Grantee Research Project Results
Final Report: Reducing Uncertainty in Estimating Toxaphene Loading to the Great Lakes
EPA Grant Number: R825246Title: Reducing Uncertainty in Estimating Toxaphene Loading to the Great Lakes
Investigators: Swackhamer, Deborah L. , Hites, Ronald A.
Institution: University of Minnesota , Indiana University - Bloomington
EPA Project Officer: Chung, Serena
Project Period: November 15, 1996 through November 14, 1999
Project Amount: $296,996
RFA: Air Quality (1996) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Air
Objective:
There were three objectives for this research project. These were:
Objective 1. Sediment and water were sampled from Lakes Michigan and Superior (see Figure 1) to determine toxaphene concentrations and estimate fluxes between media. Data for t-chlordane, c-chlordane, t-nonachlor, and c-nonachlor also were included. Although previously established methods were used for sample collection, several new approaches were utilized in clean-up and analytical quantitation, thereby requiring extensive method optimization and validation. In addition to providing a descriptive data set for the lakes, this information forms the basis for future predictions of toxaphene fate using environmental models.
Objective 2. The relative importance of atmospheric versus non-atmospheric loadings of toxaphene to northern Lake Michigan was estimated using sediment core analysis from Green Bay along with air and water concentrations. These data, taken with existing Lake Michigan data then were used to validate a mass balance model for Lake Michigan.
Past results (Pearson 1996; Pearson, Swackhamer, et al., 1997), showed that levels of toxaphene in Lake Michigan sediments were higher in the northern basin compared to the southern basin. Pearson, Swackhamer, et al., suggest that, beginning in the mid-1980s, a source of 100-200 kg/yr of total toxaphene may have resulted from non-atmospheric loading. Because sediments can provide a reliable historical profile of deposition, numerous cores were taken in Green Bay. One was taken in the northern basin of the open lake for this study. This helped to reduce uncertainty in toxaphene inventories in sediments and provided an evaluation of tributary inputs by sampling depositional zones that capture tributary outflow. Sediment cores were collected from near the mouth of the Fox River north to the top of Green Bay and in the northern part of Lake Michigan. Sediment cores are more useful than water measurements for this purpose because they provide a current as well as past record of any significant toxaphene inputs.
Objective 3. Using the STELLA model, a multilevel framework that allows the user to construct dynamic multivariate mass balance models tailored to a specific system, dynamic mass balance models for toxaphene in both Lake Superior and Lake Michigan (Swackhamer, Schottler, et al., 1999) was refined and used to generate future predictions of water and sediment concentration. (The STELLA model allows for the specification of the type and number of input data, the equations used for the input and loss processes to and from the lake, and the time increment.) The model outcomes were compared to measured results and previous modeling efforts for reasonableness. The final model and the data generated from it contributed to testing the objectives. In addition, the model output was used to estimate the half-life of toxaphene in Lake Michigan and Lake Superior and provided insight into the time required for concentrations to decline to levels consistent with those utilized in establishing fish consumption advisories.
Summary/Accomplishments (Outputs/Outcomes):
The current research examined toxaphene in Lake Superior and Lake Michigan to determine spatial and temporal trends in concentration, homolog distribution, water/particle partitioning (Kp), fluxes between media, and the potential for the existence of significant non-atmospheric sources. There were no clear spatial trends in either lake (except Green Bay) and toxaphene net air/water flux showed consistent volatilization during all times tested.
In general, volatilization decreased epilimnetic concentrations during stratified conditions, and lake-wide mixing during overturn increased surface concentrations. The mean dissolved toxaphene concentration in Green Bay was significantly lower than in the open lake during isothermal conditions, due to minimal hypolimnetic recharge. Average surficial dissolved toxaphene concentrations in Lake Superior averaged 0.91 ng/L in 1997, and 0.73 ng/L in 1998. Dissolved toxaphene concentrations for Lake Michigan averaged 0.40 ng/L in 1997, and 0.35 ng/L in 1998.
Toxaphene homolog distributions showed significant differences between dissolved and particulate phases in water and sediments. The homolog distribution in Lake Michigan water was enhanced in lower chlorinated homologs compared to Lake Superior. Toxaphene Kp varied with season, and showed varying dependence on particulate organic carbon (POC) (no correlation in Lake Superior; seasonal correlation in Lake Michigan). The overall average log Kp value for Lake Superior was 4.93 and for Lake Michigan was 4.49. Sediment analysis in Lake Michigan demonstrated that Green Bay is not a likely significant source of toxaphene to the northern basin of the lake. However, there were elevated concentrations and inventories of all compounds studied at the southern end of Green Bay, reflecting likely point-source inputs. Concentration trends from a dynamic mass balance model were consistent with measured values. Factors such as model results, a lack of spatial differences in dissolved and sediment concentration, seasonal concentration fluctuations, and a homolog fingerprint similar to air supported the assertion that no significant non-atmospheric toxaphene sources in Lake Superior existed. Model results also were used to estimate toxaphene half-lives of 19.5 years for Lake Superior and 9.5 years for Lake Michigan, and were the basis for evaluating fish consumption advisory trends.
Toxaphene has been identified in several locations in the Duluth/Superior harbor at the western edge of Lake Superior at concentrations up to 140 ng/g (Schubauer-Berigan and Crane, 1997). Swackhamer, et al. (1998), summarized toxaphene data for the lake, reporting dissolved water concentrations of 1.12 ng/L for Lake Superior, 0.38 ng/L for Lake Michigan, and a water particulate concentration of 9.4 ng/g for Lake Michigan. They also reported that sediments had similar concentrations among the lakes (around 15 ng/g) and for filtered water particulates (around 9 ng/g), but had different homolog compositions among the lakes (see Table 1). The bioaccumulation factor (BAF) is determined by dividing the level in the organism by the level found dissolved in water, normalized to lipid (zooplankton and fish) or organic carbon (OC; phytoplankton).
Concentration, ng/g dw | Concentration, ng/g lipid or OC | mean normalized log BAF | |
Phytoplankton | 51.3 ± 30.4 | 295 ± 197 | 5.82 ± 0.24 |
Zooplankton | 243 ± 139 | 843 ± 566 | 6.53 ±0.24 |
Mysis | 92.4 ± 32.8 | 843 ± 566 | 6.29 ± 0.24 |
Bythotrephes | 162 ± 57.2 | 1530 ± 918 | 6.70 ± 0.33 |
Diaporeia | 411 ± 57.2 | 3710 ± 3020 | N/A |
Sculpin | 225 ± 101 | 4300 ± 1934 | 6.58 ± 0.34 |
Lake trout | 2370 ± 1450 | 10300 ± 3920 | 6.96 ± 0.42 |
N/A = not applicable; n.d. = not determined. |
Existing studies of toxaphene in the Great Lakes include levels in surface sediments and water (Pearson 1996; Pearson, Swackhamer, et al., 1997). Current concentrations in sediments ranged from a low of 2.8 ng/g in a nearshore core of Lake Superior to a high of 45 ng/g in a core taken from northern Lake Michigan. Except for cores from northern Lake Michigan, the current toxaphene levels in sediment cores from depositional zones of the Great Lakes were 15 + 4 ng/g, which was similar to the current concentration in one of the control lakes. Current accumulations ranged from 0.065 ng/cm2 yr in a control lake (e.g., only long-range atmospheric input) to 1.0 ng/cm2 yr in one of the northern Lake Michigan cores. Sediment inventories ranged from 2.8 - 12 ng/cm2 yr in a control lake (e.g., only long-range atmospheric input) to 1.0 ng/cm2 yr. The homolog compositions of the sediment cores changed with depth consistent with degradation of toxaphene. However, the degradation rates were shown to be slow, with a half-life of about 50 years. One exception to the weathering of toxaphene in sediment was in Lake Superior, where two adjacent cores showed homolog compositions similar to parent toxaphene.
Comparison of the accumulations and concentrations in the Great Lake cores to those in the control lakes indicated that the atmosphere is the dominant source of toxaphene to the Great Lakes. One of those exceptions was northern Lake Michigan, which was estimated to be receiving about 30 percent-50 percent of its current inputs from non-atmospheric sources (Pearson 1996).
Figure 1. Sample Locations
There were few data outliers or missing data for Lake Superior. Samples were not collected in the spring of 1998 at Station 22/1dd due to scheduling limitations. The filters used to determine the fraction organic carbon for the particulate phase for Station 19 in spring 1998 yielded a higher carbon mass per liter than total particulate mass per liter. To compensate for this, the average concentration of particulate organic carbon for the two stations geographically bounding Station 19 was inserted, and a fraction of the organic carbon value was estimated. All other data related to Lake Superior met the QA/QC requirements. Variability was examined using the General Linear Model (GLM).
The most significant factors relating to variability in toxaphene concentration are associated with temporal differences and covariates that tend to change with time (e.g., water temperature). These factors, in turn, are highly correlated with changing air concentrations as a result of changing air mass sources. This indicates that toxaphene is not at a constant concentration over time, and that temporal characterization of a sample is crucial to understanding the significance of a given observation with respect to the lake ecosystem. There also are spatial differences with respect to the particulate phase that result from changing conditions within the lake.
Roughly 97 percent of the mass of total toxaphene in Lake Superior water was present in the dissolved phase. The majority of toxaphene in Lake Superior entered the lake in the dissolved phase because the main source of loading is from air/water transfer. The concentration remains high in the dissolved phase due to the relatively low water/particle partition coefficient. In addition, there are limited removal mechanisms, such as low sedimentation rate and minimal tributary outflow.
The optimal set of parameters for the GLM have an overall R2 value of 0.765. When the covariates water temperature or air temperature are included in the model, both factors were significant, but produced a lower R2 value (0.74 and 0.72, respectively). Therefore, while the physical parameters describing temperature did contribute to determining dissolved toxaphene concentration, the most significant factors involved temporal characteristics of the samples. There is no significant contribution from the factor station, indicating that the lake is mixed well horizontally. This is consistent with the lack of a current, significant point source of toxaphene to the water column and reinforced the assertion that atmospheric transport is the main source to the lake. There is a significant difference (p = 0.001) of dissolved total toxaphene between years, with 1998 being lower in average concentration.
To investigate potential trends within a specific year, individual cruises are compared. Many significant differences are found both between years and individual cruises. Differences between individual cruises were defined using a separate GLM model with cruise number as the only factor. Because the sequential cruise number incorporates both year and season, this model is equivalent to those using season, year, and season x year and is as robust (cruise # p = <0.001; model R2 = 0.765). The most significant differences are noted between cruises during isothermal conditions (e.g., cruises 1, 4, and 5) and those during stratified conditions (cruises 2, 3, 6, 7, and 8). There were lower concentrations observed during stratified conditions because of losses from the epilimnion by volatilization. There is a gradual decline in concentration over time during stratified conditions in both years. When the lake vertically mixes in the fall, hypolimnetic water containing higher concentrations of toxaphene mixes with the epilimnetic water, raising concentration at the surface. Recharge of the surface with hypolimnetic toxaphene was likely responsible for the increase in dissolved toxaphene between cruises 2 and 3. It also explained why cruise 3 was more consistent with the isothermal cruises rather than the stratified cruises. The absence of major upwelling events to recharge the surface toxaphene caused a decline in dissolved concentration over time in 1998. The sustained lower concentration in the October 1998 sample was the result of continued stratified conditions late in the season. The spring values for each year are statistically similar, even though the concentration trends in each year are different. This is because the lake was mixed well vertically during spring turnover. There is a slight decline from the spring 1997 value (1.01 ng/L) to the 1998 spring value (0.94 ng/L), which reflects the net mass lost from the system between 1997 and 1998.
