5

In-Lake Processes: Dissolved Oxygen and Nutrients

This chapter continues the committee’s analysis of trends in water quality in Coeur d’Alene (CDA) Lake by focusing on dissolved oxygen, the nutrients phosphorus and nitrogen, and parameters indicative of the Lake’s productivity such as chlorophyll a. The chapter also specifically addresses the bullet in the statement of task that asks whether reduced levels of zinc entering the Lake as a result of the upgrade to the Central Treatment Plant and other upstream activities are removing an important control on algal growth.

EUTROPHICATION, PRODUCTIVITY, AND OXYGEN DEPLETION

Like other ecosystems, lake ecosystems are driven by the inputs and cycling of essential chemical elements that are often supplied at insufficient rates to support growth. These are “limiting nutrients.” Among the biologically important elements, nitrogen (N) and phosphorus (P) are most commonly identified as limiting nutrients in lakes (Elser et al., 2007), although there remains discussion (Conley et al., 2009; Schelske, 2009; Schindler and Hecky, 2009) about the timescales over which nitrogen can be limiting due to potential compensation by nitrogen fixation (see below). Overall, elevated nutrient inputs (“loading”) accelerate the slow natural processes by which lakes increase in primary productivity (in terms of internal carbon fixation by phytoplankton, periphyton, and macrophytes) over time, leading to the generally undesirable outcome of anthropogenic eutrophication. A eutrophied lake has diminished water clarity due to accumulation of phytoplankton biomass in the water column. Depending on conditions, the phytoplankton community of a eutrophic lake can become dominated by undesirable species such as cyanobacteria, which can produce chemical compounds that are toxic to aquatic life as well as to pets and humans. Eutrophication can also lead to anoxia.

Decreases in Dissolved Oxygen

Elevated production of phytoplankton biomass due to increased nutrient loading can often lead to depletion of oxygen in the bottom waters of a lake, which occurs when organic matter produced by nutrient-driven phytoplankton sinks into the bottom layers and is decomposed by bacteria that consume oxygen. Respiration in benthic sediments also consumes oxygen and leads to its depletion in bottom waters. The term “hypoxia” is used to describe situations in which oxygen is depleted below saturation levels, where saturation refers to the maximum amount of dissolved oxygen that water can hold as a function of temperature and pressure. Hypoxia is usually defined

as a depletion that is detrimental to organisms of interest. Dissolved oxygen concentrations that are considered hypoxic are in the range of 2–4 mg/L (Diaz and Rosenberg, 2008; Howell and Simpson, 1994; Paerl et al., 2006); obviously, the exact concentration below which dissolved oxygen has negative effects varies depending on the organism of interest, the environment, and the duration of exposure (e.g., Hrycik et al., 2017; Tellier et al., 2022). In practice, hypoxia is often defined in terms of measurable consequences reflected in the ecosystem, such as the oxygen concentration at which fisheries start to collapse (Renaud, 1986) or a particular biological function becomes impaired (Diaz and Rosenberg, 1995).

The term “anoxia” refers to situations when oxygen is reduced to very low levels (undetectable or < 1 mg/L; Nürnberg, 2002). At these low dissolved oxygen levels, not only is aquatic life (fish, mollusks) threatened but chemical conditions at the sediment–water interface can be altered, releasing nutrients back into the water column and creating a highly undesirable self-amplifying loop of eutrophication. Specifically, when oxygen levels at the sediment–water interface decline low enough (to < 1 mg/L) to allow redox potential to fall below ~200 mV (Mortimer, 1942, 1971), oxidized forms of iron (“ferric”) are reduced to their “ferrous” form by iron-reducing microbes and are solubilized. This dissolution of iron releases the PO₄-P that was bound to the oxidized ferric compounds. Release of other metals from the sediments back to the water column can also be enhanced due to development of anoxia (see, e.g., Atkinson et al., 2007). These processes are of high relevance to CDA Lake given concerns about how future eutrophication might affect the legacy of metals stored in the Lake sediments.

Nitrogen Cycling

Nitrogen in lakes can be found in various chemical forms, both organic (i.e., associated with a carbon-based molecule) and inorganic. Organic nitrogen is found in living things (bacteria, phytoplankton, zooplankton, fish) where it is an essential component of proteins, nucleic acids, and other molecules. Organic nitrogen can also be found in nonliving forms (including detritus particles) or dissolved molecules derived from and processed by microorganisms. The dominant inorganic forms of nitrogen are nitrate (NO₃) and ammonia/ammonium (NH₃/NH₄) with nitrite (NO₂), usually at low concentrations. Also present in the water is dinitrogen gas (N₂).

Nitrogen has a relatively complex cycle driven largely by specialized microorganisms that use nitrogen not only to build their biomass (e.g., to make protein or nucleic acids) but also as a source or a receiver of electrons during metabolic processes. The cycle can be thought to begin via “nitrogen fixation,” in which certain bacteria (of special importance in lakes) and cyanobacteria convert N₂ gas into biologically useful forms via large expenditures of energy. Once in organic form, this nitrogen can be released into inorganic form as NH₃/NH₄ in a process called “ammonification,” when organisms are eaten or their biomass is decomposed. In turn, ammonia can be chemically converted to nitrate (via nitrite) by certain bacteria in a process called “nitrification.” Both nitrate and ammonia can return to the cycle by being re-assimilated by phytoplankton or heterotrophic microbes for use in biomass production. Finally, nitrate can be used as an electron acceptor for respiration when oxygen is depleted, resulting in production of N₂ gas. This process is called “denitrification” and largely completes the cycle. (Note that the greenhouse gas nitrous oxide [N₂O] can be formed during denitrification and nitrification. There are other processes of nitrogen cycling that can occur, but for the sake of relative simplicity, they are not described here.)

Fixation of N₂ from the atmosphere is not the only way that nitrogen can enter a lake’s water column. Transfer of water from the watershed or other regions of the lake can carry nitrogen in various forms into the lake. Most nitrogen loaded to lakes and rivers in undisturbed catchments originates from nitrogen fixation by symbiotic bacteria–plant root associations in the surrounding forests and grasslands. In disturbed catchments, nitrogen from agricultural runoff or point source discharges is often the dominant source. Nitrogen, especially as NH₃/NH₄, can be released from sediments and reenter the water column. Nitrogen can also enter a lake from the atmosphere, in wet deposition as NO₃ or NH₃/NH₄, in gaseous transfer as NH₃/NH₄, or in particulate form as organic nitrogen (pollen, eroded soil). Finally, denitrification is not the only vector by which nitrogen can leave the water column. It can also be assimilated by periphyton, settle to the lake bottom attached to particles, or be discharged from the lake via its outflow.

Phosphorus Cycling

Like nitrogen, phosphorus in lakes can be found in various organic and inorganic forms. Whether in free dissolved form or in organic molecules, phosphorus is generally bound to four oxygen atoms as phosphate (PO₄). Organic phosphorus in living things includes nucleic acids, adenosine triphosphate (ATP), and membrane lipids. As with nitrogen, organic phosphorus can be found in nonliving detritus or as dissolved molecules derived from and processed by microorganisms. By far the dominant inorganic form is phosphate, but under many conditions, concentrations of phosphate are too low for reliable detection. Inorganic forms of phosphorus are also associated with inorganic particles, such as dust, glacial flour, clays, or eroded soil, often in mineral complexes with calcium, iron, and aluminum.

Compared to nitrogen, the cycling of phosphorus is simple. Phosphorus enters a lake via river and stream inflows and atmospheric deposition, although the direct biological availability of that phosphorus can vary and depends on its chemical form. Phosphorus can also enter a lake water column via processes of “internal loading” or release of phosphorus from sediments—a process that is greatly amplified when oxygen concentrations at the sediment–water interface are depleted to less than 1 mg/L. Once in the lake, cycling is dominated by assimilation of PO₄ by phytoplankton and microbes; release of dissolved forms of phosphorus; mineralization of organic phosphorus back to PO₄ by heterotrophic microbes, zooplankton, and fish; and re-assimilation. Under phosphorus-limited conditions, these internal cycling processes can be extremely fast (on the order of minutes), and phosphorus atoms cycle many times within water column biota before being lost via sedimentation or discharge from the lake.

