Grantee Research Project Results
Final Report: Biogeochemical Indicators of Watershed Integrity and Wetland Eutrophication
EPA Grant Number: R827641Title: Biogeochemical Indicators of Watershed Integrity and Wetland Eutrophication
Investigators: Reddy, Konda R. , Lowe, E. F. , DeBusk, William F. , Fisher, M. M. , Graham, William M. , Prenger, Joseph P. , Keenan, L. W. , Ogram, A.
Institution: University of Florida , St. Johns River Water Management District
EPA Project Officer: Packard, Benjamin H
Project Period: October 1, 1999 through September 30, 2002
Project Amount: $639,410
RFA: Ecological Indicators (1999) RFA Text | Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Aquatic Ecosystems
Objective:
The objective of this research project was to develop sensitive, reliable, rapid, and inexpensive indicators of ecological integrity for use in large-scale ecosystem management and restoration. The premise was that wetlands will more efficiently indicate the ecological integrity of the entire watershed for the following reasons: (1) wetlands are critical areas of the landscape; (2) wetlands, as low-lying areas in the landscape, receive inputs from all adjacent uplands; and (3) the response of a wetland to inputs from the uplands is patterned. Biogeochemical processes are good candidates for efficient indicators of ecological integrity because they are potentially very sensitive. They also are likely to be highly reliable in that ecological changes at this fundamental level will affect all species utilizing the ecosystem. The central hypothesis of the proposed research is that rates of biogeochemical cycling of carbon, nitrogen, and phosphorus (C, N, and P) in wetlands can be used to indicate the ecological integrity of wetlands, and that the concentrations of certain forms of these elements can accurately predict the rates of ecologically important processes. To develop efficient indicators of eutrophication, we addressed several objectives: (1) identify the key biogeochemical processes impacted by nutrient loading, and measure the rates of these processes along the nutrient gradient; (2) develop relationships between a process and its related easily measurable indicator; (3) determine the spatial and temporal distribution of easily measurable indicators for a test wetland ecosystem; (4) determine the spatial variations in biogeochemical processes, and develop spatial maps for various processes to determine the extent of impact and risk assessment; and (5) validate the predictability of empirical relationships by making independent measurements of biogeochemical processes in different wetland ecosystems.
The study area is Blue Cypress Marsh Conservation Area (BCMCA), an approximately 8,000-hectar freshwater marsh in the headwaters of the St. John’s River, in east-central Florida. The project area is predominately a mosaic of marsh communities. Approximately 75 percent of the headwater floodplain has been drained for cattle, citrus, and row-crop usage. Virtually all nutrient runoff into the marsh has been eliminated; however, vegetation changes from native sawgrass stands and maidencane flats to cattail and willow communities are still evident in areas of impact, and sediment accretion rates have been altered. This area has three advantages for use as a study area: (1) it exhibits gradients from severely impacted at some margins to apparently pristine in the interior; (2) it has multiple-input points for pollutants with information on the flow, water quality, and upstream land uses for each source; and (3) large amounts of ancillary ecological data exist for the area.
Summary/Accomplishments (Outputs/Outcomes):
Determination of Spatial Variability and Interrelationship of Biogeochemical Processes and Efficient Indicators
Three zones of the BCMCA were used as study sites: the Northeast (NE) and Southwest (SW) areas as impacted sites (nutrient enriched) and the Northwest (NW) area as an unimpacted site. Two types of samples were obtained at each station: soil and detritus. Vegetation types in the sampling areas were classified into the followingsix groups: Typha, Typha/woody mix, Cladium, Cladium/woody mix, Panicum, and others. Woody mix includes the following vegetation types: Salix sp., Myrica sp., Cephalanthus sp., and Royal fern. Biogeochemical processes related to C, N, and P cycling were measured in plant litter and soil samples, and all samples were characterized for basic physicochemical properties. Microbial indicators were analyzed on a subset of samples and correlated with the basic physicochemical properties.