The dissolved phase concentration for total toxaphene correlates with measured air, water temperature, and the temperature of the air/water interface (Schwab, Leshkevich, et al., 1999). The concentration of dissolved toxaphene decreased with increasing interface temperature. This is because the Henry's Law (H) constant increases with increasing temperature, thereby enhancing flux from water to air, and depleting the dissolved water phase in the upper zone of the lake. The interface temperature is lowest during the highest measured concentration, reflecting isothermal conditions and the mixing of hypolimnetic toxaphene to the surface. The upward trend in 1997 was a result of breaks in the stratification that occurred over summer, while the October sample was taken after the fall overturn. In 1998, the consistent downward trend reflected the stratified conditions throughout the summer, while the October sample was taken before the fall overturn.
The concentration values presented here for total dissolved toxaphene are consistent with previous reports from Lake Superior (Pearson, Swackhamer, et al., 1997; Swackhamer, Pearson, et al., 1998) and show that the levels are gradually declining with time. The average dissolved phase concentration in 1996 was 1.12 ng/L (Swackhamer, Pearson, et al., 1998), which is about 17 percent higher than the observed value for 1997 from this study. The first order overall loss rate constant (k) for total toxaphene may be calculated using the average concentration; A is the 1996 concentration; and k is the slope. This yields a value of k of -0.21yr-1 with an R2 of 0.999. This yields a half-life of 4.73 years for dissolved toxaphene in Lake Superior (see Figure 2).
Figure 2. Average Dissolved Total Toxaphene (Diss tox) by Individual Cruise in ng/L.
Bars represent 95 percent confidence intervals. N = 7 for April/May, 1997, August/September 1997 and 1998; N = 2 for June/July 1997, and October 1997 and 1998; N = 3 for Jun/Jul 1998. All June/July, August/September, and October 1998 cruises are in stratified conditions; others are isothermal. Solid line represents the monthly average air/water interface temperature as determined from satellite data (Schwab, Leshkevich et al., 1999).
There is no significant difference between stations for the dissolved toxaphene, indicating that the toxaphene identified in the Duluth harbor (Schubauer-Berigan and Crane, 1997) does not have a significant influence on the concentration of dissolved toxaphene in the western arm of the lake. Also, the additional particulate load from the south shore of the western arm (Scholz, 1985) did not appear to affect the dissolved concentration, and indicated no significant input from other tributaries or erosion.
When examining toxaphene homolog concentration trends in the dissolved phase, the same factors were significant in explaining variability as those found for total toxaphene. The lower chlorinated homologs (6-, 7-, and 8-Cl) accounted for greater than 90 percent of the total dissolved toxaphene mass. This is consistent with an atmospheric deposition source. In general, the predominance of the lower chlorinated homologs reflects their higher solubility and vapor pressure, which would allow them to preferentially pass the air/water boundary. The presence of lower chlorinated homologs also supports atmospheric deposition, because those compounds are found most commonly in the air (Barrie, Bidleman, et al., 1993).
When the general nature of the particles (suspended particulate matter [SPM], POC) in the Lake Superior water column is examined, several factors are found to be significant. The R2 for the POC model is 1.0, while the R2 for the SPM model is 0.998. Homogenous groups are determined to consist of Stations 22, WA, and 19 (the western stations), and Stations 17, 12, 8, and 1 (the eastern stations). The SPM tends to be higher and the POC lower in the western arm, which is consistent with the higher erosional loading of inorganic clay particles to the western arm (Scholz, 1985).
The POC content varies significantly with the total amount of particulates present. As the amount of particles decreases, the fraction of organic carbon increases. The slope of the linear regression line doubles (R2 = 0.42; slope = -1.116) when the stations from the western arm (22, WA, and 19) are removed (see Figure 3). This may be because the biologically oriented particles (e.g., relatively high in OC) remain fairly constant throughout the lake, while the input of inorganic particulates increased on the western portion of the lake. This is consistent with the western arm of Lake Superior; it constitutes only 5 percent of the lake surface area, but receives up to 60 percent of the sediment load. The main source of these particles is erosion of red clay from the southern shore between Duluth and the Apostle Islands (Scholz, 1985).
Figure 3. Regression of SPM versus POC for Lake Superior. Equation is the computer-generated best-fit linear regression trend line. Data are from all cruises and all stations. N = 35.
Toxaphene concentration in the particulate phase in Lake Superior showed much greater variability over both time and space than the dissolved phase as shown by the lower significance for the term season and the presence of the term station in the GLM model for particulates. The difference exists most likely because the dissolved phase adjusts to changes in ambient air concentration faster than the particulate concentration adjusts to changes in water concentration. This is reflected in the presence of significant differences at the year/time scale, in addition to the season/time scale. The significant statistic differences between individual cruises for the particulate phase also are much less frequent and less significant than those observed for the dissolved phase. This is either because of the higher variability in the individual results or because of the most significant input/loss pathway, which has the most influence on the dissolved pathway. The particulate concentration for 1997 is significantly (p = 0.005) higher than that for 1998. This is consistent with the trend observed for the dissolved phase.
In contrast to the dissolved phase, however, the particulate concentration did significantly vary with location. Station 1 (ave + 95 percent CI = 106 ng/g + 29 g/g), located in the easternmost portion of the lake, is significantly (p = 0.022) higher than Station 22 (ave + 95 percent CI = 43 ng/g + 1 ng/g), which is located in the westernmost portion. This is consistent with the higher sediment load in the western portion of the lake, which tends to dilute the toxaphene concentration on a per gram basis.
Analysis of variability in particulate phase toxaphene homolog concentration reveals similar trends for the 7 and 8 chlorinated homologs as those seen in the total toxaphene particulate phase. For both total and individual homologs, location plays a role in explaining variability in particulate concentration for both total and individual homologs. The amount of variability explained by these models varied greatly, ranging from a low of 17 percent up to a relatively high 68 percent. The differences between homologs were a result of varying chemical properties and their effect on behavior in the particulate phase. The 6-Cl homolog had the highest solubility, and therefore, is bound least tightly to the particle. Although there is a relatively high proportion of 6-Cl homolog in the air and dissolved phase, it is not as readily partitioned into the particle phase as the lower solubility homologs. The 7 and 8-Cl homologs closely mirrored the response of total toxaphene in the particulate phase due to their lower solubility (and therefore greater affinity for the particles) and high prevalence in both air and dissolved water phase. The highest chlorinated homologs showed poor correlation and significance because of their low concentration and inherent greater variability associated with quantifying compounds near the limit of detection.
As with the dissolved phase, concentrations of particulates in 1997 were consistently higher than those of 1998. Differences across stations for the 7 and 8-Cl generally followed the same pattern as total toxaphene, with the western locations having a slightly lower concentration than those in the eastern portion of the lake. Differences across seasons also mirrored the total toxaphene response, with the concentration gradually decreasing over the course of the sample season, and then increasing back to near an initial level in spring. This is consistent with the particulate concentration; it is highly dependant on the dissolved concentration.
When the total toxaphene particulate concentration is plotted against the POC, no correlation (p = 0.326) is observed. This indicates that the toxaphene has not attained steady state conditions with respect to the carbon in the particles, because the partitioning theory suggests that the concentration of a hydrophobic compound on a particle should correlate with the amount of organic carbon in the particles (Karickhoff, Brown, et al., 1979). The lack of correlation supports the inference that the dissolved and particulate concentrations changed at different rates and are not at equilibrium with respect to each other.
When analyzing the homolog distribution in the dissolved phase, neither a factor or covariate was found to be significant. Therefore, the variability in these homologs could not be explained by the factors utilized in this study. The final models for the 7-Cl, 8-Cl, and 9-Cl homologs included the single factor water temperature (p < 0.001; R2 = 0.491 for 7-Cl and 0.542 for 8-Cl homologs) or the single factor cruise number (p = 0.001; R2 = 0.578). The model for the 9-Cl homologs contained the single factor cruise number (p = 0.012, R2 = 0.468). These individual factors give similar results because water temperature and cruise number were highly (p < 0.001) correlated, with lower temperatures in earlier seasons.
When analyzing dissolved toxaphene homolog distribution patterns as a function of water temperature, the 7-Cl homologs increased (p = <0.001), while the 8-Cl homologs decreased (p = <0.001) with increased water temperature. The 9-Cl homologs also showed a negative trend with increasing water temperature, but with a lower significance (slope p = 0.09). The linear regression lines for the 7-Cl and 8-Cl homologs are nearly exact opposites (slopes of 0.0038 and -0.0039). The slope of the 6-Cl homolog is not significantly different from 0 (p = 0.184). Therefore, the mass lost from the 8-Cl homologs is offset by an increase in 7-Cl homolog as temperature increased.
This preferential enhancement of the lower chlorinated homolog with increasing water/air temperature most likely occurs because the ambient air concentration increases simultaneously with temperature. Because air has a higher percentage of lower chlorinated homologs, this increase in air concentration acts to decrease the flux of the lower chlorinated homologs out of the water. This causes the 8-Cl homolog to offgass faster than the 7-Cl homolog as temperature increases, resulting in a lower relative percentage being present. The 7-Cl homolog does not actually inflate with increasing temperatures; rather, it decreases less quickly, appearing to increase relative to the other components. Because the lake was consistently offgassing, the increase in 7-Cl homologs was not the result of increased flux from air to water with higher temperatures.
For particulate toxaphene homolog distribution, the trend is similar to the dissolved phase, with the lower chlorinated homolog (6-Cl) increasing over the course of a year (p = 0.032), and the higher chlorinated homolog (8-Cl) decreasing over the same period (p = 0.024). Significant trends were not observed for the 7-Cl, 9-Cl, or 10-Cl homologs. Therefore, it appears that the mass lost from the 8-Cl is replaced by the 6-Cl homolog. Similar plots (with slightly higher significance) are obtained when homolog distribution is plotted against air temperature. The decline in significance is attributed to the particles that are directly influenced by the dissolved concentration and the water temperature, and only indirectly by the air trends. In addition, there is much more scatter in the data (e.g., lower R2 values for the linear regression lines). This scatter is from increased variability in the particulate concentration, due to a lack of steady state conditions. The negative slope for the 8-Cl homolog distribution with temperature is nearly identical to the value for the dissolved phase, indicating that loss from the dissolved phase drives loss from the particulate phase. Sedimentation also may reduce higher chlorinated homologs in the water column by preferentially removing more hydrophobic fractions.