Interactions among Elements, Including Metals

Biogeochemical cycles of nitrogen and phosphorus do not operate independently of each other. For example, nitrogen and phosphorus are used to produce phytoplankton and bacterial biomass within a relatively narrow range of stoichiometric N:P ratios. Suspended organic matter in lakes has an average N:P ratio of ~20–30:1 (atomic) (Sterner et al., 2008), with variation driven by the role of nitrogen or phosphorus limitation in constraining growth (Sterner and Elser, 2002). The cycling of nitrogen and phosphorus are not independent of other elements, especially various metals that play important biological and geochemical roles. For example, the trace elements iron (Fe) and molybdenum (Mo) are essential for biological nitrogen fixation as well as other processes. Metals also play a role in modulating nutrient availability, as when oxidized (ferric) forms of iron form insoluble ferric hydroxides under aerobic conditions, co-precipitating phosphate from the water column. In turn, depletion of oxygen in sediments and at the sediment–water interface can reduce iron to its soluble ferrous form, releasing the co-precipitated phosphate and contributing to internal nutrient loading (see Chapter 7 for details).

ANALYSES OF DISSOLVED OXYGEN TRENDS

The committee analyzed dissolved oxygen data collected in CDA Lake by the U.S. Geological Survey (USGS), the Idaho Department of Environmental Quality (IDEQ), and the CDA Tribe over the past 30 years. The dissolved oxygen vertical profile datasets were first examined for evidence of trends over time at the four major sampling stations in the Lake (C1, C4, C5 and C6—see Figure 2-1). The committee also conducted a time series analysis for evidence of any long-term seasonal trends in dissolved oxygen concentrations in CDA Lake. Finally, the committee performed a Mann-Kendall test of trends in the monthly-averaged dissolved oxygen data in bottom waters at all four sites.

To conduct these analyses, it was necessary to standardize the profile data with respect to sample collection methodologies, sample depth, and temporal frequency because the available data were quite heterogeneous with regard to these dimensions. For example, data collected during sonde upcasts were discarded, the profile data were interpolated and binned by 1-meter and monthly increments, missing maximum depth data were estimated from available adjacent depths, and diurnal fluctuations in dissolved oxygen were confirmed to be small¹ (see Appendix B for greater detail).

___________________

¹ The committee hypothesized that diurnal variations in dissolved oxygen in CDA Lake would be small because of surface water mixing and because of the absence of light penetration in the hypolimnion, where anoxia is of concern. This was confirmed by analyzing buoy data on dissolved oxygen at C5 on eight separate days during July–August 2011 and 2015. For both the topmost (1 m) and bottommost (17 m) samples at C5, the average standard deviations of dissolved oxygen data were 0.074 mg/L and 0.34 mg/L, respectively.

Observations of Dissolved Oxygen Data

The first analysis compared the dissolved oxygen percentiles for the CDA Lake sampling stations C1–C6 based on all of the raw profile data collected from 1991 to 2021. As shown in Figure 5-1, there was little evidence of low oxygen concentrations at the C1, C2, C3, and C4 stations. There was some evidence of low oxygen concentrations at the C5 station and clear evidence of depressed dissolved oxygen at the C6 station. However, because these percentile distributions are based solely on extant data, and these data were not always collected in a systematic manner with regard to time and depth, this is only a preliminary analysis. For example, in the earlier years, dissolved oxygen depth profiles were often collected at a finer spatial resolution (e.g., 1 or 2 m) in the upper water column and at a coarse resolution (e.g., 5 m) in the hypolimnion. Because the deeper depths were sampled with a coarser resolution, the frequency of low dissolved oxygen concentration in these percentile estimates was probably somewhat underestimated.

The second analysis characterized the within-year and depth variability at the C1, C4 and C6 stations. These stations were the focus of these analyses because C1 and C4 were the deeper stations that had the most data (compared to C2 and C3) and the C6 station had the clearest evidence of seasonal dissolved oxygen depletion. For station C1, it is evident that the main mode of variability for the intra-annual data is time (Figure 5-2). Although not shown, very similar results were obtained for C4. Higher oxygen concentrations are observed throughout the water column during the winter and spring, and lower concentrations are observed during the summer and fall, when CDA Lake is warmer at the surface. For example, both C1 and C4 dissolved oxygen concentrations averaged 11.2 mg/L from December to May and 8.7 mg/L from July to November. There is also a slight tendency for lower dissolved oxygen concentrations in the hypolimnion in July–November at both stations. For example, oxygen concentrations averaged 9.0 mg/L in the upper 20 m of the water column and 8.2 mg/L at depths of ≥ 31 m during this time.

Finally, at both stations there is evidence suggesting a deep chlorophyll maximum, and associated higher dissolved oxygen concentrations, as evidenced by the spike in dissolved oxygen concentrations in the metalimnion (8–14 m) during mid-July. However, the chlorophyll data were not granular enough to enable the committee to determine which stations and/or sampling dates had a deep chlorophyll maximum present.

In contrast to C1 and C4, the vertical dissolved oxygen profiles for C6 showed much clearer evidence for hypoxic and even anoxic conditions during the summer stratified period (Figure 5-3). From July to mid-September,

dissolved oxygen concentrations only averaged 1.7 mg/L, at depths of 8–11 m, whereas dissolved oxygen concentrations averaged 8.6 mg/L in the upper 5 m of the water column during this same time.

The committee also examined the buoy data collected by the CDA Tribe at station C5 (data not shown). These data were available at a very high temporal resolution (i.e., four or six profiles per day) for most of the period from July to November for the years 2011 and 2015. Buoy data are also available for 2019, but oxygen data are not available for that year. Based on a comparison of the oxygen percentiles for 2011 and 2015, it is evident that on average the two years had similar dissolved oxygen concentrations, but the lowest concentrations were lower during 2015. For example, the bottom quartile of data had an average of 1.5 mg/L lower dissolved oxygen concentrations during 2015. Because 2015 had a tendency for lower dissolved oxygen concentrations, the committee also analyzed the buoy data collected by the CDA Tribe at C5 for the entirety of the temporal and depth distribution. This analysis showed that dissolved oxygen concentrations were lowest below a depth of 15 m during the second half of September and all of October. During this time and at these depths, dissolved oxygen concentrations averaged 5.0 mg/L.

As shown in Tables 5-1 through 5-4, the month-by-year matrix of observed dissolved oxygen concentrations was incomplete for the C1, C4, C5, and C6 sampling stations.² For example, the data availability at C1 for the months of November to February was only 24 percent, which increased to 79 percent during the months of July

___________________

² These four paragraphs, and text throughout the report, were edited after report release to reflect corrections to Table 5-3, the previous version of which had data inadvertently shifted by two months.

to October. Additionally, the data availability during late summer and early fall has been good since 1995, and data availability in the spring and early summer has improved substantially since 2004.

The average monthly dissolved oxygen concentrations for the C1 station showed clear evidence of modest monthly and depth patterns in the dissolved oxygen concentrations (Table 5-1). A comparison of the monthly average oxygen concentrations in the upper 10 m of the C1 station to the bottom 10 m (i.e., 31–40 m) showed that the entire water column at this station had higher oxygen concentrations during the winter and spring and lower concentrations during the summer and fall (Figure 5-4). Surface and bottom dissolved oxygen concentrations were similar from December to August, and peaked at ≈ 12.0 mg/L in March and April. In October and November, bottom oxygen concentrations averaged 1.9 mg/L less than surface concentrations.

Similar to the C1 station, the long-term monthly mean oxygen concentrations for the bottom of station C4 (30–39 m) showed strong seasonality (as expected) (Table 5-2). Additionally, the average dissolved oxygen concentration in the surface and bottom waters of station C4 were similar during the months of December through August, and peaked at about 11.6 mg/L from January through April (Figure 5-5). During October and November, dissolved oxygen concentrations were about 1.7 mg/L lower on the bottom of station C4 compared to the surface.

Station C5 is considerably shallower than stations C1 and C4, and its seasonal dissolved oxygen dynamics are also substantially different from the deeper stations in several regards (Table 5-3 and Figure 5-6). In particular, the seasonal oxygen depletion evident in the bottom depths (i.e., 14–17 m) was more pronounced and subsided one month earlier in the year at this station. From November to June, monthly dissolved oxygen concentrations at

TABLE 5-1 Long-Term Monthly Averaged Dissolved Oxygen Concentrations (mg/L) for the Bottom 10 m (i.e., 31–40 m) at the C1 Sampling Station

Note: Red shading denotes less favorable conditions (e.g., low dissolved oxygen), while blue shading denotes more favorable conditions (e.g., high dissolved oxygen). The coloring, created in Excel, is not consistent across Tables 5-1 to 5-4.

