substantial differences between surface- and deep-water concentrations of zinc and cadmium (such as potential benthic flux and biogenic cycling) have limited effects on lead. This is consistent with the low bioavailability of lead to phytoplankton (Chen et al., 2000) and the strong binding of lead to particulate material.

Another observation from these data is that the overall trend is distinctly non-monotonic, not unlike the temporal pattern of flow-normalized flux of total phosphorus at Harrison (described in Chapter 3). Over the period 2004–2020, the smoothed lake concentrations rise to a maximum in 2012 then fall from 2012 to 2020. The flow-normalized river flux rises from 2004 to 2009 and then falls to 2020. The estimated input flux record at a daily or monthly time scale could be used in conjunction with the observed Lake concentration data to calibrate a model of how the Lake lead levels respond to changing inputs. Such a model, using simplified physical process simulation with statistical parameter estimation, could be used to project future Lake concentrations as a function of various degrees of control of inputs to the Lake. What seems clear from the simple empirical analyses described in this report is that lead concentrations in the Lake are responsive to changing lead inputs at time scales of months to years. The aggregate trend over the entire 2004–2020 period, which is non-monotonic, is a change in mean concentration of −41 percent, or −3 percent per year. Even though it is a period of increase followed by decrease, the overall trend is statistically significant (p = 0.04). When evaluating the most recent ten years, the change is −74 percent, which is −13 percent per year and is highly significant (p < 0.001).

Dissolved Lead Concentrations at C4

A similar analysis of the dissolved lead concentration at C4 was also conducted (results not shown). Dissolved lead concentrations follow a similar non-monotonic trend pattern, but the rates of increase in the first part of the record and the rates of decrease in the most recent years are smaller in magnitude than those described above for total lead. The same pattern of seasonality exists, and the relationship of high winter and spring concentrations being associated with the years of high lead input also exists. Unlike zinc or cadmium, the dissolved lead annual mean concentrations are substantially lower than the total lead concentrations. Typically, the ratio of dissolved to total lead concentrations is in the range of 0.16 (also see Table 6-1).

Total Lead Concentrations at C1

At C1 and farther from the mouth of the CDA River, the patterns of total lead seasonality and trends are similar to those observed at C4 in surface waters. Figure 6-18 shows the most recent six years of surface water data. Comparing it to Figure 6-14 (which uses the same vertical scale), one can see that the general seasonal pattern is similar to C4, with the highest concentrations typically in the earliest samples in the year, but the concentrations tend to be lower (mean of all of the annual mean values is 4.8 μg/L at C4 and 1.4 μg/L at C1) and the amplitude of the seasonal cycle is reduced (mean annual difference between maximum and minimum is 19.7 μg/L at C4, compared to 4.3 μg/L at C1). Figure 6-19 shows the patterns for both surface and deep water for the full period of 2004–2020. The overall trend for the full 17-year period is a change of −11 percent (−1 percent per year), but

it is not significant (p = 0.12). For the past ten years of the record, the change is −62 percent (or −9 percent per year), and it is highly significant (p < 0.001).

Total Lead Concentrations at C5

The only other site with a dataset of sufficient sample size and sufficient numbers of values above the detection limit is C5. The record for surface water concentrations for 2015–2020 is shown in Figure 6-20. The data are intentionally shown on the same scale as is used in Figures 6-14 and 6-18 to facilitate comparison with the two northern Lake locations.

The first thing to note is that annual maximum concentrations are consistently lower than those observed at C1 or C4. For example, over the years 2015–2020, the annual maximums at C4 are in the range of 2.5–50 μg/L, at C1 they are in the range of 1.2–8 μg/L, and for C5 they are in the range of 0.7–1.5 μg/L. In addition, the seasonal pattern at C5 is very different from that of the other two sites. The difference between the annual minimum and annual maximum is only 1.55 μg/L versus 4.3 (C1) and 19.7 (C4). What one generally sees at C5 is that in many years the trend through the year is rising values from the early spring through the summer months. The explanation for this is that the spring snowmelt season at C5 is dominated by water coming from the St. Joe River, which is relatively free of lead. Only later in the year, when the water movement is driven more by internal dynamics of the Lake, does one see increases in lead as water from the CDA River and northern portions of the Lake circulate south to C5.

Another important difference between C5 and either C1 or C4, apparent in Figure 6-21, is that the difference between total lead concentrations near the surface and concentrations at the bottom are more pronounced at C5. Statistical trends (Figure 6-21) indicate that total lead concentrations in deeper water are commonly more than double the concentrations observed in surface water. The mean total lead concentrations in 2021 near the bottom of C5 are similar to C4 (~ 1 μg/L) but higher than at C1 (~ 0.5 μg/L vs. 1 μg/L, respectively). Surface water mean concentrations at C5 in 2021 are similar to C1 but lower than C4.