Examination of the average toxaphene homolog distributions for dissolved, particulate, and sediment revealed that each media and homolog is different from each other. A higher percentage in the lower chlorine homologs (e.g., 6 and 7) generally represents more weathered material (Muir and de Boer, 1993; Fingerling, Hertkorn, et al., 1996). These data could be the result of a higher magnitude of transfer of more volatile components (e.g., the lower chlorinated compounds (Rice, 1993)) across the air/water interface, or the recycling of higher chlorinated homologs in the sediments as the carbon is broken down. The most likely explanation involves a combination of volatilization, which is the most significant transfer pathway for toxaphene in the lake. Recycling has been shown to occur for other hydrophobic organic compounds in the lake (Baker, Eisenreich, et al., 1991; Pearson, 1996; Hietela, 1999). The relatively small confidence intervals on each of the homolog values indicate that the distribution was relatively stable within a given media. The homolog distributions established for Lake Superior by this study are consistent with values reported previously for the Great Lakes (Swackhamer, Pearson, et al., 1998).
Partition coefficients for toxaphene were calculated using the dissolved concentration, or the particulate concentration (in ng/g). Partitioning of an organic compound such as toxaphene between the dissolved and particulate phases in water, can be influenced by the lipophilicity and solubility of the compound and the physical characteristics of the particles (Manahan, 1994). The factors found to be significant in explaining variability in total toxaphene log Kp are station (p = 0.063) and cruise (p = 0.064), producing an R2 value of 0.638. This observed difference across time and space was not consistent with previous assumptions that the partition coefficient is a relative measure that should not change with changing concentrations (e.g., that changes in dissolved phase concentration would be offset by changes in the particulate phase). However, the statistical correlation was relatively weak.
Net air/water fluxes were calculated for each cruise. The flux equation also was used to estimate the percent saturation of the lake based on equilibrium partitioning and the estimated air concentration. Percent saturation is the actual measured dissolved concentration divided by the postulated dissolved water concentration based on given air concentrations. Estimates of saturation are all well above 100 percent. The degree of saturation tends to be the mirror image of the net air/water flux, with the lowest saturation occurring at the time of highest flux out of the water. These two processes are intertwined, with the degree of saturation depending on the amount of volatilization. The finding of highly saturated conditions was consistent with other investigations (Hoff, Bidleman, et al., 1993; Swackhamer, Schottler, et al., 1999).
The factor cruise number (p = 0.041) and water temperature (p = 0.002) were the only parameters included in the optimal GLM model for characterizing variability in flux estimates. The R2 for this model was 0.702. This revealed that time of year and lake conditions are key factors affecting the flux of toxaphene. All results showed a negative flux (e.g., from water to air); therefore, the overall toxaphene concentration in the water decreased, and the lake was a source for toxaphene to the surrounding air. The degree of that contribution varied with the time of year.
The flux out of the water appears to peak in late summer and then to decrease slightly until the October sample. This increase in volatilization from season 1 to 3 is likely the result of the H increase that comes with increasing temperature and the decrease in air concentration in late summer. The increase in H is a result of the greater increase in vapor pressure that occurs compared to the increase in water solubility as temperature increases. The air concentration decreases from a spring high level to a lower fall level. In addition, the dissolved water concentration generally decreases during this period as a result of the increased volatilization (e.g., toxaphene is transferred out of the water and into the air). Volatilization then decreases towards late fall as the air concentration remains low, and the temperature declines causing a decrease in H. The toxaphene concentration trend is exacerbated by the stratified conditions, which prevent dissolved toxaphene from the hypolimnion from mixing with the epilimnion.
In general, because toxaphene was always offgassing, dissolved concentrations in the upper water gradually decreased throughout the year (as was shown by the 1998 data). In spring, volatilization again increased as the lake was mixing for two main reasons: (1) toxaphene was redistributed from the hypolimnion, thereby raising the dissolved concentration; and (2) air temperature increased (which increases H). In 1997, there was a significant increase in dissolved concentration from spring to fall, due to periodic upwelling in the eastern and central basins of the lake. As a result, it destratified the lake and allowed toxaphene from the hypolimnion to be mixed throughout the lake.
The dissolved water concentrations also were used to back-calculate an air concentration that would result in steady state conditions across the air/water boundary. When the dissolved water concentrations from this study were used to generate these air concentrations, they were much higher than previous estimates, which were utilized in estimating the current net air/water flux. However, these estimated air vapor toxaphene concentrations are similar to those used to model the peak input interval for toxaphene in Lake Superior (Swackhamer, Schottler, et al., 1999). These elevated air concentrations facilitated transfer from air to water and provided the past source of toxaphene for Lake Superior.
Because the air levels were much lower, the lake was constantly offgassing toxaphene. The cold water temperatures and low particulate levels in the water limited the amount of toxaphene that can be removed from the lake each year. The high toxaphene water saturation values are an echo of past high toxaphene air concentrations and provide strong evidence that the lake is not at steady state with respect to air/water transfer. In addition, the homogenous concentration and even distribution of toxaphene homologs in the water column strongly support the assertion that the main source of toxaphene to Lake Superior has been atmospheric transfer.
Surface water concentrations varied significantly by year and season. They correlate to the air/water interface temperature. Concentrations are strongly correlated with the thermal state of the lake, being elevated during isothermal conditions (both in spring and fall) as a result of mixing from the hypolimnion. The yearly overall average concentration for all compounds declined slightly from year to year, due to losses from volatilization. Volatilization only occured from the epilimnion during stratified conditions, which is likely to result in a vertical concentration gradient. However, the dissolved phase concentrations did not vary by horizontal location within the lake, which supports the assertion that the main source to the lake was air/water transfer rather than point-source discharges.
There were significant differences in particulate toxaphene concentration across years, seasons, and stations, indicating that the particulate concentration changes over both time and space, depending on local conditions. Concentrations were lower in the western basin as a result of dilution from a higher particulate load.
The total toxaphene Kp gradually increased within a sample year, due to the slow uptake of toxaphene into the particulate phase relative to changes in the dissolved concentration. In addition, the Kp for different homologs showed no significant trend with additional chlorines and does not correlate with organic carbon content of particulates. The lack of a trend/correlation is most likely because the lake is not at a steady state with respect to toxaphene. This imbalance is the result of seasonally changing limnological conditions in the lake and a significant decline in air concentration. OC pesticide Kp values correlated with organic carbon, providing evidence that they are at a steady state with respect to the dissolved/particulate phases.
The net air/water flux of toxaphene was consistently negative, indicating transfer from water to air throughout the year. Volatilization was the main toxaphene loss pathway. As a result of minimal loss pathways and lowering air concentrations over time, the lake was highly saturated throughout the year. This effect is an additional indication that the lake is not at a steady state with respect to total toxaphene.
The dissolved concentrations for total toxaphene and each of the homologs had a normal distribution. The particulate concentrations for toxaphene and toxaphene homologs, expressed on a per gram dry mass basis, are not normally distributed (Kolmogorov-Smirnov p < 0.006; Shapiro-Wilk p < 0.01; n = 43). Therefore, log transformed particulate data are used to examine statistical trends.
Although the SPM data (n = 44) initially were not normally distributed (Shapiro-Wilk p < 0.01), a detailed review of the data indicated that this result was being driven by a high bias resulting from three outliers. When the three outliers were removed from the data, the overall mean did not change significantly (Student's t-test, p = 0.227). When normality was checked without the outliers, the data were found to be normally distributed. Therefore, the SPM data are considered normally distributed.
The POC data were found to have a trimodal distribution based on when the sample was taken. For this reason, when the entire set of POC data were checked for normality, they were shown to be abnormal (Shaprio-Wilk p < 0.01). An ANOVA of POC data by cruise number indicated that there were significant (p < 0.001) differences between cruises, with the spring 1997, spring 1998, and the fall cruises forming separate groups. When the POC data were grouped into these catagories, each was found to be normal (Shapiro-Wilk p > 0.286). Because the homogenous groupings of POC had a normal distribution, and the POC data are not used as a group (but rather are applied to individual concentration data), no transformation was performed on the POC data.
The dissolved concentration phases from Stations 18 and 27 from spring 1998, were collected and extracted, but did not show a measurable response due to problems in the extraction procedure.
The relative percent differences for three duplicate toxaphene Stations (50b, 4/98; 24, 8/98; and 27, 8/98) were above the 50 percent limit established in the QAPP. However, each of these samples were from the particulate phase, and were very close to the MDL. Therefore, small differences in measurement and analysis can lead to very large relative differences between sample concentration estimates. The reported concentrations for each of these stations were consistent with other stations from the period, and a review of collection and analysis records did not reveal any reason for the large differences. Therefore, these samples are flagged but maintained in the database.
The surrogate recovery on four samples (XAD 50b, 4/97; XAD 24, 4/98; particulate 18, 9/97; and particulate 24 8/98) was less than the 50 percent required by the QAPP. However, the surrogate-corrected results from these stations were consistent with similar stations and times. The surrogate-corrected results will be flagged in the data tables, but maintained in the final database. All other data met required QA/QC criteria.
A general univariate linear model (GLM) was used to examine the sources of variability in total toxaphene concentration and for each individual toxaphene homolog (see Table 2). Models were developed by optimizing the significance and correlation of a grouping of factors and covariates. Data for the dissolved and particulate phases were followed by an analysis of homolog distribution, an estimation of water/particle partitioning, and an estimation of fluxes between media.
Parameter | Type | Description |
Basin | Fixed Factor | 1 = Northern basin; 2 = Green Bay3 = Southern Basin |
Station | Fixed Factor | Numerical location designator |
Cruise number | Fixed Factor | Sequential order of major cruises; 1-4 |
Year | Fixed Factor | Sample year; 1997;1998 |
Season | Fixed Factor | 1 = isothermal; 4 = stratified |
SPM | Covariate | Suspended Particulate Matter |
POC | Covariate | Particulate Organic Carbon |
Water temp (°C) | Covariate | Surface water temperature |
The cruise numbers are: 1 = April/May 1997; 2 = September 1997; 3 = March/April 1998; 4 = August/September 1998. The three basins were defined as: Northern basin = Stations 52b, 47, and 41; Southern basin = Stations 27, 23, 18, and 5; and Green Bay = Stations 24 and 50B. These basin groupings were based on relative proximity to each other and circulation patterns within the lake that tend to isolate these areas. The seasons represent spring unstratified conditions (season 1) and late-summer stratified conditions (season 4; used to be consistent with Lake Superior).
Ninety-five percent of toxaphene mass in Lake Michigan is present in the dissolved phase, and three-quarters of that is present as 6-Cl and 7-Cl homologs. Dissolved total toxaphene concentrations are more variable than those observed in Lake Superior and ranged from 0.14 to 0.80 ng/L.