TABLE 5-2 Long-Term Monthly Averaged Dissolved Oxygen Concentrations (mg/L) for the Bottom 10 m (i.e., 30–39 m) at the C4 Sampling Station

Note: Red shading denotes less favorable conditions (e.g., low dissolved oxygen), while blue shading denotes more favorable conditions (e.g., high dissolved oxygen). The coloring, created in Excel, is not consistent across Tables 5-1 to 5-4.

TABLE 5-3 Long-Term Monthly Averaged Dissolved Oxygen Concentrations (mg/L) for the Bottom 4 m (i.e., 14–17 m) at the C5 Sampling Station^a

Note: Red shading denotes less favorable conditions (e.g., low dissolved oxygen), while blue shading denotes more favorable conditions (e.g., high dissolved oxygen). The coloring, created in Excel, is not consistent across Tables 5-1 to 5-4.

^a Table 5-3 was replaced after report release to correct a formatting error in which the data had been inadvertently shifted by two months.

TABLE 5-4 Long-Term Monthly Averaged Dissolved Oxygen Concentrations (mg/L) for the Bottom 3 m (i.e., 8–11 m) at the C6 Sampling Station

Note: Red shading denotes less favorable conditions (e.g., low dissolved oxygen [DO]), while blue shading denotes more favorable conditions (e.g., high DO). The coloring, created in Excel, is not consistent across Tables 5-1 to 5-4.

14–17 m averaged ≈ 0.6 mg/L lower than the corresponding concentrations in the upper 5 m of the water column, and during August through October the bottom concentrations averaged ≈ 3.1 mg/L less than the surface concentrations. The deep water dissolved oxygen concentrations at C5 rebounded in November when these concentrations averaged 10.6 mg/L, which was ≈ 4.4 mg/L higher than the preceding month. In contrast, the bottom dissolved oxygen concentrations at C1 and C4 rebounded one month later in December.

Compared to stations C1, C4 and C5, the data availability for station C6 is more limited (Table 5-4). This site has detailed oxygen profile data from 1991–1992, and then again from 2004 to the present. This site also has the most severe seasonal dissolved oxygen depletion in its bottom waters, with concentrations routinely averaging well below 1 mg/L at 8–11 m depth during the month of August, and concentrations only averaging 2.9 mg/L in July and September. During these months, the bottom dissolved oxygen concentrations also averaged 7.0 mg/L lower than the surface concentrations (Figure 5-7). Similar to station C5, dissolved oxygen concentrations rebounded rapidly in the fall, albeit in October (instead of November as was the case for C5, or December as for C1 and C4). The mean monthly bottom dissolved oxygen concentrations were also much more variable at this station, with large outliers within months and dramatic changes in concentrations from one month to the next. Given its unique hydrodynamics (discussed in Chapter 1), oxygen concentration trends at the C6 station are of unclear relevance to the main body of the Lake as represented by the C1–C5 sampling stations.

As expected, there are obvious seasonal patterns in the monthly dissolved oxygen concentrations at both the surface and bottom of CDA Lake. The differences in the surface and bottom dissolved oxygen concentrations were usually quite small at the deeper profile sampling stations (i.e., C1 and C4). The bottom water dissolved oxygen concentrations at the C6 station were very low during the late summer and of concern for fisheries habitat and for redox reactions at the sediment–water interface, which would cause both nutrients and dissolved metals to be released from the Lake sediments.

In general, the available dissolved oxygen data for CDA Lake are quite patchy, but data coverage has improved markedly since 2004. Winter data are particularly sparse, but this time of the year is of less concern since dissolved oxygen concentrations are consistently high during these months. To improve future time series analyses, it would be best to sample every other winter month on a consistent basis.³

The analyses reported herein used a coarser sampling resolution for the bottom samples than is ideal because the most important information is the oxygen concentration immediately above the sediment–water interface. However, there was some variability in the maximum depth sampled from one vertical profile to another, especially during the earliest years of this time series. Some of this variability may be due to intra-annual and interannual changes in the surface level of CDA Lake,⁴ which would in turn affect that maximum depth from which oxygen data can be collected. When inferring redox conditions at the sediment–water interface for these profiles, it would be beneficial to know the actual bottom depth associated with each profile so that the bottom dissolved oxygen data can be presented relative to their distance from the sediment–water interface.

In early 2022, the committee was shown data suggesting increasing trends in anoxia at C6 and SJ1 (Chess, 2022). Figure 5-8 shows the number of days each year that the hypolimnion was anoxic (defined as dissolved oxygen concentrations < 1 mg/L) at the C6 and SJ1 sampling sites for the 2011–2021 time period. Whether this plot actually indicates a declining oxygen trend is confounded by the fact that 2011 was an anomalously high year for hypolimnetic oxygen concentrations at the C6 sampling site. Table 5-4 shows the 8–11 m dissolved oxygen concentration at C6 averaged 3.7 mg/L in August of 2011, but 0.7 mg/L in August for all years at this sampling station (n = 19). The year 2011 was also anomalous because it was an unusually high flow year for CDA Lake. In addition, the hypolimnetic anoxia analysis shown in Figure 5-8 started in 2011 (instead of 2004 when consistent profiling at this station started) because a switch to twice monthly profiling occurred in 2011. To test whether the significant trend for days of hypolimnetic anoxia at C6 was simply due to starting this time series at an anomalously high year, the committee showed that had this time series been started in 2012 instead, the trend in increasing days of hypolimnetic anoxia at C6 would not be significant.

___________________

³ This paragraph was edited after report release to remove duplication with the section on the Mann-Kendall DO trend analysis.

⁴ The committee assumes that the level of CDA Lake varies by ca. ± 1 m within a year and probably less from one year to another. These data presumably exist for the outflow of the Lake, since that controls the flow of water into the Spokane River.

Mann-Kendall Dissolved Oxygen Trend Analysis

The committee conducted a formal statistical analysis of all of the bottom-water monthly average data presented in Tables 5-1 through 5-4. The analysis is structured in a manner such that trends over all months are evaluated, but results for each of the individual months can also be considered. The goal of this analysis was to determine if there is an ongoing trend in the direction of anoxia, particularly summertime anoxia, in the Lake. The datasets included in this analysis cover calendar years 1995–2020 plus the months January–March of 2021 at C1 and C4. There are data for 1991–1992 at some of the sites, followed by a gap of several years. Given that there are some concerns about the comparability of the 1991–1992 data to the later years, the committee deleted the 1991–1992 data from this analysis, although an alternative analysis was run with these data included and the results did not change substantially (results not shown).

For each of the four sites, the data were evaluated by doing individual Mann-Kendall trend tests for the time series for each of the 12 months (provided that there were more than three observations in that month during the total record). The significance level was evaluated and categorized as follows. For p < 0.01, the trends were considered highly significant; for 0.10 > p > 0.01, they were considered moderately significant; and for p > 0.1, they were considered not significantly different from zero. The Mann-Kendall test was selected for three reasons. It is robust against the influence of a few data points that are widely separated in time from the main body of the data. It is robust against strong asymmetry of the data. It can be combined readily into an overall test of trends across all months (the Seasonal Kendall test described by Hirsch et al., 1982) with a minimum set of assumptions (i.e., there is no need to specify a model of seasonality), but it is resistant to the seasonal differences in the distributions.

For each of the months, the slope of the trend was evaluated using the Theil-Sen slope estimator (Sen, 1968; Theil, 1950), which is closely related to the Mann-Kendall test (such that the sign of the Mann-Kendall test statistic is never the opposite of the Theil-Sen slope) and is highly resistant to the effect of outliers in the data. The results of all of the Mann-Kendall trend tests, for all months and each of the four sites, are shown in Table 5-5. The Seasonal Kendall test (and associated Seasonal Theil-Sen slope) sums the results of the individual monthly results and evaluates the statistical significance of this sum; it also calculates a trend slope across all months. These Seasonal Kendall results are given in the last column of Table 5-5.