The committee investigated the reason for the difference in lead concentration in surface versus bottom waters at C5. Given lead’s chemistry and lack of biogenic cycling, the most likely process creating this difference is hydrodynamic advection of lead from the northern Lake, with desorption from suspended particles and/or upward transfer from the Lake bed to the water column being less likely processes. To confirm this, the committee graphed the ratio of dissolved lead concentrations (bottom water: surface water) using data from 2017 to 2020 grouped by month (see Figure 6-22). If there is desorption of lead from the bed sediment surface and sinking particulate

material, concentrations of dissolved lead in bottom waters would be higher than in surface water during the summer period of stratification. The ratio near 1 for the summer months suggests that stratification (evidenced by dissolved oxygen and temperature) does not affect lead in June through September, such that there is little dissolved lead input to isolated bottom waters either from desorption from settling particles or from the bed sediments. This is very similar to what one observes at C1 and C4. The high ratios in April and May (see Figure 6-22) must, by the process of elimination, reflect advection of bottom waters to the south from the north as discharge rates from the St. Joe River begin to decline. Advection to the south of bottom waters would have to exceed that of surface waters to explain such results.

An important distinction that should be made across the sites is that the amplitude of seasonal and interannual variability at C5 is lower than at the other two sites. Peak concentrations originating from the CDA River are more damped before they make it to C5 as compared to C1. It is possible that a substantial amount of particulate lead has settled out before it reaches C5.

The trend over time at C5 does not show the same obvious non-monotonic pattern as is seen at the other two sites, but this may be an artifact of this being a shorter record than the other two. For the period for which there are data, one can generally say that there is very little change from 2007 to about 2016, and that the last few years of the record are suggestive of a downward trend. For the full 13-year record, the downward trend is a change of about −42 percent, and for the last ten years of the record it is similar (−45 percent). Most of this decline takes place in the last five years of the record. Both of these trends are significant, with p-values of 0.001 or lower. It is reasonable to assume that the decreased inputs from the CDA River are driving these downward trends at C5, but the influence of what appear to be sources that reflect a redistribution of lead within the Lake (internal sources) add to the complexity of the response and perhaps delays the decline.

***

A summary of the total lead trends at all three CDA Lake sites is shown in Table 6-8. For the longer two records (at C1 and C4), there is evidence of non-monotonic trends, i.e., rising values up to about 2012 and then declining to 2020. The downward trends over the period 2011–2020 are 9 percent per year at C1 and 12 percent per year at C4 (greater than observed for other metals) and statistically significant. At all three sites considered, the downward trend over the past ten years is categorized as statistically highly significant. These patterns are reminiscent of the non-monotonic pattern of trend in flow-normalized flux of total lead for the CDA River near Harrison (see Chapter 3).

The committee’s hypothetical conceptual model for total lead dynamics is relatively simple in the northern part of CDA Lake (C4 and C1). Large loads of total lead enter the Lake with the rising limb of discharge in the spring, generating high concentrations in the water column. Currently, only 1 percent of the lead leaves the Lake via the Spokane River, such that most of the incoming lead is trapped in the Lake sediments. Lead concentrations in the water column progressively decline as discharge recedes and throughout the rest of the year, presumably as the 99 percent of the lead that remains in the Lake settles out of the water column to the sediments. Stratification makes little difference, as evidenced by similar total lead concentrations in surface water and bottom water

TABLE 6-8 Summary Table for Trends in Total Lead at Sites C1, C4, and C5

NOTES: See Appendix B for methods of calculation. The shading indicates the significance level of the result. Dark blue is highly significant (p < 0.01), light blue is significant (0.01 < p < 0.1), and no shading indicates a nonsignificant trend. Results are based on trends in both surface and bottom water data.

concentrations throughout the year. The spring peak in lead concentrations declines with distance away from the mouth of the river, evidenced by lower concentrations in spring at C1 than at C4. This decline is most likely driven by a combination of dilution and settling of particles with distance from the source. The dynamics of dissolved lead track total lead, but represent only 14–25 percent of the lead in the water column. Declining inputs of total lead from the watershed result in declining concentrations of total lead in the water column and declining flux to the sediments. Net lead flux from the sediments back to the water column at C4 and C1 is an undetectable fraction of the lead concentrations driven by inputs at this point in time, as evidenced by similar lead concentrations in surface and bottom waters.

The dynamics of total lead in the southern Lake are more complex. No spring peak in concentrations is observed. Discharge from the St. Joe River dilutes lead concentrations throughout the water column in spring, and then total and dissolved lead concentrations increase from May through October. Total lead concentrations in bottom waters exceed concentrations in surface waters after stratification begins, but dissolved lead concentrations are not different between bottom and surface waters for much of the summer/fall, consistent with the strong binding of lead to particulate material and a lack of benthic flux of lead from the sediments to the water column. Hence, where there are differences in dissolved lead between surface and bottom waters, it could be attributable to north-to-south advection of lead to C5 during the period of low discharge. If advection (internal redistribution) dominates the seasonal increase in lead at C5, then declines of lead inputs to the Lake will manifest at C5 in a similar way as they do at C1 and C4.