A univariate GLM model was optimized to describe variability in the total dissolved toxaphene concentration. This model had an R2 value of 0.780. The basin term also was significant (p < 0.001) when substituted for station, yielding an R2 of 0.645. This indicates that both time and location played a significant role in explaining variability in the dissolved toxaphene concentration in Lake Michigan, in contrast to the lack of location dependence in Lake Superior. An examination of location differences reveals that the overall average for stations located in Green Bay is significantly lower than either the northern (p = 0.001) or southern basins (p = 0.002). In addition, there is no significant difference between the northern and southern basins (p > 0.88). The data are consistent with the assertion that the open lake locations decrease in concentration later in the year, due to the increased flux from water to air that results from the lower air concentration, higher water temperature, and the reduced effective volume of the lake (only the epilimnion is available to the surface for exchange with air during stratified conditions). The concentration within Green Bay does not appear to vary significantly over the seasons because the hypolimnion in Green Bay constitutes a much smaller percentage of the total water volume in the bay compared with the hypolimnetic/epilimnetic volume ratio in the main basin of the lake. The decline in the main basin is consistent with the negative correlation (slope = -0.01; p < 0.001) between dissolved concentration and water temperature. This correlation is directly related to the fact that higher surface water temperatures were associated with stratified conditions and promoted volatilization. In addition, colder surface water temperatures are associated with isothermal (and therefore elevated toxaphene) conditions.
The high concentrations in the open lake basins in spring were caused by mixing during spring isothermal conditions, bringing toxaphene from the hypolimnion up to the surface. Although there was an apparent decline in overall average dissolved toxaphene concentration from year to year, this trend is not statistically significant. These data are comparable to dissolved concentrations for 1994-1996 (0.38 + 0.12 ng/L) (Swackhamer, Pearson, et al., 1998), and it appears that the overall average concentration in the lake is not significantly changing. This shows that Lake Michigan recently has been near steady state conditions and only recently has begun consistently offgassing toxaphene as the air concentration has decreased.
In general, the pattern for the homologs is similar to that found for total toxaphene, with no significant difference between sample years, high levels in spring, low levels in Green Bay, gradual (but not statistically significant) decline between sample years, and a negative correlation with water and air temperature. Low levels for the 6-Cl homologs for spring 1998 may be the result of the relatively mild winter of 1997-1998, allowing the preferential offgassing of the more volatile 6-Cl homologs. This would not have significantly affected the trends in the total toxaphene concentration because the 6-Cl homolog is only 8 percent of the total dissolved mass.
Variability in the descriptive characteristics SPM and POC for Lake Michigan is explained by three main factors. Both SPM and POC changed significantly as time and location changed on the lake. For example, the POC changed between spring cruises (average = 12.6 percent) and fall cruises (average = 27.9 percent). This may be the result of greater input of inorganic particulates during the spring run-off period and higher primary productivity in warmer periods.
The variability in the particulate phase toxaphene concentration is three times that of the dissolved phase as indicated by the relative 95 percent confidence interval of all the data. The optimal model (R2 = 0.779) describing variability included the factor's cruise number (p < 0.001) and POC (p = 0.006). In contrast to the dissolved phase, there is no significant difference between basins or stations. Therefore, the particulate concentration varies over time, but not over space. In addition, the amount of organic carbon in the particle significantly affects the toxaphene concentration. This is consistent with toxaphene being a hydrophobic compound, and the fact that Lake Michigan is nearly at steady state conditions.
When the concentration trend for total toxaphene was examined, a significant decline between seasons was shown for both sample years. There also was a significantly higher concentration in the spring of 1997 than the remaining three cruises. The dissolved toxaphene concentration for April/May cruises, however, were not dramatically different between years. The particulate concentration negatively correlated with water temperature (slope = -1.14; p <0.001). The spring 1997, results were higher than the average particulate concentrations of 9.4 + 1.2 ng/g for 1994-1996 (Swackhamer, Pearson, et al., 1998). The higher concentrations were obtained across all stations and sample times, ruling out bias from a single, high result. It is not readily apparent at this time why levels were elevated in early 1997.
Both the slope and the intercept for samples collected during isothermal spring conditions are higher than those collected during stratified conditions. This indicates that the particles present in the spring have a higher overall concentration both per gram dry weight and per gram carbon, and that the concentration increases more rapidly as a function of organic carbon. As a result, some of the particles in spring may have been suspended in the water column for the entire winter by isothermal conditions, allowing additional time for the toxaphene to partition from the dissolved phase to the particulate. It is likely that particles present in the summer have just recently been introduced to the system (via runoff or algae growth) and have less time to attain steady state concentrations. The spring 1998 value was low because particles had low levels of organic carbon. In general, the decrease in the particulate concentration over the course of a year is caused by a combination of the decrease in the dissolved phase, the introduction of fresh particles to the system, and the length of time required to attain steady state levels.
The optimal models for describing variability in the particulate phase for toxaphene homologs are similar to those for total toxaphene in that they have no spatial variability. When all particulate concentrations of individual toxaphene homologs are plotted against POC, no correlation is found. However, when data are grouped by stratification status, there are significant trends. The slope of the least squares linear regression line is dramatically larger for unstratified conditions compared to stratified ones. In addition, the stratified data generally has a much lower significance. This stronger correlation for unstratified data shows that the toxaphene in the carbon portion of the particles is similar to steady state conditions.
There are no significant spatial differences for either the dissolved or particulate phase with respect to homolog distribution. There are, however, significant differences between media, with the dissolved phase having the higher prevalence of lower chlorinated homologs. This pattern is consistent with air/water transfer as the main source of toxaphene. The sediment data are included as an overall average from the entire depth of the core. The average homolog distribution of the sediment surface sections (n = 6), is not statistically different from the suspended particulates. The lower chlorinated homologs gradually increased in prevalence with depth in the core, indicating that dechlorination may be occurring over time.
The distribution of the 9-Cl homologs as a fraction of the total is the only group to demonstrate a significant correlation with water temperature. The negative slope of the plot of 9-Cl homolog prevalence versus water temperature indicates that as water temperature increases, the amount of 9-Cl homolog decreases. Though it can be assumed that this transformation results in a higher percentage in the lower chlorinated homologs, the total mass of the 9-Cl homologs was so small that it made the contribution to the other homologs insignificant. The lack of a statistically significant trend in the other homologs is likely the result of scatter in the data. Both the 6-Cl and 7-Cl homologs have a positive, but not significant, slope when plotted against water temperature. In addition, the 8-Cl homologs have a negative slope, but it also is not statistically significant. These general results are consistent with the trend observed in Lake Superior, where higher water temperatures correlated with a higher percentage of lower chlorinated homologs. Although the mean levels showed a correlation with temperature, the lack of statistical significance indicates that the trend was very weak, regarding the variation of individual toxaphene homologs within a given media.
The water particle partitioning coefficient (Kp) for toxaphene in Lake Michigan contributes to the variability in the Kp; it is related to both time and space. The amount of organic carbon in the particulate phase was not found to be significant, when we used this model (p = 0.451 when POC added to other optimal factors). The R2 for this optimal model was 0.816. However, when combined with the factor's season or cruise number individually, the POC was found to be highly significant (p < 0.001). This indicates that the amount of organic carbon in the particulate phase does affect the partitioning of toxaphene, and is consistent with the fact that POC significantly affects particulate toxaphene concentration. However, POC is not included in the optimal model for Kp, because the amount of variability explained by models (R2) including POC is lower than the optimal set of parameters.
When the log Kp's for all data are plotted against POC, no significant correlation is obtained for total toxaphene. However, there is a strong correlation between the POC and the seasonal log Kp, with the spring cruises having significantly higher Kp values than the summer cruises. This is consistent with the fact that the April/May cruise particulate concentration is significantly higher on a per gram carbon basis than the August/September cruise value. The slopes of the plots are similar, but the intercept between least squares regression lines for individual seasons slightly differs from an order of magnitude, which is similar to the seasonal difference in particulate concentration, per gram carbon.
An inter-basin comparison of Kp over the course of the project shows that the southern basin was less than Green Bay (p = 0.016) during stratified conditions. The statistical significance of the difference, however, is relatively poor, because of variability in the results and the small number of observations. The northern basin and Green Bay are not significantly different from each other. The lower value in the southern basin may be the result of the increased particulate loading to the southern basin from the southern shoreline during summer compared to the northern basin, causing a dilution effect.
The water temperature is negatively correlated (p = 0.002) to the log Kp. This negative slope results because the particulate concentration declines faster relative to the dissolved concentration with increasing water temperature as the year progresses. The reasons for the decline in the particulate concentration over time within a year were explained previously. This difference between seasons persisted when the organic carbon partition coefficient (Koc) is estimated. Because the dissolved concentration declined slightly through the year, and the particulate concentration declined precipitously, both as per total mass and on a per gram carbon basis, the estimated Koc is highly dependent on the season.
The variability in the partition coefficients for the individual toxaphene homologs depends on which homolog is being examined. The partitioning is most strongly affected by changes in time (as represented by the factor cruise number) and by location within the lake for the most prevalent homologs. However, the significance of the correlation decreases (e.g., the likelihood of an actual correlation increases) with increasing chlorine content of the homologs. This is consistent with partitioning theory, which states that higher chlorinated compounds are more strongly bound to the organic fraction of a particle. The highest chlorinated homologs did not have any significant factors, indicating that variability in these values could not be explained by factors in this study.
The average Kp across each of the toxaphene homologs generally increases as the number of chlorines increases. This is consistent with partitioning theory, differs from the results from Lake Superior, and indicates that Lake Michigan is more characteristic of steady state conditions. In addition, there is a downward trend within each homolog across seasons for all but the 9-Cl and 10-Cl homologs. This is the result of the difference in particulate concentrations over time.
Toxaphene may enter Lake Michigan by one of three main routes: (1) direct application (intentionally, inadvertently, or criminally) in the past; (2) inflow from major tributaries; and (3) exchange across the air/water boundary. There was no record of toxaphene ever being used directly on Lake Michigan, and tributaries were assumed to be below detection limit due to their relatively small surface area, limited past use of toxaphene in the Midwest, and lack of current applications. A recent study of sediment near seven pulp and paper mills along rivers that flow into Lake Michigan suggested that at present, they are not significant sources of toxaphene to the lake (Shanks, McDonald, et al., 1999). It was concluded that the majority of the toxaphene in Lake Michigan was from exchange across the air/water boundary because the likelihood of other sources had been reduced.
Toxaphene may leave Lake Michigan by one of four main routes: (1) surface water outflow; (2) deposition to sediments; (3) degradation within the lake itself; and (4) volatilization. The only major surface water outflows for Lake Michigan are the Chicago river in the southern basin and the Straits of Macinac at the northern end of the lake. These two sources combined account for an outflow of 1600 m3/second (USEPA and Environment Canada, 1995). Using the average dissolved toxaphene concentration for each year sampled (mean + 95 percent CI; 0.404 + 0.222 ng/L for 1997; 0.353 + 0.104 ng/L for 1998), this yields an outflow of 20.2 kg/yr for 1997, and 17.3 kg/yr for 1998. The total mass of toxaphene in the water of Lake Michigan is 1835 kg, based on a total volume of 4290 km3 (USEPA and Environment Canada 1995) and an average toxaphene concentration of 0.378 ng/L (overall average of all data from this study). Therefore, loss via surface water outflow is about 1 percent per year.