The following are some observations about these results. First, for sites C1 and C4, all 22 of the monthly trends that could be evaluated (regardless of significance) are positive and none are negative.⁵ Five of those 22 were moderately significant upward trends and none were highly significant. However, when aggregated over the months using the Seasonal Kendall test, both C1 and C4 show highly significant upward trends in dissolved oxygen.

___________________

⁵ Text here and throughout the report was modified after report release to more precisely denote where the trends were observed.

TABLE 5-5 Trends in Monthly Mean Dissolved Oxygen in Bottom Water at C1, C4, C5, and C6, 1995–2020^a

NOTES: The values shown are the estimated slope of the trend in mg/L/yr. Months marked with an asterisk are those with three or fewer monthly mean values. Data are from 1995 to 2020 except for January, February, and March at C1 and C4, which extend to 2021. The color code relates to the significance level of the observed trend. Dark blue indicates a highly significant upward trend (p < 0.01), light blue indicates a moderately significant upward trend (0.10 > p > 0.01), no shading indicates results are not significantly different from zero (p > 0.10), and pink indicates a moderately significant downward trend (0.10 > p > 0.01). There were no highly significant downward trends (p < 0.01). The last column, labeled “All,” is the result of a Seasonal Kendall test on the data from all months. The color coding uses the same assignments as are used for the monthly results.

^a Table 5-5 was edited after report release to reflect corrections to Table 5-3, the previous version of which had data inadvertently shifted by two months.

At C5, the trends across the 11 months that could be tested were all positive. Of these, four were highly significant and three were moderately significant. The overall result using the Seasonal Kendall test had a highly significant upward trend, which was much larger in magnitude than the trends at C1 or C4.

The data for C6 were much more limited than the data for the other sites. Only nine months could be evaluated. Of these, seven had positive slopes and two had negative slopes (July and August). Of these negative slopes, July was not statistically significant and August was moderately significant. The overall result using the Seasonal Kendall test shows a moderately significant upward trend in dissolved oxygen. The downward trends indicated for July and August are important because they come at the time of year with the lowest dissolved oxygen levels.

In summary, it is difficult to make a general case that there has been deterioration in dissolved oxygen conditions in CDA Lake; rather, the data suggest the contrary conclusion. Of course, these data do not include the lateral lakes, which are known to have hypoxic or anoxic conditions at times. Unfortunately, there are no long-term dissolved oxygen observations in the lateral lakes that would support a similar trend analysis.

These results, coupled with the trend results for phosphorus and nitrogen in the Lake (see below) and entering the Lake (see Chapter 3), indicate that low dissolved oxygen is not expected to become a problem in the main body of the Lake if current trends continue. It will be important to continue to collect dissolved oxygen data, and analyze it every two years, in order to detect changing conditions.

IN-LAKE NUTRIENT ANALYSES

The committee analyzed both phosphorus and nitrogen data from CDA Lake to determine if there are any trends over time that might indicate worsening water quality. The committee plotted nutrient data from five locations in CDA Lake, it created loess smooths of the surface water and bottom water data sets, and it conducted Mann-Kendall tests of trends in the monthly averaged nutrient data at all five sites.

Trends in Total Phosphorus

For total phosphorus (TP), there are regular long-term data available for five sites (C1, C4, C5, C6, and SJ1). For C1 and C4, the data used here cover 2004–2020, and for C5, C6, and SJ1, the data cover 2007–2020. The data at all five sites show strong seasonal patterns, but these patterns are different for the different sites. The data for surface water are displayed in Figure 5-9. Only surface water samples were considered because their nutrient values can be expected to be important drivers of photosynthesis due to light availability.

At both C1 and C4, the highest total phosphorus values in the six years shown are in 2017, which is the year with the highest inflow. In general, the values decline from the first sample of the year at least through the end of the summer at both sites and all years—a pattern that is very similar to the total lead pattern at these sites (see Chapter 6). Both total lead and total phosphorus have a large component carried in the particulate fraction and are thus both responding to the pattern of inputs from the CDA River. Both of these sites show a strong decline

after 2017, with the total phosphorus values in 2019 and 2020 having much lower concentration than the prior years. The seasonal pattern at C5 is similar to C1 and C4. The maximum value in 2017 is not as large at C5, but all of the years show declines over the course of the spring and summer.

It is reasonable to ask the following question: might those low values in 2019 and 2020 simply be an artifact of two relatively dry years? Evaluating the inflow records for the January–May period of every year for the CDA River, one sees that the January–May period of 2019 was the third lowest in the past 26 years of record, and January–May of 2020 was the eighth lowest in the past 26 years. Thus, these two years were low flow years but not extremely low years. Furthermore, the total phosphorus concentrations in the CDA River observed in 2019 and 2020 were very low when compared to the other years that are also in the lowest third of the records for the past 26 years. Thus, the committee would argue that these two low concentration years are not simply an artifact of these being low flow years. Nevertheless, seeing this decline in concentrations at the end of the period of record makes it critical that frequent updates of analysis of these records be done (say every other year) to see if this strong departure from previous conditions persists.

The data suggest that river inputs drive the high total phosphorus concentrations of each year at each of these three sites, followed by some combination of biological uptake and particle settling driving the declines from the early peaks. The absence of a sharp decline at C5 for 2019 and 2020 is likely related to the differences in the history of total phosphorus flux for the CDA River versus the St. Joe River. Indeed, the flux of total phosphorus for

the CDA River for the years 2018, 2019, and 2020 was 36, 15, and 17 percent, respectively, of the same flux in 2017. In contrast, the St. Joe River fluxes for 2018, 2019, and 2020 were 70, 32, and 30 percent, respectively, of the 2017 value. The discharges for these years were very nearly the same across the two watersheds. This suggests that the impacts of the mining remediation could have been driving the declines in the CDA River and hence driving the declines at C1 and C4. This is further evidence that the Superfund remedy is having a beneficial impact on the phosphorus levels in the northern parts of the Lake. At sites C6 and SJ1 there are no regular seasonal patterns in the total phosphorus concentrations.

Figure 5-10 shows a set of plots for the entire total phosphorus record at all five sites, distinguishing between surface water (or photic zone) and deeper water. The Seasonal Kendall trend results are summarized in Table 5-6. The ten-year trends (or full period of record trends—whichever is longer) are estimated using the Theil-Sen slope estimator. In the cases of sites C6 and SJ1, the change and significance levels are estimated separately for surface water and for bottom water because their trends are distinctly different.

The results for sites C1 and C4 are very similar. The estimated change over the past decade is a decrease of 15 percent at C4 and 25 percent at C1 (the p value is about 0.22). Thus, the evidence favors the conclusion that concentrations have declined over this period, but the statistical support for a ten-year decline is weak. Both C1 and C4 show a steady rise in total phosphorus concentrations from the start of the record until about 2016 (Seasonal Kendall trend calculations for C1 or C4 show a significant upward trend during this period). However, this was followed by a decline from 2016 to the end of 2020. The flow-normalized flux record for the CDA River near Harrison (see Chapter 3) also shows a pattern of rising values followed by falling values, but the timing of the reversal in that record comes at an earlier time (around 2010). Finally, surface water and bottom water concentrations are very similar throughout the C1 and C4 records.

The increase in phosphorus concentrations at C1 and C4 up until 2016 might explain conclusions (e.g., IDEQ and CDA Tribe, 2020) that phosphorus concentrations in the Lake have been increasing. But the newer data change the sign of the decadal trend, and unless there is yet another reversal in the record, decadal trends will soon indicate a highly significant downward trend in total phosphorus concentration. Ten years is a short window in which to interpret meaningful trends in something as complicated as total phosphorus, so continued monitoring and timely syntheses are essential to separating stochastic variability from actual trends.

In contrast to C1 and C4, the trend at C5 is steadily downward from the start of the record in 2007 to the end in 2020 (a decrease of 33 percent, with p < 0.001). This is in keeping with the observed trend in flux for the St. Joe River (see Chapter 3). The concentrations at C5 are typically slightly higher than those at C1 or C4, and this is also consistent with the somewhat higher total phosphorus concentrations for the St. Joe River versus the CDA River (see Chapter 3). All three of these sites (C1, C4 and C5) have median total phosphorus concentrations that are well within the range of what can be considered oligotrophic (< 12 μg/L).