The most important conclusion to draw about the total lead trends in CDA Lake is that for the years from about 2003 to 2012, total lead concentrations were rising slowly but have declined over the past eight to ten years. The data are suggestive of an acceleration of the downward trends in the most recent five years (2016–2020), but the particular sequence of flow conditions (with 2017 being one of the highest discharge years and 2019 and 2020 among the lowest years in the record) makes the evidence rather weak. Frequent (e.g., every two years) reanalysis of the data should help to clarify if these apparent downward trends are meaningful.

The analysis also makes it clear that for C1 and C4, the highest observed total lead concentrations are directly linked to high flow conditions in the CDA River, which now deliver mostly legacy lead from past mining contamination that has been stored in the lower basin. The future trajectory of Lake concentrations will depend strongly on the extent to which this scouring of the lower basin declines. The analysis in Chapter 3 indicates that such declines are currently happening, but continued monitoring of the movement of sediment and lead and analysis of geomorphic change in the lower basin will be crucial to understanding the rate of these trends and their likely future development. Stronger inferences about the extent to which high winter and spring flows are delivering less lead than they had in past years can only come about with the accumulation of additional high-flow years and a high level of effort at river and Lake monitoring when such high-flow years take place, coupled with reanalysis of the CDA River and Lake water quality trends on a regular basis.

CONCLUSIONS AND RECOMMENDATIONS

This chapter analyzed water column data from CDA Lake over the past 30 years to reveal trends in zinc, cadmium, and lead concentrations. For many of the processes that control metals in the Lake, particularly hydrodynamic processes, benthic flux, and biogenic cycling, there were almost always insufficient data to distinguish between the processes. Based on the committee’s analysis of available datasets to better understand changes in metal concentrations in CDA Lake over the past 30 years, the following detailed conclusions and recommendations are made. These conclusions apply to the monitored areas of the Lake and cannot be extended to embayments and niches along the shoreline or other Lake locations that are poorly monitored.

1. Downward trends in dissolved zinc concentration in CDA Lake at sites C1 and C4 over the past 30 years are highly significant (p < 0.01). Furthermore, zinc concentration at these sites is decreasing at a similar rate to declines in zinc inputs to CDA Lake from the CDA River. Internal cycling of zinc (see below) has not detectably slowed the response of the northern Lake to changes in zinc input from the CDA basin.

2. Dissolved zinc concentrations have declined at C5 (in the southern Lake) over the past two decades, but the rate of decline is slower than at C1 and C4. The dynamics of zinc at C5 are more complex than at C1 and C4 and reflect spring dilution from the St. Joe River, summer/fall hydrodynamic redistribution of metals from north to south, and other internal processes (see below).
3. Dissolved cadmium concentrations at C1 and C4 declined from 2004 to 2020, with virtually all of the decline occurring after about 2014. Like zinc, cadmium trends in the northern Lake are highly responsive to changes in cadmium inputs from the CDA River.
4. Seasonal differences in dissolved zinc and cadmium concentrations between surface and bottom waters suggest there is an internal source of both metals to bottom waters during periods of stratification. For zinc, after the onset of stratification, dissolved zinc concentrations decline in surface waters at C1 and C4 but increase in bottom waters. At C5 in summer, zinc concentrations in bottom waters increase faster than in surface waters, such that bottom-water zinc concentrations at C5 eventually equal those at C4. Possible bottom-water sources of zinc include desorption from sinking particles and the surface layer of bed sediments, decomposition of settling organic particles, or hydrodynamic redistribution (at C5). For cadmium, concentrations in deeper water (> 20 m) are greater than those in surface water, by about 15 percent, suggesting a role for internal processes (as for zinc). For both metals, there are insufficient data to quantify the relative importance of these processes.
5. From about 2003 to 2012, total lead concentrations in CDA Lake rose slowly, but they have declined over the past eight to ten years. Most lead enters the Lake during periods of high discharge, such that the future trajectory of Lake lead concentrations will depend strongly on the extent of scouring of legacy sediment from the bed and banks of the CDA River. Despite thermal stratification in summer, there were only small differences between total lead concentrations in surface and bottom waters at C1 and C4. This suggests minimal flux of lead from internal sources and is consistent with strong lead binding to particulates and the lower bioavailability to phytoplankton of lead compared to zinc and cadmium. At C5, hydrodynamic redistribution of lead from north to south is the most likely process to play a role in lead dynamics, with desorption of lead from suspended particulates settling to the bottom or from bed sediments of limited importance.
6. Extremely large reservoirs of metal are present in particulate form (in suspension and in the Lake bed sediments) that could replenish dissolved concentrations in the water column whenever conditions are conducive to metal release from the particles. Desorption, decomposition, and hydrodynamics (especially at C5) all can slow rates of decline in water column metal concentrations in lakes as inputs from the watershed decrease. To date, that does not appear to be the case north of the CDA River confluence (at C1 and C4), but those processes could be responsible for the slower rate of decline of cadmium, lead, and zinc at C5. Detailed understanding and monitoring of the processes that influence interactions between particulate metal and dissolved metal are critical to evaluating future changes in metals concentrations in the Lake as inputs from the watershed continue to decline.

REFERENCES