Values for the air side mass transfer coefficient (ka) and air vapor concentration (Ca,v) are the same as those used for Lake Superior. The values used in the flux calculations are averaged over an individual cruise to maintain consistent time intervals for all parameters. The air concentrations used were monthly averages, ka was chosen to match the reported seasonal values (Hoff, Bidleman, et al., 1993), and the air/water interface temperatures were based on Schwab, et al., 1999.
The final net air/water flux estimates indicate that toxaphene is nearly always offgassing from Lake Michigan, as represented by the negative flux values. In May 1998, review of Station 50 indicates that this was the only station that showed a positive flux. This was likely the result of the combination of very low water concentration (0.11 ng/L), high air concentrations, and relatively low temperature. The percent saturation of the lake also may be estimated by setting the flux equal to zero, generating a water concentration based on the estimated air concentration, and then comparing that water concentration to the measure value for that location.
The flux out of the stations in Green Bay was significantly (p = 0.001) less than that for the open lake basins. This occurs because the dissolved concentration in Green Bay is lower compared to the open lake. As a result, during summer stratified conditions, the open lake locations lose significantly greater amounts of toxaphene than those in Green Bay, resulting in all areas of the surface of the lake having similar concentration.
The negative correlation of the dissolved phase concentration with both water and air temperature is likely the result of the increase in H that occurs with increasing temperature. In addition, the range of day/night temperature change is buffered by the lake. The increase in H promotes the movement from water to air by increasing the second term in the net air/water flux equation (e.g. Cw,d x (H/RT)). The increased transfer from water to air acts to lower the surface concentration.
The high estimates of Lake Michigan's saturation are in contrast to previous reports (Swackhamer, Schottler, et al., 1999), which showed summer saturation to be 50 percent and the annual average saturation to be 82 percent. This discrepancy results from the significantly lower air vapor concentrations used in this study compared to previous estimates. The lower air concentration results in higher volatilization and saturation values. Progressively lower air concentrations are expected as production gradually diminishes (due to regulatory restrictions) and use declines, and as historic stockpiles and applications degrade/weather.
It has been proposed that northern Lake Michigan has received significant non-atmospheric toxaphene input in the past (Pearson, 1996). Possible sources for this non-atmospheric input may include the pulp and paper industry, residuals from past piscicidal uses in the basin, trace releases from hazardous waste sites (MDPH, 1997), agricultural runoff, or incidental, accidental, or unreported spills in the past. Data from the water column in this study (which represents more recent conditions) are consistent with an atmospheric source for the toxaphene in Lake Michigan, and a lack of current point-source inputs. This is based on the lack of spatial differences for particulate toxaphene, lower dissolved concentrations in Green Bay, a homolog distribution similar to the air profile, year-round offgassing, highly saturated conditions, and seasonal variations that correlate to a gradually declining, non-replenished toxaphene sink. Point-source inputs (e.g., paper mills, runoff, spills, releases from hazardous sites, and one-time piscicidic application) would have to have been massive and sufficiently long ago to allow the lake to have become evenly mixed.
Toxaphene concentrations in individual sections reflected the highly variable depositional environment in Green Bay. The analytical limit of dectection was reached in cores 52b, 24, and 137, and a significant drop-off obtained in core 32 (e.g., most of the toxaphene inventory was likely captured). Core 137 had more variable trends with depth due to mixing and the possible presence of sand lenses. Core 50b reached a tight clay layer at 16 cm, indicating that the entire length of the core has been mixed, while core 13 still had significant levels of toxaphene at 20 cm.
Sediment concentration was multiplied by the average mass sedimentation rates to determine the accumulation rate for toxaphene in ng/m2 yr. For core 24, the sedimentation rate appeared to be constant, allowing for the use of a single average value. In core 52b, there were two distinct rates, with a higher slope in more recent history. This was likely due to an increase in the eutrophic status of the bay with increasing discharges from sewage, industry, and growing population. The individual average slopes for these two zones were used, with the average value of the two slopes applied to the section at the slope change. Since the 210Pb horizon was defined in each of these cores, a single focusing factor was used to generate the estimates.
The total toxaphene mass accumulations gradually decreased with depth in the core, and the surface results were only slightly lower than those obtained for Lake Superior. The lower concentration in the water column of Lake Michigan was offset by the higher deposition rate, resulting in fairly similar accumulation rates. The deepest measured toxaphene accumulation regions (2-2.5 cm at Station 52b; 2-3 cm at Station 24) are associated with sediment from about 60 years ago (e.g., late 1930s). This correlates with the time that toxaphene was first introduced and used, and indicates that toxaphene began appearing in Lake Michigan almost immediately after large-scale agricultural use started.
These accumulations are significantly lower than previously reported values (Pearson, Swackhamer, et al., 1997) as a result of lower sediment surface concentrations (15-45 ng/g versus 2-5 ng/g). This decline in sediment concentration is likely the result of a combination of three main factors. One factor is an actual decline in concentration as a result of the reduction in the total toxaphene burden in the water column of the lake over the 6 years between samples. Secondly, most of the samples were taken in Green Bay rather than in the open lake basin. Furthermore, differences in quantitation methods may have reported different concentrations by rejecting given peaks in the chromatogram, using the current method that the previous method would have included as acceptable toxaphene.
Total toxaphene inventories are used to investigate whether an area has had significant excess toxaphene deposition compared to other areas of the lake. For example, the series of five sediment cores taken along Green Bay appear to show a decrease in inventory from the southern end of the bay (e.g., at the outlet of the Fox River) to the outlet at the northern portion of the lake. As with the accumulations, these inventory estimates are significantly lower than previously reported results (Pearson, Swackhamer, et al., 1997).
Over 95 percent of the mass of toxaphene in the water column of Lake Michigan is in the dissolved phase, with a homolog distribution that reflected air/water transfer as the main source. The dissolved concentration was significantly lower in Green Bay compared with the open lake, but there was no significant difference in dissolved toxaphene concentrations between the northern and southern basins. Concentration also varied over seasons, showing higher levels in the open lake in spring isothermal conditions, and is correlated to air/water interface temperature. This is consistent with the trend observed in Lake Superior, whereby hypolimnetic toxaphene is mixed to the surface during isothermal conditions.
The spring (isothermal) total toxaphene particulate concentrations, on a mass per gram carbon basis, are nine times higher than the summer (stratified) concentration. This disparity is the result of the long time required for the particles to reach steady state conditions with respect to dissolved toxaphene. Both the dissolved and particulate concentrations in Lake Michigan are much less than those observed for Lake Superior.
The homolog distribution of each of the media samples was significantly different, with the sediment having the highest percentage of 6-Cl homologs. The dissolved phase had the greatest proportion of 7-Cl homologs, and had a distribution that mimicked the air phase. The increase in the higher chlorinated homologs in the particulate phase reflects the preferential volatilization of the more soluble homologs.
There was a strong correlation with the seasonal log Kp and POC, with the spring cruises having a significantly higher Kp value. In addition, the southern basin was significantly lower than the rest of the lake, due to the increased particulate loading from high erosion zones. Consistent with general partitioning theory, the average log Kp, and the correlation with POC increased as the number of chlorines increased for the toxaphene homologs.
Air/water flux was the most significant loss route for toxaphene from Lake Michigan. The flux was from water to air throughout the year, and was driven by the combination of high water concentrations and decreasing air concentrations. This resulted in saturated conditions throughout the year with respect to ambient air levels.
Although sediment cores were highly mixed and difficult to date, some trends could be estimated. Toxaphene inventory in the sediment of Green Bay was found to demonstrate a distinct north/south concentration gradient, with highest sediment inventories in the southern portion of the bay. Samples taken at the northern end of Green Bay were similar to cores taken in the open northern basin, indicating that the bay is not likely to be a significant source of toxaphene to the main lake.
Although both lakes are large, oligotrophic bodies of water, significant differences exist between the amount of SPM and the composition of sediment. Lake Superior had a much lower burden of suspended particles and less carbon in sediment. The lower SPM for Lake Superior is due to a combination of the low degree of erosion that results from the minimal development, hardrock watershed base, and the lower primary productivity. Spatial differences in SPM for both lakes (e.g., higher SPM in eastern Lake Superior and southern Lake Michigan) are related to erosional/tributary inputs from the watershed and proximity to population centers. Lake Superior has a stronger relationship between SPM and POC (slope p < 0.001) compared to Lake Michigan (slope p = 0.018). This was most likely because Lake Michigan has more varied particulate input due to shoreline and watershed characteristics along with higher primary productivity.
There also was a slight difference in the measured seasonal water temperature, with Lake Superior being colder than Lake Michigan. This is likely due to its more northern location, greater volume, and is related to its lower SPM (less particles to absorb light energy). Satellite data and measurements from buoys in the lakes confirmed that the water surface temperature of Lake Michigan was consistently warmer than Lake Superior for the past 5 years (Schwab, Leshkevich, et al., 1999). The similarity in air temperature between lakes is consistent with earlier assumptions that air masses over each lake were alike, and supports the use of a single air concentration for a given time when estimating flux.
There was a significant difference between the two lakes with respect to toxaphene content. Although both lakes received toxaphene input from the same pathway (e.g., exchange across air/water boundary (Hoff, Bidleman, et al., 1993; Pearson, Swackhamer, et al., 1997)), the lower temperature and smaller removal routes in Lake Superior have acted to maintain a higher average toxaphene concentration over time (Swackhamer, Schottler, et al., 1999). However, both lakes were offgassing toxaphene during all cruises. The saturation values in excess of 100 percent occur because the air concentration has diminished faster than the lakes can release toxaphene from the water. In effect, the current high water concentrations are reflections of past higher air concentrations.
The dissolved phase did not show spatial variation in either Lake Superior or in the main basin of Lake Michigan. The only location difference that is statistically significant was found in Green Bay. As previously discussed, these lower values are thought to be the result of a lack of influx of toxaphene from the hypolimnion during isothermal conditions. The lack of spatial variation in the main basin is consistent with both atmospheric deposition as the main source of toxpahene to the lakes and the absence of a current non-atmospheric source. Both lakes display similar temporal trends in dissolved concentrations, reflecting lower levels during stratified conditions, higher levels during isothermal conditions, and gradually decreasing overall levels over time.
The particulate toxaphene concentration was significantly higher in Lake Superior on both a per gram dry mass and per gram carbon basis compared to Lake Michigan. This is consistent with the higher dissolved concentration. Lake Michigan had a higher particulate toxaphene concentration on a volumetric basis (e.g., per L) due to higher SPM. The particulate total toxaphene concentration for Lake Michigan during isothermal conditions (247 ng/g carbon) was similar to the value for Lake Superior, indicating that particles that were retained in the system for a period of several months reached a similar concentration. The particulate concentrations in Lake Michigan are positively correlated (p < 0.02) with POC when individual seasons are considered. In contrast, Lake Superior particulate concentrations are not correlated (p = 0.326). This supports the inference that Lake Michigan was nearly at equilibrium regarding dissolved/particulate partitioning, while Lake Superior displays conditions that are not consistent with equilibrium.