Farther south in the Lake, sites C6 and SJ1 have more complex temporal patterns of total phosphorus and substantial differences between surface water and bottom water (with bottom water having median total phosphorus concentrations around 30 μg/L). There is much more variability in the record at these two sites compared to C1, C4, and C5, and trends in the deeper water at these two sites are highly uncertain. The surface water at both sites appears to follow the general trend of inputs from the St. Joe River, but the bottom water records at these two sites are much less related to river inputs and show signs that they are responding to internal sources of phosphorus in the Lake bed. Bottom waters at both these locations are anoxic in late summer and fall, a likely cause of the mobilization of phosphorus (see Chapter 7).

The conclusion drawn from the trends analysis is that total phosphorus conditions in CDA Lake are rapidly changing, such that frequent reanalysis (as often as every two years) is crucial. The evidence as of the end of 2020 does not support the hypothesis that total phosphorus concentrations in the Lake are trending upward. For surface water samples at all five sites, the most recent trends in total phosphorus are downward, especially from 2016 to 2020 (see Figure 5-10). The substantial differences in total phosphorus concentrations between surface water and bottom water at the two sites in the southernmost parts of the Lake (C6 and SJ1) indicate that internal processes are significant in the latter areas.

The recent declines in total phosphorus in the Lake at C1, C4, and C5 appear to be linked to the declines in total phosphorus flux for the two main tributaries of the Lake. The possible drivers of these changes (discussed in Chapter 3) are likely the diminution of mining and the effectiveness of the Superfund remedy in terms of soil loss,

TABLE 5-6 Summary of Trends in Total Phosphorus Concentration at Five Monitoring Sites in CDA Lake

NOTES: Amounts of change and their significance are determined by the Theil-Sen slope estimator and the Seasonal Kendall test. Dark blue shading indicates a highly significant downward trend (p < 0.01), light blue shading indicates trends that are moderately significant (0.01 < p < 0.1), and the unshaded trends are not statistically significant (p > 0.1). Median values in 2020 are determined using the loess smoothing shown in Figure 5-10.

improved forest practices, and other efforts carried out under the Lake Management Plan to reduce streambank erosion in the watersheds of the CDA and St. Joe Rivers. The ability to project future changes in phosphorus inputs and concentrations in the Lake will depend on research aimed at better understanding the causative mechanisms of these recent trends. Future trends in total phosphorus will depend on a combination of factors, including continuing improvements in landscape stability in formerly mined parts of the watershed and in the channel of the CDA River, changes in rain intensity and river discharges, and the implementation of additional phosphorus control measures under the Lake Management Plan. Forecasting how these changes will influence the delivery of phosphorus to the Lake will depend on the creation of predictive models that take these processes into consideration and utilize the history of the river and Lake total phosphorus records to evaluate the ability of the models to simulate trends.

Trends in Total Nitrogen

Analyses of seasonality and trend for total nitrogen (TN) were run at all five CDA Lake sites, although the records at C5, C6, and SJ1 are all shorter than ten years. The data from the past six years are displayed in Figure 5-11.

The seasonal patterns differ among the sites. At C1 and C4, the patterns are very similar to the total phosphorus record at these sites. The highest values come with the high inflows of the late winter and spring, with concentrations declining through the remainder of the year. The highest values in this period are associated with the year of highest CDA River inflow, 2017. The total nitrogen pattern at site C5 is more complex. The highest values are generally at the first sampled month in the year (such as March or April), but then concentrations are much lower in the summer and appear to rise up to the last sample in the fall. At sites C6 and SJ1 the total nitrogen concentrations tend to climb throughout the entire sampling season and not be so highly related to inflow.

Figure 5-12 shows the trends at all of the sites. Trends in total nitrogen shown in Table 5-7 are quite different across the five sites. These results are consistent with the trends in flow-normalized nitrogen fluxes for the CDA River and St. Joe River reported by Zinsser (2020).

TABLE 5-7 Summary of Trends in Total Nitrogen at Five Monitoring Sites in CDA Lake

NOTES: Amounts of change and their significance are determined by the Theil-Sen slope estimator and the Seasonal Kendall test. The tests include data from all months and from both surface water and bottom water samples. Dark blue shading indicates that each is a highly significant downward trend (p < 0.01), pink shading indicates upward trends that are moderately significant (0.01 < p < 0.1), and the unshaded trends are not statistically significant (p > 0.1). Median values in 2020 are determined using the loess smoothing shown in Figure 5-12. * indicates that the record is seven years rather than ten years.

Changes in atmospheric deposition may also be an important driver of changing nitrogen concentrations in water. The role of atmospheric deposition in this watershed is a topic that has not been studied, but should be part of future evaluations of nitrogen dynamics of the lake and watershed. Across all of the stations evaluated, the bottom waters have higher concentrations (by amounts like 25 percent) than the surface waters. Although phosphorus concentrations have been the primary focus in discussions of possible eutrophication of CDA Lake, nitrogen plays a role as well. All of the available Lake data sets indicate a general improvement in conditions in recent years, but an understanding of the causes is important to being able to evaluate likely future trends. Gaining a better understanding both of the nitrogen cycle, and of nitrogen trends in the Lake and rivers is an important topic that needs to be included in the research agenda supporting adaptive management of CDA Lake trophic status.

IN-LAKE PRODUCTIVITY ANALYSES

To complement the preceding analyses of dissolved oxygen and nutrients, the committee analyzed additional measures of lake productivity—trophic status and chlorophyll a. It also tested the hypothesis that zinc in the water column may be suppressing productivity, in order to better understand whether the Superfund remedy might be removing an important control of eutrophication in CDA Lake.

Trophic Status of CDA Lake with Respect to Nutrients and Chlorophyll a

Data from the 2017 National Lakes Assessment were used to assess CDA Lake with respect to trophic status. The National Lakes Assessment is a regionally stratified random sample of lakes (larger than 1 ha and deeper than 1 m) and thus provides an accurate picture of summer lake conditions in the continental United States. Data for total nitrogen, total phosphorus, and chlorophyll (chl) for more than 1,110 unique study sites were available in the 2017 database (EPA, 2021). The data can also be partitioned by region or state. In Tables 5-8 to 5-10, these data are compared to those obtained for CDA Lake, which were monthly epilimnetic or euphotic averages for summer (June–September) for both IDEQ and tribal data at stations C1, C4, C5, C6, and SJ1. Note that the CDA

TABLE 5-8 Nitrogen Data from U.S. Lakes, Northwest Lakes, and CDA Lake

SOURCES: Top two rows: 2017 National Lakes Assessment data, https://www.epa.gov/national-aquatic-resource-surveys/nla. CDA Lake data courtesy of IDEQ and the CDA Tribe.

TABLE 5-9 Phosphorus Data from U.S. Lakes, Northwest Lakes, and CDA Lake

SOURCES: Top two rows: 2017 National Lakes Assessment data, https://www.epa.gov/national-aquatic-resource-surveys/nla. CDA Lake data courtesy of IDEQ and the CDA Tribe.

Tribe’s dataset had total nitrogen records recorded as “< 50 μg/l”; in these cases, the value was set to 50 μg/L. In the tribal and IDEQ data sets, chlorophyll values coded as “< X” were set to X. Thus, average total nitrogen and chlorophyll values are likely overestimates.

The committee also compared total nitrogen:total phosphorus (TN:TP) ratios in CDA Lake in relation to those in the National Lakes Assessment database. Averaged across stations C1, C4, C5, C6, and SJ1 since 2007, for which paired observations of TN and TP are available, average TN:TP ratios are 40.2 (atomic ± 1.3 s.e.). This value is somewhat higher than the median value of 30.8 (atomic) for lakes in the NLA. A transition to onset of nitrogen limitation occurs when TN:TP ratios are lower than a threshold of 31:4 (atomic) (Downing and McCauley, 1992). Thus, nitrogen limitation of phytoplankton growth seems possible in CDA Lake at sites and times when the TN:TP ratio is low. In general, TN:TP ratios were somewhat lower in the southern part of the Lake (C5, C6, SJ1), implying that nitrogen limitation of phytoplankton growth might be more frequent there than in the northern part of the Lake (C1, C4).

According to the trophic status parameters of Carlson (1977), CDA Lake’s median values for total phosphorus (11.1 μg/L) and chlorophyll (1.80 μg/L) concentrations place the Lake in the oligotrophic (< 12 μg P/L, < 2.6 μg chl/L) category.