Differences in particulate concentration also were reflected in the water/particle partitioning coefficients (Kp and Koc), which were significantly (p < 0.012) lower for Lake Michigan based on either mass or carbon content. One reason for this difference could be that only surface samples were collected. If water and particles from the hypolimnion were collected (e.g., a zone not subject to volatilization where particles have long retention times), the estimated K values would likely be stable throughout the year and be similar to values obtained during isothermal conditions. The slightly different homolog distribution of the two lakes, with the higher amount of lower chlorinated homologs in Lake Michigan (which are more water soluble) causes the coefficient to be lower. This also may account for the differences in particulate concentration.
Sediment samples in the two lakes were collected in significantly different regions (Lake Superior stations were in the open lake basin, while five of six Lake Michigan stations were located in Green Bay, which is not representative of the entire lake). Consequently, interlake comparisons of the sediment data from this study are not valid.
A review of the homolog distribution for each lake showed that there was significantly more of the lower-chlorinated homologs in Lake Michigan compared with Lake Superior for both the dissolved and particulate phases. Both phases also were enhanced in lower chlorinated homologs compared with the technical standard, because of the preferential uptake of the more soluble homologs across the air/water interface. Lake Michigan had a higher percentage of lower chlorinated homologs as a result of the combination of its greater sedimentation rate and degradation in the sediments. The higher sedimentation rate in Lake Michigan preferentially removes the higher chlorinated homologs. This removal of higher chlorinated homologs increases the amount of lower chlorinated homologs in Lake Michigan because the homolog concentration is relative to the total. Degradation in the sediments shifts the homolog distribution towards the lower chlorinated homologs. The important role of sedimentation in determining lake-wide homolog distribution is apparent because most significant differences in the particulate phase are in the higher chlorinated homologs.
Both Lake Michigan and Lake Superior are large, oligotrophic lakes that contain toxaphene. However, a comparison of measured parameters from this study revealed several significant differences. Lake Superior had a significantly higher concentration of toxaphene in the dissolved and particulate water phase due to its lower temperature and limited toxaphene removal routes. Both the Kp and Koc for total toxaphene were lower for Lake Michigan due to slight differences in homolog distribution. Lake Michigan had a higher relative percentage of lower chlorinated homologs due to its higher temperature and greater sedimentation rate.
Although Lake Superior retains a greater absolute burden of toxaphene than Lake Michigan, both lakes were consistently over-saturated and offgassing toxaphene during all periods sampled. This is the result of the continuing decline in air concentration of toxaphene, and the fact that the water takes longer to adjust to changing inputs than air. Volatilization occurred only from the epilimnion during stratified conditions. Because Lake Michigan stratified more intensely (e.g., a more shallow epilimnion) and with a more consistent yearly pattern compared with Lake Superior, the concentration gradient over seasons in Lake Michigan was greater and followed a more consistent pattern.
A previously published mass balance model for toxaphene in Lakes Michigan and Superior (Swackhamer, Schottler, et al., 1999) was used as the starting point for examining trends in current and potential future data. That model was based on an earlier mass budget approach (Hoff, Bidleman, et al., 1993) and adapted to operate in the STELLA software program to allow for ease of manipulation and calibration. The best-fit atmospheric input function from the beginning of toxaphene use (1950) was constructed, using the production history and available past water and air data. The final model used here was updated based on data from this study (sediment/water partition coefficient, current air concentration) and other recent publications (Glassmeyer, Brice, et al., 1999). In addition, the time scale was extended to the year 2050, to examine potential trends in water and sediment concentrations. The potential impact of particular input values were assessed by conducting sensitivity analyses.
The model follows a basic mass balance approach that estimates the overall change in toxaphene concentration with time by subtracting outputs from inputs in a time-step function. In general, the following inputs and losses were considered in the mass balance model:
dC/dt = (wet deposition + dry deposition + gas absorption + inflow + non-atmospheric inputs) -(sediment accumulation + volatilization + outflow + degradation).
Wet and dry deposition are small in the Great Lakes (Hoff, Bidleman, et al., 1993). As a result, input parameters are estimated from literature values of toxaphene in rain and annual average rainfall from toxaphene concentrations in air particulates, and a representative deposition velocity. Gas absorption and velocity is determined as a net air-water exchange flux from air and water data using the two-film model (Liss and Slater, 1974) as applied by Eisenreich and others (Achman, et al., 1993; Hornbuckle, et al., 1993; Hornbuckle, et al., 1994; Hornbuckle, et al., 1995; Hoff, et al., 1993b). Sedimentation is determined from sediment analyses of toxaphene and accumulation rates from 210Pb dating (Robbins and Edgington, 1975). Degradation rates are taken from those reported by Pearson, et al. (1997), and Howdeshell and Hites (1996). Groundwater flow, other tributary inputs, and losses are thought to be negligible and, therefore, are ignored.
Many of the factors used in the model are assumed to be constant throughout the year. Most of the factors are used without alteration from previous versions of the model. The time factor was expanded from 1950-1995 to 1950-2050, and the Kd was estimated as the average surface sediment concentration (ng/kg) divided by the average water concentration (ng/L) using the values from Lake Superior from this study. Values for Lake Michigan were estimated based on differences in the carbon content of particles between lakes (e.g., 1.36). Use of these values will account for any recycling from the sediments back into the water column by estimating the total net transfer between water and sediments.
Although it was shown previously that the surface water/particle partition coefficient (Kp) was variable over the year, the Kd used here represents the total net partitioning between water and sediments. Because both the sediment and the hypolimnion concentrations are essentially constant throughout the year, the Kd also will be constant.
There also were several parameters that are allowed to vary on a quarterly basis. These generally varied as a function of temperature. The values for H are slightly less than the cruise-specific average values generated in previous chapters due to the higher temperatures observed during 1997-1998. The values are based on long-term average temperatures, which are better estimates when evaluating trends over several decades.
Because the toxaphene in the Great Lakes is delivered via atmospheric transport (Hoff, Bidleman, et al., 1993; Swackhamer, Schottler, et al., 1999), the air vapor concentration (Ca,v) was a critical parameter in modeling trends. This model utilizes a curve based on total toxaphene production over time to estimate ambient air concentrations over the lakes (e.g., higher production = higher air level). The air concentration values in the model are consistent with measured results from 1995 to 1998 (Glassmeyer, Brice, et al., 1999), and with the work of other researchers on this study (James, McDonald, et al., 2001).
The model is designed to produce an estimate of water concentration, sediment concentration, and sediment accumulation for the period from 1950 to 2050. These estimates then are checked against measured values to determine the validity of the input parameters. A key assumption in assessing future concentration is that the shape of the air concentration curve will remain predictable over time. To address several possible scenarios, the model was run using different period (e.g., 2000-2050) air concentration curves. Estimated air concentrations from 1950-2000 are the same in all scenarios. The different conditions examined include the gradual elimination of all toxaphene in air by the year 2020 ("Ban"), maintenance of the current air concentrations through 2050 ("Current"), and three future air concentrations between these two boundary scenarios (for a total of five scenarios). By using a range of possible input air levels, a more comprehensive analysis of potential future concentrations is developed. Final results for the year 2050 for each of these scenarios shows that the air concentration input function has a significant effect on the estimated water and sediment concentration in each lake. Each of the curves are similar in shape over time (e.g., an exponential decay), with different final concentrations depending on the amount of toxaphene present in the air.
The plot of total toxaphene concentration over time in the water column assuming current air concentrations are maintained through 2050, shows that levels generally mirror the average air concentration, with a peak occurring in 1977 for Lake Michigan and in 1979 for Lake Superior. Both lakes reach a similar maximum concentration, but Lake Michigan loses its toxaphene more rapidly than Lake Superior. This is most likely because of its higher average temperature (which shifts the H value up slightly, resulting in greater flux out of the water) and greater sedimentation.
A review of the modeled sediment concentration of toxaphene over time shows that Lake Michigan is initially higher than Lake Superior, and is lower after the early 1980s. This was because Lake Michigan had a much higher sediment deposition rate. As volatilization and sedimentation gradually decreased the dissolved water concentration in Lake Michigan, the sediment concentration also decreased. The lower sediment concentrations after the 1980s shows that the dissolved concentration decreased in Lake Michigan faster than it did in Lake Superior. Concentration is used to compare toxaphene between lakes because toxaphene has a relatively low (<105) Kp (Pearson, et al., 1999).
The modeled results for dissolved toxaphene in water for individual lakes for 1997-1999 are compared to the measured results for Lake Superior and Lake Michigan. In general, there is consistency between the values. The measured values for stratified conditions are significantly lower than the estimated values because the model utilizes the entire volume of the lake to generate concentration estimates. This situation mimics isothermal conditions in the lakes. It has previously shown that volatilization during stratified conditions occurs only from the epilimnion, which has much less volume (and therefore would demonstrate a greater concentration drop for a given mass loss to volatilization).
A comparison of the focus-corrected measured sediment concentration for Lake Superior (J. McDonald and R. Hites, Indiana University, 2000, unpublished data) to the model results for the period 1950-1998, shows that results were fairly similar. Peak concentrations in measured data are 5-10 years later than predicted by the model, with the data from Station 17 showing the closest fit in both amplitude and timing of the peak concentration, and in the overall shape of the concentration gradient. However, the overall concentration trend for each station is relatively similar to the predicted values. This consistency between measured and model data for sediment concentration supports the validity of the model and indicates that virtually all toxaphene mass may be accounted for using current parameters (e.g., no non-atmospheric input). Therefore, the assertion that Lake Superior does not receive non-atmospheric input of toxaphene is supported by this model, consistent with previous reports (Swackhamer, Schottler, et al., 1999). Sediment samples from Lake Michigan were intended to test the impact of Green Bay, and were not representative of the main basin of the lake. Therefore, measured values from Lake Michigan are not compared to model results.
The toxaphene concentration decline also is used to estimate a half-life for toxaphene for each lake. The slope that resulted from the plot of ln(A/Ao) versus time is used to estimate the half-life, where A is the toxaphene concentration at time t, and Ao is the initial concentration. These data provide a better estimate of the actual half-life in the water column because they cover a larger amount of time. In contrast, data from this study only provide a slope for a 2-year period. The estimated values for Lake Superior and Lake Michigan are 19.5 (R2 = .996) and 9.5 (R2 = .990) years, respectively. These results support the observation that Lake Michigan has declined in toxaphene concentration quicker than Lake Superior.
Previous versions of the model identified the air concentration, H coefficient, air-side mass transfer coefficient, and water/particle partitioning coefficient as the model parameters having the greatest influence on final predictions for water and sediment toxaphene concentrations (Swackhamer, Schottler, et al., 1999). Parameters were varied systematically to determine their impact on final concentration estimates for both lakes.