Chlorophyll Trend Analysis

The committee evaluated the chlorophyll data collected in surface water at sites C1, C4, C5 and C6. For sites C1 and C4, the dataset runs from July 2002 through October 2020. (Data from 1991 and 1992 are not considered because of the long time gap prior to the main body of the dataset.) About 6 percent of these values are censored, meaning that the samples were recorded as “<1 μg/L,” and were treated as being equal to half their reporting limit. For sites C5 and C6, the dataset runs from July 2007 through December 2020. Of these data, about 15 percent are censored and were handled in the same manner as the censored data from C1 and C4. Site SJ1 was not considered because more than 50 percent of the data are censored.

For each of the four sites, the data were evaluated by doing individual Mann-Kendall trend tests for the time series for each of ten “seasons.” The “seasons” are generally the calendar months, except that January and February data are considered to be one season, and November and December are considered to be one season (because these months had relatively few observations). In the case of C5 and C6, there were insufficient data to conduct the trend analysis on the January–February data.

There was a total of 38 cases, each a combination of season and site, which could be evaluated for trends (4 sites × 10 seasons—C5/C6 Jan/Feb cases). For 9 of the 38 cases that could be evaluated, positive trends were

TABLE 5-10 Chlorophyll Data from U.S. Lakes, Northwest Lakes, and CDA Lake

SOURCES: Top two rows: 2017 National Lakes Assessment data, https://www.epa.gov/national-aquatic-resource-surveys/nla. CDA Lake data courtesy of IDEQ and the CDA Tribe.

TABLE 5-11 Trends in Chlorophyll by Month in Surface Water at C1, C4, C5, and C6, 2002–2020

NOTES: The values shown are the estimated slope of the trend in μg/L/yr. Months marked with an asterisk are those with three or fewer observations. The color code relates to the significance level of the observed trend. Light blue indicates a moderately significant downward trend (0.10 > p > 0.01), and no shading indicates that results are not significantly different from zero (p > 0.10). The last column, labeled “All,” is the result of a Seasonal Kendall test on the data from all months.

observed, but none of them had even moderate levels of significance (p < 0.1). The remaining 29 cases had negative trends; five of them were of moderate significance (0.01 < p < 0.1) but none were highly significant (p < 0.01).

In addition to those analyses, a Seasonal Kendall test was run for each of the four sites by combining data across all seasons at the site. For sites C1, C4, and C6 the results showed moderately significant decreases (p-values of 0.043, 0.012, and 0.097, respectively). For C5, the sign of the trend was also negative, although it was far from being significant. For each of the months, the slope of the trend was evaluated using the Theil-Sen slope estimator.

The results of all of the Mann-Kendall trend tests for all months and the results of the Seasonal Kendall test for each of the four sites are shown in Table 5-11.

The committee makes the following observations about the results. For sites C1 and C4, three of the monthly trends that could be evaluated (regardless of significance) were positive while 17 were negative. Three of the negative values were moderately significant (0.01 < p < 0.1). When aggregated over all of the months using the Seasonal Kendall test (last column), both sites show moderately significant downward trends in chlorophyll.

At C5, the trends in five of the nine months that could be tested were positive, but none of these were even moderately significant. The other four months had negative trends. One of these (the trend in April) was a moderately significant downward trend. The overall trend as determined by the Seasonal Kendall test was slightly negative, but not even close to being significant. At C6, one month (May) had a positive slope (not significant), the other eight months that could be evaluated had a negative slope, and only one of these was significant (April). The overall result for C6 using the Seasonal Kendall test shows a moderately significant downward trend in chlorophyll.

In summary, it is very difficult to make a general case that there is a deteriorating (i.e., increasing) chlorophyll a level in the Lake over the most recent 20 years. In fact, the data suggest the contrary conclusion. Of course, these data do not include the lateral lakes, which are known to sometimes have high chlorophyll concentrations. Unfortunately, there are no long-term chlorophyll observations in the lateral lakes that would support a similar trend analysis. Given these results, coupled with the generally declining trends for phosphorus in the Lake and entering the Lake, there has been no tendency toward eutrophication of CDA Lake in the past 20 years. These results, coupled with the lack of increasing trends for phosphorus in the Lake and entering the Lake, show there has been no tendency toward more eutrophication of CDA Lake in the last 20 years. Eutrophication is not expected to become a problem if current trends continue.

The Question of Zinc Suppression in CDA Lake

Zinc is essential at low concentrations to most living things (used as an enzyme cofactor), but it is toxic at high concentrations, when it can bind nonspecifically with biologically important molecules and interfere with their function (Gonçalves et al., 2018). Thus, if zinc concentrations reach levels toxic to phytoplankton, zinc can conceivably offset the eutrophying effects of nitrogen and phosphorus loading (Kuwabara et al., 2007). Indeed, undesirable phytoplankton blooms are often, though temporarily, addressed by additions of toxic metals, especially copper (McKnight et al., 1983). Zinc and other metals can also be toxic to biota higher in the water column food web, such as zooplankton and fish.

An important question that the committee was specifically asked to address is the possibility that eutrophication in CDA Lake might occur if zinc concentrations decline to levels sufficient to release lake phytoplankton from inhibition by current high levels of potentially toxic zinc. In theory, this could set up a positive feedback in which increased algal production would lead to depletion of oxygen in the Lake’s hypolimnion, leading to release of sediment phosphorus and further amplifying eutrophication. Such concerns are plausible given the high zinc concentrations in some parts of the Lake, along with early laboratory experiments showing that high concentrations of free zinc ion reduced the growth of some phytoplankton taxa isolated from CDA Lake and inhibited their response to phosphorus addition (Woods and Beckwith, 1997; Kuwabara et al., 2007). These responses are concordant with previous experiments with cultured algae, as well as with natural phytoplankton in Lake Ontario (Wong et al., 1978; Wong and Chau, 1989). The committee recognized that it is notoriously challenging to conclusively test or refute such hypotheses with field observations alone. But sufficient weight of evidence from the field can be invaluable in identifying tendencies in the direction predicted by the hypotheses. This element of the statement of task could have important implications for future management, but it has been 15 years since the hypothesis was proposed. The committee felt an analysis, even if limited given constraints imposed by the existing data, would be appropriate.

To evaluate the likelihood of the zinc inhibition hypothesis, the committee performed three analyses with the available field data on chlorophyll that were gathered since the original hypothesis was reported:

1. It compared CDA Lake’s chlorophyll values relative to its total phosphorus concentrations and compared these to global relationships. If high zinc concentrations in CDA Lake inhibit phytoplankton biomass, then one would expect CDA Lake chlorophyll values to be lower than those observed in other lakes with similar total phosphorus values. Furthermore, one would expect that deviation to be most pronounced in parts of the Lake where zinc levels are highest.
2. It performed multiple regression analyses to predict chlorophyll as a function of total phosphorus and zinc concentrations for CDA Lake. If zinc is important in reducing chlorophyll concentrations, one would expect negative associations of chlorophyll with zinc once total phosphorus was accounted for in a regression model.
3. Since chlorophyll concentrations are likely driven by multiple factors beyond just total phosphorus and zinc, the committee assessed the shape of the relationship between chlorophyll concentration and zinc concentration in different parts of the Lake in order to determine if potential threshold or envelope effects describe how zinc affects chlorophyll–total phosphorus relationships.

A notable limitation of the field data is that all three of these approaches use overall chlorophyll concentrations as an index of phytoplankton performance or proliferation in response to zinc levels in the Lake. Chlorophyll concentration is an aggregate of chlorophyll found in cells across the range of taxonomically and physiologically diverse taxa present. Thus, bulk chlorophyll has the potential to obscure responses of zinc-sensitive taxa that may be ecologically important. Only detailed taxonomic assessments of the phytoplankton community can provide indications of potential zinc impacts on different taxa. However, as with most other lakes, such data for CDA Lake are limited in temporal and spatial scope.

Expected Chlorophyll Concentrations in CDA Lake

The committee sought to assess how CDA Lake conforms to known relationships between nutrient supply (as indexed by total phosphorus concentration) and chlorophyll concentration. Figure 5-13 plots data from a survey of published data (Havens and Nürnberg, 2009) for 389 lakes from North America, Europe, Asia, and New Zealand (all shown in orange) showing the well-known positive log–log relationship between these two parameters. Summer (June, July, August, September) monthly mean values of each parameter for each the years 2007–2020 for CDA Lake are plotted separately for IDEQ (blue) and tribal data sources (black and white). CDA Lake data lie in the low-intermediate range for both parameters relative to the global lake sample. The two parameters are positively correlated, as seen in previous studies. However, the relationship has a shallower slope, and the CDA Lake data have notably lower chlorophyll concentration for any given total phosphorus concentration, especially at higher

total phosphorus levels. This reduction in observed chlorophyll concentrations for any given level of total phosphorus is consistent with the hypothesis that high concentrations of zinc (and other metals) depress phytoplankton biomass, moderating the eutrophication response that accompanies nutrient loading.

The committee estimated CDA Lake’s expected chlorophyll concentration if the system behaved according to the global relationship reported by Havens and Nürnberg (2009). Using the median of total phosphorus data for CDA Lake (9.53 μg/L; tribal and IDEQ data, summer only), the predicted value from the global chl–TP relationship is 3.50 μg chl/L. In contrast, the observed median chlorophyll value in CDA Lake is 1.55 μg chl/L (summer). From these data, one can infer that, on average, chlorophyll levels in CDA Lake are ~2.25 times lower than expected based on the Lake’s total phosphorus concentration. Indeed, if the Lake had the predicted chlorophyll concentration, it would be classified as mesotrophic, not oligotrophic. However, this discrepancy is not uniform across the Lake, as indicated in Figure 5-13. That is, because the slope of the chl–TP relationship is shallower than the one for the global lake data, the discrepancy is larger for samples with higher total phosphorus concentrations.

As seen in Figure 5-13, samples from the southern end of the Lake have generally higher total phosphorus concentrations and chlorophyll concentrations than those at the northern end of the Lake. Thus, in this analysis, southern sites have a greater discrepancy between observed total phosphorus concentration and the chlorophyll concentration expected from the global chl–TP relationship. This holds when considering individual stations within the southern part of the Lake. For example, site C6 is the furthest south and zinc concentrations are < 5 μg/L throughout the recorded data (lowest in the Lake). However, examination of Figure 5-13 shows that it is not distinct in terms of the chl–TP relationship for the Lake as a whole. That is, C6 also shows a lower chlorophyll concentration given its total phosphorus level relative to the global lake dataset. Thus, this lack of a within-lake, zinc-dependent pattern in the chl–TP relationship is not consistent with the hypothesis that high zinc concentrations inhibit phytoplankton biomass in CDA Lake.⁶

___________________

⁶ Our conclusion (lack of support for Zn suppression) from looking at C6 might be tempered by the occurrence of anoxia (and thus unusually low chl and high TP) on some of the dates for C6.

Multiple Regression Analysis

To explore these possibilities further, the committee developed multiple regression analyses of chlorophyll concentrations in CDA Lake as a function of total phosphorus and zinc using datasets provided for sites C1, C4, C5, and C6. First, for each of the hundreds of individual months that might be present in the records (near monthly for most of 1991, 1992, and 2002–2020), the committee extracted all the surface water data (“epi” or “photic zone,” always less than 20 m) collected at a given site for a given month. For each month at each site, the median value of total phosphorus, total zinc, and chlorophyll a was computed. Cases where there was a “less than” value were coded to be half the reporting level (there were not very many of these except at C6, where most of the zinc values were “less than” values). For purposes of the regression analysis, only sites and months in which all three values were available were used; this comprised a dataset of 458 valid cases. Based on the frequency distributions of each of the variables, all data were natural log-transformed.

The most general model considered looked at all 458 cases (without regard to site or time of year). The form of the model was as follows:

ln(Chl) = β_o + β₁ · ln(TP) + β₂ · ln(TZn) + ε

The expectation based on the zinc inhibition scenario is that β₁ should be positive (more total phosphorus leads to more chlorophyll) and that β₂ should be negative (more total zinc leads to less chlorophyll). However, when this model is fit, it has a very low R² value (0.05), but it is statistically significant overall (because the sample size is relatively large). The coefficient on ln(TP) is positive (+0.25), as expected, and it is highly significant. However, the coefficient for ln(TZn) is positive (+0.02), not negative, and is not significant. Hence, there is no evidence of zinc inhibition from this analysis.

To consider this hypothesis more deeply, the committee also assessed subsets of the data (e.g., combinations of particular sites or months). In doing so, various outcomes, some supporting the zinc inhibition hypothesis—for example, an analysis of all sites in October yielded a statistically significant positive association for ln(TP) and a statistically significant negative association for ln(TZn)—and some contradicting it, were found. However, such efforts represent a form of “p hacking” (searching among a multitude of correlations for one that fits a preconceived notion) and is not recommended practice. Thus, this regression approach did not prove fruitful in shedding light on the zinc inhibition effect.

Shape of Zinc versus Chlorophyll Relationships

As one final assessment of a possible influence of zinc on chlorophyll concentrations, the structure of the data distribution was considered following Schmidt et al. (2012). This is based on the idea that ecological responses (e.g., chl a) to a well-characterized inhibitory factor (e.g., zinc) in the presence of other limiting factors often result in a wedge-like pattern across the gradient of that limiting factor (Terrell et al., 1996; Schmidt et al., 2012). Interpretations of such relationships are vulnerable to coincidental occurrences of events and confounding variables, of course (e.g., Cloern, 2022).

Figure 5-14 shows the relationship between total zinc and chlorophyll for all available data for the photic zone (i.e., data from C1, C4, C5, C6, and the lateral lakes, Swan and Thompson). For the overall dataset, there is a statistically significant (p < 0.0001) but very weak negative (r = −0.18) relationship between chlorophyll and total zinc (Figure 5-14), consistent with zinc inhibition. However, for stations in the main body of the Lake (C1, C4, C5), there is a statistically significant (p < 0.02) but very weak positive (r = 0.13) relationship (data not shown).

High chlorophyll concentrations (> 5 μg/L) in locations with zinc concentrations less than 10 μg/L are evident in the data distribution. However, the only environments with chlorophyll concentrations > 5 μg/L and zinc < 10 μg/L are C6 and some lateral lakes. These, coincidentally, are the most shallow and nutrient-rich locations in the lake for which relevant data are available. At C5, on the other hand, dissolved zinc was also less than 10 μg/L in April–June in 2010, 2011, and 2018 (see Chapter 6), but these were the years with the lowest chlorophyll a at C5 (< 1 μg/L). The co-occurrence of low zinc and high chlorophyll a at C6, therefore, could represent a coincidence tied to the type of environment, rather than evidence that release of zinc inhibition results in high chlorophyll a.

In those years when zinc concentrations in the euphotic zone were greater than ~60 μg/L (all before 2004), chlorophyll a concentrations were always below 5 μg/L and declined as zinc concentrations increased. This could represent inhibition of chlorophyll a by zinc, but the threshold for such effects was ~60 μg/L, a concentration higher than currently found in surface waters.

Overall Conclusions Regarding Zinc–Total Phosphorus–Chlorophyll Relations

It has been plausibly hypothesized, based on laboratory experiments, that high zinc concentrations (> 10 μg/L; Wong et al., 1978; Wong and Chau, 1989) inhibit phytoplankton growth in CDA Lake specifically by suppressing the response of some phytoplankton species to phosphorus (Woods and Beckwith, 1997; Kuwabara et al., 2007). This is an important concern because mitigation of zinc loading via the Superfund remedy might release phytoplankton from that inhibition, accelerating eutrophication. Indeed, chlorophyll a concentrations in CDA Lake are lower than predicted from total phosphorus concentrations compared to other lakes globally (Figure 5-13), an observation that could be supportive of possible zinc inhibition of phytoplankton growth response to phosphorus supplies. However, from the multiple regression analysis, it is clear overall that zinc inhibition of chlorophyll in the Lake is not consistently detectable in a statistical sense in data from multiple sites across multiple years. Nor is it consistent with the zinc inhibition hypothesis that locations in the Lake with low zinc concentrations also fall below the global chlorophyll–total phosphorus relationship.