Factors that have minimal impact on final toxaphene concentrations include the air vapor/particle partition coefficient, the concentration in rain, and the dry deposition velocity. Changing any of these inputs by as much as a factor of three results in a change in the final toxaphene concentration in water or sediment of less than 10 percent from the original estimate. These factors did not play a significant role in determining long-term toxaphene concentration in the lakes. It is consistent with the conclusion that the main input pathway is vapor exchange across the air/water boundary.
Alterations in the ambient air toxaphene concentration causes significant changes in the estimated water and sediment concentrations. The response is correlated positively across the entire time used in the model. An increase in the air concentration increases the toxaphene available for transfer across the boundary; thereby increasing the water concentration. In turn, this leads to increased deposition to the sediments.
Changing the air-side mass transfer coefficient has a dramatic effect on the historical levels of toxaphene in both lakes, but did not dramatically effect the estimates of the current concentration. Therefore, the toxaphene enters the lakes faster and is volatilized. This rapid offgassing causes the water concentration to decline, and to be roughly equivalent to measured values by the late 1990s. A similar trend is observed for the sediments, with a higher maximum concentration in the 1970s, but a rapid decline during the 1980s. This results in a higher total toxaphene inventory for sediments across the lake.
The final parameter that has a significant effect on the air/water transfer of toxaphene is the H value. Because H varies as a function of temperature, it also takes into account fluctuations in the temperatures between seasons and lakes. It is inversely related to the water and sediment concentration. An increase in H inflates the partitioning of toxaphene from water into air relative to previous conditions, thereby lowering the water concentration. A decrease in H has the opposite effect, resulting in an increased propensity for toxaphene to remain in the water phase. The magnitude of the change in current toxaphene concentrations is similar to that observed for air concentration.
Other values that can significantly alter the sediment concentration profile are the focusing factor, sediment mass deposition rate, and the water/sediment partition coefficient (Kd). However, because the value for Lake Superior is based on measured values, and the organic carbon adjustment for Lake Michigan is similar to the factor used in previous versions of the model, their uncertainty is relatively low. Although they do affect sediment profiles, their effect on total water concentration is relatively minor, because sedimentation is a small removal pathway for toxaphene compared to volatilization.
The main contributions of this study to the general knowledge in the field of toxaphene in the Upper Great Lakes include:
- The seasonal nature of toxaphene concentration in the air and water of the Upper Great Lakes was characterized by variation in the water/particle partitioning homolog distribution, net air/water flux, and consistent saturated conditions for both Lake Michigan and Lake Superior.
- There were no non-atmospheric sources of toxaphene to Lake Superior and the current elevated concentrations were a result of limnological characteristics of the lake.
- Based on both water and sediment results, it is unlikely that Green Bay is a source of toxaphene for the main body of Lake Michigan.
- A previously published mass-balance model for Lake Superior and Lake Michigan was refined and used to examine current and future trends of toxaphene. The field data generally supported the model. The half-life for toxaphene was estimated, seasonal trends were examined, and time trends were modeled for concentration in water and fish regarding possible public health advisories.
Many crucial characteristics of the lakes' chemistry were shown to be seasonal. The surface concentration of toxaphene in both Lake Superior and Lake Michigan fluctuated significantly throughout the year. Concentrations tended to be highest during isothermal conditions, when toxaphene from the hypolimnion is mixed throughout the entire lake. After stratification, volatilization occurs only from the epilimnion. This smaller volume available for mass transfer results in greater concentration changes than would be expected if the entire lake volume were available. It is expected that the hypolimnetic toxaphene concentration remains relatively stable, declining slightly with the fall turnover as more dilute epilimnetic water is mixed throughout the lake. Over time, the lakes lose toxaphene as air concentrations continue to decline. However, the decline in water concentration does not follow a smooth curve; rather, it descends faster during stratified conditions resulting in the observed seasonal variations. As a result of the currently low air concentrations, each lake is super-saturated with toxaphene with respect to air, and therefore, is consistently offgassing (e.g., the net air/water flux is negative). This volatilization, combined with sedimentation, has altered the homolog distribution in Lake Michigan and Lake Superior. Lake Michigan loses its 8-Cl homologs faster than Lake Superior and appears enhanced in 6-Cl homolog. Water particle partitioning also varies as a function of season. In Lake Superior, the Kp gradually increases through the stratified season as particles equilibrate with dissolved toxaphene. The Kp decreases during fall turnover, due to the resulting increase in dissolved concentration. In Lake Michigan, particulate concentration is highly correlated to organic carbon and season, with much higher values early in the year. These characteristics are very important to understand the chemodynamics of the lakes. Because each variable depends on time of the year, it is crucial that sampling is conducted at appropriate times.
There are several factors that reinforce the conclusion that there is no non-atmospheric source of toxaphene to Lake Superior. They include a lack of spatial differences in dissolved concentration, seasonal fluctuations which are highly correlated to major limnological events (e.g., turnover), similar concentrations in sediments taken at three locations across the lake, a homolog fingerprint consistent with the characteristics of toxaphene in the air, and the success of the mass-balance model to account for toxaphene mass.
Considering all toxaphene data for Lake Michigan from this study, it does not appear that Green Bay is a source of toxaphene for the northern basin. This conclusion is based on the lack of spatial differences for particulate toxaphene, lower dissolved concentrations in Green Bay, a homolog distribution similar to the air profile, year-round offgassing, highly saturated conditions, and seasonal variations that correlate to a gradually declining, non-replenished toxaphene sink. Historical non-atmospheric deposition may have occurred as a result of runoff from the Fox River. However, this concentration gradient did not extend out of Green Bay, indicating that the input did not affect the open basin of the lake, and did not provide a significant non-atmospheric source for the northern basin.
One of the most important aspects of environmental research is that it provides data that form the basis for making sound predictions of future conditions. This allows resource managers the potential to have valuable insight into problems before they become crises. We expanded the model used for this study to provide estimates until the year 2050. As expected, the air concentration input was the most crucial parameter in determining long-term toxaphene concentration in water and sediment. The time frame provided data to produce an estimate of the half-life of toxaphene in each lake. Lake Superior had a significantly longer half-life because of its low temperature, minimal sedimentation, and low productivity. The field data for dissolved toxaphene correlated with model estimates for isothermal conditions, but were significantly lower for stratified conditions. This was due to differences in the volume of the lake available for volatilization during stratified conditions. Future predictions of water concentration show that it is unlikely that toxaphene water levels will decrease sufficiently enough to allow for the unrestricted consumption of game fish by the general public. However, this assertion is limited by the time frame examined, uncertainty associated with toxaphene risk characterization, and the complexity of generating sound fish consumption advice. In addition, toxaphene may not even be the most signficant toxic factor in generating protective exposure levels.
References:
Achman DA, Hornbuckte KC, et al. Volatilization of polychlorinated biphenyls from Green Bay, Lake Michigan. Environmental Science and Technology 1993;27(1):75-87.
Appleby PG, Oldfiekt F. The calculation of 210Pb dates assuming a constant rate of supply of unsupported 210Pb to the sediment. Catena 1978;5:1-8.
Atkins DH, Eggleton AE. Studies of atmospheric washout and deposition of BHC, dieldrin, and DDT using radio labeled pesticides. International Atomic Energy Agency, Vienna, Austria, 1971:521.
Agency for Toxic Substances and Disease Registry. Toxicological Profile for Toxaphene. U.S. Department of Health and Human Services, Atlanta, GA, 1996:215.
Baker JT, Eisenreich SJ, et al. Sediment trap fluxes and benthic recycling of organic carbon, PAHs and PCB eongeners in Lake Superior. Environmental Science and Technology 1991;25(3):500-509.
Barrie L, Bidleman T, et al. Atmospheric toxaphene in the high Arctic. Chemosphere 1993;27(10):2037-2046.
Bennett EB. Characteristics of the thermal regime of Lake Superior. Journal of Great Lakes Research 1978;4(3-4):310-319.
Bidleman T, Wideqvist U, et al. Organochlorine pesticides and polychlorinated biphenyls in the atmosphere of southern Sweden. Atmospheric Environment 1987;21(1):641-654.
Bidleman TF, Oiney CE. Long range transport of toxaphene insecticide in the atmosphere of the west North Atlantic. Nature 1975;257:475-477.
Chu I, ViBeneuve D, et al. Toxicity of toxaphene in the rat and beagle dog. Fundamental and Applied Toxicology 1986;7(3):406-418.
de Boer J, Wester PG. Determination of toxaphene in human milk from Nicaragua and in fish and marine manomals from the northeastern Atlantic and the North Sea. Chemosphere 1993;27(10):1879-1890.
Desaiah D, Koch R. Influence of solvents on the pesticides inhibition of ATPase activities in fish and insect tissue homogenates. Bulletin of Environmental Contamination and Toxicology 1977;17(1):74-78.
Eakins JD, Morrison RT. A new procedure for the determination of 210Pb in lake and marine sediments. International Journal of Applied Radiation Isotopes 1978;29:531-536.
Eisenreich SJ, Hollod GJ, et al. Accumulation of polyehiorinated biphenyls (PCBs) in surficial Lake Superior sediments: atmospheric deposition. Environmental Science and Technology 1979;13(5):569-573.
El-Sebae AH, Zeid MA, et al. Status and environmental impact of toxaphene in the third world: a case study of African agriculture. Chemosphere 1993;27(10):2063-2072.
Fingerling G, Hertkom N, et al. Formation and spectroseopic investigation of two hexachlorobornanes from six environmentally relevant toxaphene components by reductive dechlorination in soil under anaerobic conditions. Environmental Science and Technology 1996;30(10):2984-2992.
Glassmeyer S, Brice K, et al. Atmospheric concentrations of toxaphene on the coast of Lake Superior. Journal of Great Lakes Research 1999;25(3):492-499.
Glassmeyer ST, De Vault DS, et al. Toxaphene in Great Lakes fish: a temporal, spatial, and trophic study. Environmental Science and Technology 1997;31(1):84-88.
Great Lakes Health Effects Programs. Toxicity profiles for Great Lakes chemicals. Department of National Health and Welfare, Environmental Health Center, Ottowa, Canada 1992:623.
Golden KA. Organochlorine confounds in Lake Michigan sediments. Civil Engineering, University of Minnesota, Minneapolis, MN, 1994.
Gunderson EL. FDA total diet study, April 1982-April 1984, dietary intakes of pesticides, selected elements, and other chemicals. Journal of the Association of Official Analytical Chemistry 1988;71(6):1200-1209.
Hietela A. Recycling of contaminants in the benthic nepheloid layer of southern Lake Michigan. Environmental Health, University of Minnesota, Minneapolis, MN, 1999.
Hill EF, Heath RG, et al. Lethal dietary toxicities of environmental pollutants to birds. United States Fish and Wildlife Service, Washington DC, 1975;61.