Several considerations might explain why the experimental suggestions of negative effects of zinc on phytoplankton response to phosphorus may not result in detectable and consistent relationships among chlorophyll, total phosphorus, and zinc in the euphotic zone.

1. Chlorophyll responds to multiple variables in CDA Lake (e.g., nutrient concentrations including possible nitrogen limitation, temperature, turbidity, residence time, zooplankton grazing, etc.). To drive chlorophyll concentration in a consistent manner, effects of zinc would have to overwhelm these influences. These confounding variables

1. are not present in the laboratory studies of Woods and Beckwith (1997) and Kuwabara et al. (2007). For example, nitrogen limitation of phytoplankton may also contribute to the shallow slope observed for the chl–TP relationship (Figure 5-13) as nitrogen limitation would constrain chlorophyll biomass below that expected for a given supply of phosphorus.
2. Even if zinc inhibited growth of some green algal taxa or other taxa, compensation in terms of increased production of chlorophyll by diatoms and other less-sensitive species would not be surprising, nor would this result be completely inconsistent with experimental results. Determination and analysis of taxa-specific distributions in the lake could be used to directly test the hypotheses of Kuwabara et al. (2006). That study found less of the zinc-sensitive taxa (the chlorophyte Chlorella spp.) at C5 (where zinc concentrations were higher) than at C6 and SJ1 (where zinc concentrations were < 5 μg/L). In contrast, the zinc-tolerant diatom (Asterionella spp.) was more abundant at C5 and less abundant than Chlorella spp. at C6/SJ1. A preliminary analysis by the committee of Chlorella sp. and Asterionella sp. from the most recent surveys found the diatom A. formosa to be more abundant than the chlorophyte Chlorella sp. at both C5 and C6; Chlorella sp. were only found at these locations in 2007 through 2009.⁷
3. Zinc concentrations in the photic zone (or epilimnion) are spatially and temporally variable, driven by multiple factors (see Chapter 6). Those influences differ with season and between the northern and southern Lake (see discussion above) and could confound any simple relationships among chlorophyll, total phosphorus, and zinc.

In short, zinc inhibition of some taxa of phytoplankton at the concentrations observed at some places and at some times in CDA Lake cannot be excluded. However, the weight of evidence available to the committee does not support an overall influence of zinc on the Lake’s trophic status at present. Nor is there evidence that zinc is currently inhibiting chlorophyll to the extent that reducing zinc inputs to the Lake would cause an increase in chlorophyll concentrations sufficient to cause eutrophication and subsequent anoxia in bottom waters. The reason chlorophyll concentrations in CDA Lake are low relative to phosphorus concentrations remains unclear.

These analyses highlight the difficulties of inferring the impacts of possible zinc toxicity on eutrophication in CDA Lake from either single species experiments or observational data alone. The nature of the Zn–chl relationship in the Lake is dependent on what dataset is selected, and interpretation of observed dependences is complicated by the intervention of various confounding factors. More confidently establishing the nature of this relationship would require controlled field experimentation in which zinc concentrations are enriched in low-zinc zones of the Lake (a feasible experiment). An experiment in which zinc concentrations are selectively lowered in high-zinc zones of the Lake seems less feasible at present. However, this experiment could occur “naturally” over decades if remediation continues to reduce zinc inputs at the present pace. Field experiments testing for zinc effects on particular taxa of CDA Lake phytoplankton across seasons of interest could be especially important given that overall chlorophyll concentrations might mask ecologically important impacts. Given the physiological plausibility of zinc inhibition of phytoplankton growth, continued monitoring and investment in both passive and active adaptive management will be critical tools in the future.

CONCLUSIONS AND RECOMMENDATIONS

This chapter analyzed water column data from CDA Lake over the past 30 years to reveal trends in dissolved oxygen, nutrients, and lake productivity. Based on the committee’s analysis of available datasets to better understand changes in dissolved oxygen and nutrient concentrations in CDA Lake, the following detailed conclusions and recommendations are made. It should be noted that these conclusions apply to the main body of the lake, where long-term monitoring data were collected, and not to nearshore or littoral areas, for which data are absent.

1. The evidence that dissolved oxygen concentrations are worsening in the bottom waters of the C1, C4, C5 and C6 stations is equivocal at best. Rather, there is an increasing dissolved oxygen concentration trend particularly

___________________

⁷ This paragraph was updated after report release to clarify the committee’s preliminary analysis and to better explain the dates when Chlorella data were available.

1. evident at C5.⁸ The analyses reported herein used a coarser sampling resolution for the bottom samples than is ideal because the most important information is the oxygen concentration immediately above the sediment–water interface. It would be beneficial to know the actual bottom depth associated with each profile so that the bottom dissolved oxygen data can be presented relative to their distance from the sediment–water interface. Coupled with the recent trends in phosphorus concentrations in the Lake (see below) and phosphorus loading to the Lake (see Chapter 3), low dissolved oxygen is not a current problem in the main body of the lake, nor it is expected to become a problem if current trends continue. However, if climate change were to strengthen thermal stratification substantially in the future, dissolved oxygen concentrations at the sediment–water interface would likely decrease, reversing current trends.
2. For the most recent decade, the data indicate declines in total phosphorus concentration in CDA Lake, although this decline was not statistically significant in the northern lake. Declines in total phosphorus in the Lake are consistent with the declines in total phosphorus from the two major rivers entering the Lake. For the northern portions of the Lake, these trends are a reversal of prior increasing trends similar to the reversal of the trend in CDA River total phosphorus inputs. In the southern parts of the Lake at C5, the trends have been steadily downward for more than a decade, similar to the steady downward trend in St. Joe River total phosphorus inputs. The only exception to these patterns is that for deeper water at C6 and SJ1 the total phosphorus trends are upward, suggesting that internal sources of phosphorus from the Lake bed may be important in these areas. The ability to project future changes in phosphorus inputs and concentrations in the Lake will depend on research aimed at better understanding the causative mechanisms of these recent trends in phosphorus delivery from the watershed to the Lake.
3. Trends in total nitrogen are more complex than those for total phosphorus. Gaining a better understanding of the nitrogen cycle (such as assessing possible nitrogen fixation in the water column spatially and temporally or identifying hot spots for denitrification), and analyzing chemical speciation of and trends in nitrogen concentration in the Lake and rivers, is an important topic that needs to be included in the research agenda supporting adaptive management of the Lake’s trophic state.
4. CDA Lake is characterized by low total phosphorus and chlorophyll concentrations. According to these widely used metrics of lake trophic status, the Lake can be classified as oligotrophic. Indeed, median values of total phosphorus and chlorophyll are seven times and five times lower, respectively, in CDA Lake than median values in lakes in the National Lakes Assessment.
5. The available field evidence does not support the concept that the current high zinc concentrations in CDA Lake suppress chlorophyll a. CDA Lake supports lower amounts of chlorophyll a per unit phosphorus than generally observed for lakes worldwide. Although this is consistent with zinc suppression of phytoplankton biomass, beyond this observation there is little evidence from more detailed analysis of field data that the current high zinc concentrations suppress chlorophyll a. For example, chlorophyll a levels are disproportionately lower than expected in the southern part of the Lake in both zinc-enriched and zinc-poor locations. Consistent with this, multiple regression analysis relating chlorophyll to total phosphorus and total zinc finds a strongly positive and statistically significant association for total phosphorus but a highly nonsignificant association for total zinc. Thus, zinc suppression of overall phytoplankton biomass is not consistently supported, although zinc suppression of individual species of phytoplankton cannot be discounted. Further research involving field experimentation would help develop greater confidence in predicting the response of chlorophyll concentrations and particular taxa of phytoplankton to potential reductions in legacy metal contamination in the Lake.
6. The possibility exists that nutrient enrichment could take place in shallow water areas at the edges of CDA Lake, particularly in areas where land use change could bring about increased inputs of nutrients. These areas could,

___________________

⁸ This conclusion was edited after report release to reflect only the dissolved oxygen trend analyses shown in the report, which were for bottom waters.

1. in the future, become areas of localized growth of submerged aquatic vegetation or other forms of organisms that could lead to localized hypoxia when they die off. Attention to monitoring water-column nutrients, chlorophyll, and oxygen in embayments and near-shore environments should be considered, even though the main body of the Lake continues to be oligotrophic and not likely to experience dissolved oxygen depletion.

REFERENCES

This page intentionally left blank.