Hoff JT, Mackay D, et al. Partitioning of organic chemicals at the air-water interface in environmental systems. Environmental Science and Technology 1993;27(10):2174-2180.
Hoff RM, Bidleman TF, et al. Estimation of PCC loadings from the atmosphere to the Great Lakes. Chemosphere 1993;27(10):2047-2055.
Hoff RM, Muir DCG et al. Measurement of PCCs in air in southern Ontario. Chemosphere 1993;27(10):2057-2062.
Horne AJ, Goldman CR. Limnology. New York, NY: McGraw Hill, Inc., 1994. 480 pp.
Howdeshell MJ, Hites RA. Historical input and degradation of toxaphene in Lake Ontario sediment. Environmental Science and Technology 1996;30(1):220-224.
James RR, McDonald JG, et al. Loss processes of toxaphene from the upper Great Lakes. Environmental Science and Technology (in review, 2001).
Jannson B, Wideqvist U. Analysis of toxaphene (PCC) and chtordane in biological samples by NICI mass spectrometry. International Journal of Environmental Analytical Chemistry 1983;13(4):309-321.
Jarnuzi G, Matsuda M, et al. Environmental contamination by toxaphene and toxaphene-like confounds generated during chlorobleaching process II: congener patterns of toxaphene-like compounds in striped mullet. Journal of Environmental Chemistry 1992;2:557-568.
Jeremiason JD, KC Hombuckle, et al. PCBs in Lake Superior, 1978-1992: decreases in water concentrations reflect toss by volatilization. Environmental Science and Technology 1994;28(5):903-914.
Johnson TC, Evans JE, et al. Total organic carbon in Lake Superior sediments: comparison with hermpelagic and pelagic marine environments. Limnology and Oceanography 1982;27(3):481-491.
Kada J, Heit M. The inventories of anthropogenic lead, zinc, arsenic, cadmium, and the radionuclides cesium-137 and excess lead-210 in lake sediments of the Adirondack region. Hydrobiologia 1992;246(3):231-241.
Kariekhoff SW, Brown DS, et al. Sorption of hydrophobic pollutants on natural sediments. Water Research 1979;13:241-248.
Kidd KA, Earner JE, et al. Spatial variability of chlorinated bornanes (toxaphene) in fishes from Yukon lakes. Chemosphere 1993;27(l0):1975-1986.
Kucklick JR, McConnell LL, et al. Toxaphene contamination in Lake Baikal's water and food web. Chemosphere 1993;27(10):2017-2026.
Larson RA, Martey KA. Sunlight photochlorination and dark chlorination of mionoterpenes. The Science of the Total Environment 1988;77(2-3):245-252.
Lawrence JE, Casida JE. Interaction of lindane, toxaphene, and cyclodienes with brain specific t-butylbicyclophosphorothionate receptor. Life Sciences 1984;35(2):171-178.
Liss PW, Slater PG. Flux of gases across the air-sea interface. Nature 1974;247(5438):181-184.
Manahan SE. Environmental chemistry. Boca Raton, FL: Lewis Publishers, 1994.
McCarthy JF, Bartell SM. How the trophic status of a community can alter the bioavailability and toxic effects of contaminants. In: Cairns J, Pratt JR, eds. Functional Testing of Aquatic Biota for Estimating Hazards of Chemicals. Philadelphia, PA, 1988, 242.
McConnell LL, Kucklick JR, et al. Long-range atmospheric transport of toxaphene to Lake Baikal. Chemosphere 1993;27(10):2027-2036.
Minnesota Department of Health. Health risk limits for groundwater. Minnesota Department of Health, St. Paul, MN, 1996;6.
Minnesota Department of Health. Minnesota Fish Consumption Advisory. Minnesota Department of Health, St. Paul, MN, 1999;96.
Mehendale HM. Pesticide-induced modification of hepatobiliary function: hexachlorobenzene, DDT, toxaphene. Food Cosmetics and Toxicology 1978;16(1):19-31.
Meister E. Farm chemicals handbook. Willoughby, OH: Miester Publishing Co., 1976.
Miskimmin BM, Muir DG, et al. Chlorobornanes in sediments and fish 30 years after toxaphene treatment of lakes. Environmental Science and Technology 1995;29(10):2490-2495.
Mortimer CH. Environmental status of the Lake Michigan region, Volume 2: physical limnology of Lake Michigan. United States Energy Research and Development Administration, Argonne, IL, 1975;121.
Minnesota Pollution Control Agency. Annual average air concentration limits and corresponding screening emission rates for contaminants potentially having carcinogenic effects. Minnesota Pollution Control Agency, St. Paul, MN, 1995;6.
Muir DG, de Boer J. Toxaphene: analytical chemistry. Chemosphere 1993;27(10):1827-1834.
National Oceanic Atmospheric Administration. Great Lakes Coastwatch, ftp://coastwatch.glerl.noaa.gov/glsea.
O'Brien R. Insecticides, action, and metabolism. New York, NY: Academic Press, 1967.
Occupational Safety and Health Administration. Toxic and hazardous substances: air contaminants. Washington DC, 1989; 29CFR Part 1910.
Paasivirta J, Rantio T. Chloroterpenes and other organochlorines in Baltic, Finnish, and Arctic wildlife. Chemosphere 1991;22(1-2):47-56.
Pearson RF. Concentrations, accumulations, and inventories of hydrophobic organochlorine compounds in sediments of the Great Lakes and in the benthic nepheloid layer of Lake Michigan. Department of Environmental Health, University of Minnesota, Minneapolis, MN, 1996.
Pyysalo H, Antervo K. GC profiles of chlorinated terpenes (toxaphene) in some Finnish environmental samples. Chemosphere 1985;14:1723-1728.
Rapaport R, Eisenrerch S. Atmospheric deposition of toxaphene to eastern North America derived from peat accumullation. Atmospheric Environment 1986;20(12):2367-2379.
Ribick MA, Dubay GR, et al. Toxaphene residues in fish: identification, quantification, and confirmation at part perbiffion levels. Environmental Science and Technology 1982;16(6):310-318.
Rice CP. Homotogue concentrations of chlorobornanes in selected samples from the Bering/Chukchi Sea. Chemosphere 1993;27(10):1937-1947.
Robbins JA, Edgington DN. Determination of recent sedimentation rates in Lake Michigan using Pb-210 and Cs-137. Geochemica et Cosmochimica Acta 1975;39:285-304.
Roussel JS, Clover DF. Resistence to the chlorinated hydrocarbon insecticides in the Boll Weevil. Journal of Economic Entomology 1957;50:463-468.
Safe S, Hutanger O. Polychlorinated Dibenzo-p-dioxins and -furans: sources, environmental impact, epidemiology, mechanism of action, and health risks. In: Safe S, Hutzingerand O, Hill T, eds. PCDDs and PCDFs: Sources and Environmental Impacts. Berlin: Springer-Verlag, 1990, p. 145.
Saleh MA. Toxaphene: chemistry, biology, toxicity and environmental fate. In: Ware GW. Reviews of Environmental Contamination and Toxicology. New York, NY: Springer-Verlag 1991, pp. 1-85.
Saleh MA, Casida JE. Reductive dechlorination of the toxaphene component 2,2,5-endo, 6-exo, 8,9,10 heptachloroboroane in various chemical, photochemical, and metabolic systems. Journal of Agricultural Food Chemistry 1978;26:583-590.
Sanborn JR, Metcalf RL, et al. The fate of chlordane and toxaphene in a terrestrial aquatic model ecosystem. Environmental Entomology 1976;5(3):533-538.
Scholz CA. Sediment distribution and sedimentation rates in the western arm of Lake Superior using 3.5 kHz seismic reflection profiles and 210Pb geochronology. Chemical Engineering. University of Minnesota, Minneapolis, MN, 1985.
Schubauer-Berigan M, Crane J. Executive summary to survey of sediment quality in the Duluth/Superior Harbor: 1993 sampling results. Minnesota Pollution Control Agency, St. Paul, MN, 1997.
Schwab DJ, Leshkevich GA, et al. Automated mapping of surface water temperature in the Great Lakes. Journal of Great Lakes Research 1999;25(3):468-481.
Shanks KE, McDonald JG, et al. Are pulp and paper nails sources of toxaphene to Lake Superior and northern Lake Michigan? Journal of Great Lakes Research 1999;25(2):383-394.
Soto AM, Chung KL, et al. The pesticides endosulfan, toxaphene, and dieldrin have estrogenic effects on human estrogen-sensitive cells. Environmental Health Perspectives 1994;102(4):380-383.
Spencer EY. Guide to the chemicals used in crop protection. Research Branch, Agriculture Canada, 1982.
Stuthridge TR, Wilkins AL, et al. Identification of novel chlorinated monoterpenes formed during kraft pulp beaching of Pinus radiata. Environmental Science and Technology 1990;24(6):903-908.
Swackhamer D, Charles J, et al. Quantitation of toxaphene in environmental samples using negative ion chemical ionization mass spectrometry. Analytical Chemistry 1987;59(6):913-917.
Swackhamer DL, Armstrong DE. Distribution and characterization of PCBs in Lake Michigan water. Journal of Great Lakes Research 1987;13(1):24-36.
Swackhamer DL, McConnell LL, et al. Workgroup report on environmental transport and fate. Chemosphere 1993;27(10):1835-1840.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
Other project views: | All 2 publications | 2 publications in selected types | All 2 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
James RR, McDonald JG, Symonik DM, Swackhamer DL, Hites RA. Volatilization of toxaphene from Lakes Michigan and Superior. Environmental Science & Technology 2001;35(18):3653-3660. |
R825246 (Final) |
Exit |
|
Swackhamer DL, Pearson RF, Schottle SP. Toxaphene in the Great Lakes. Chemosphere 1998;37(9-12):2545-2561. |
R825246 (Final) |
|
Supplemental Keywords:
particulate organic carbon, POC, bornanes, bornenes, pinenes, camphenes, bioaccumulation factor, BAF, homolog, epilimnion, STELLA, Lake Michigan, Lake Superior, water quality., RFA, Scientific Discipline, Air, Geographic Area, Waste, Water, Ecosystem Protection/Environmental Exposure & Risk, Limnology, Water & Watershed, air toxics, Contaminated Sediments, exploratory research environmental biology, Environmental Chemistry, Ecosystem/Assessment/Indicators, Chemical Mixtures - Environmental Exposure & Risk, Ecosystem Protection, State, Ecological Effects - Environmental Exposure & Risk, Air Deposition, Ecological Effects - Human Health, tropospheric ozone, Atmospheric Sciences, EPA Region, Great Lakes, Watersheds, Ecological Indicators, atmospheric processes, ecological exposure, fate and transport, mass balance model, exposure and effects, air-water exchange, atmospheric inputs, resource management, contaminated sediment, ecological modeling, ecological assessment, human exposure, aquatic ecosystems, Toxaphene, Lake Superior, water quality, atmospheric contaminants, lake ecosystem, Lake Michigan, Region 5, fish contamination, toxaphene loading, atmospheric deposition, toxicsProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.