Physicochemical and microbial parameters varied significantly by region and by vegetation type. The impacted areas (NE and SW) had higher total phosphorus (TP) than the unimpacted area (NW) for both detritus and soil layers. TP levels also varied significantly among some vegetation types; however, TP was significantly higher in impacted areas than unimpacted areas regardless of vegetation type. The total carbon (TC) of detritus at SW was significantly higher than NE, although there was no significant difference between NW and SW and NE. For soil TC, there was no significant difference among all three sites. Total nitrogen (TN) of detritus at NE was significantly higher than NW and SW. The increase in TP concentration resulted in a lower C:P ratio in impacted areas. In addition, the C:P ratio in Cladium, Cladium/woody mix, and Panicum areas were significantly higher than Typha and Typha/woody mix for both detritus and soil layers. These results indicate higher P content in impacted areas, especially in Typha sp. areas.
Microbial biomass nutrients varied significantly among impacted and nonimpacted areas. Microbial biomass carbon (MBC) in soil at NW was significantly higher than both NE and SW soil. In the detritus layer, however, there was no significant difference in the MBC content among the three areas. The MBC in soil in Panicum area was significantly higher than most vegetation types: Cladium/woody mix, Typha, and Typha/woody mix. For both the soil and detrital layers, microbial biomass nitrogen (MBN) was not found to be significantly different between the three regions, even when considering vegetative communities. Microbial biomass P in the NW soil was significantly higher than the concentrations found in the NE and SW. There was no significant difference in the size of the microbial biomass P associated with the detrital layer between the three regions. The labile organic pools also differed significantly among sites. In the detrital layer, polymorphonuclear (PMN) and extractable NH4+ concentrations were found to be significantly higher in the NE region than the NW and the SW regions. Within the soil, PMN rates were significantly different between all three regions, with the NW region having the highest activity. However, extractable NH4+ concentrations within the soil were significantly higher for the NE and SW regions than the NW region. Higher PMN activity was associated with lower C/N values occurring primarily in the NE region.
Microbial activities varied by area. The activity of C acquisition enzyme -glucosidase associated with Typha detritus was elevated relative to that in detritus of other vegetation types, indicating more labile, easily degraded material. The sediment oxygen demand (SOD) of detritus at NW was significantly lower than that at NE and SW, and there was no difference in SOD of detritus between NE and SW. Labile organic carbon (LOC) showed an identical pattern. When the SOD values were compared by vegetation types, the SOD of detritus at Typha and Typha/woody mix areas was significantly higher than Cladium and Cladium/woody mix areas. There was no significant difference in SOD between Panicum and other vegetation types. Anaerobic respiration, as measured by CH4 production rate, was significantly higher in detritus at the SW site than the NW site. The CH4 production rate from NE detritus was not significantly different than either that of NW or SW. On the other hand, the NW soil produced a significantly lower amount of CH4 than NE and SW soil, but there was no significant difference between NE and SW soil CH4 production. When the CH4 production data was compared by vegetation types, the CH4 production rates in detritus from Typha and Typha/woody mix areas were significantly higher than the detritus from Cladium and Cladium/woody mix. An increase in anaerobic CO2 production rate was observed in both detritus and soil layers of the SW-impacted site as compared to the reference site (NW). When the anaerobic CO2 production rate was compared by vegetation types, the detritus at Typha area had a significantly higher CO2 production rate than Cladium area. As a result, the Typha dominant area produced higher CO2 and CH4 than the Cladium-dominant area. Ratios such as MBC/TC were compared by vegetation types, and were significantly higher for detritus in Typha; although in soil, the Panicum area had significantly higher MBC/TC ratio than Typha, Typha/woody mix, and Cladium/woody mix. The MBC/Total LOC ratio for soil was significantly higher in Typha and Typha/woody mix areas than Cladium, Cladium/woody mix, and Panicum areas. Results indicated that microbial parameters and their ratios were affected by both nutrient level and vegetation types.
In general, acid phosphatase activity (APA) showed no definitive trends with regard to nutrient impact when compared among the three areas. Comparison of microbial biomass phosphorus (MBP) to TP suggested the microbial pool in the NW region accounted for a larger portion of the total P at the 0-10 cm depth than in the NE and SW regions. The same regional trends were seen in the detritus, with MBP accounting for a greater proportion of TP. These findings indicate that the microbial community in the nonimpacted region was more P efficient than the ones in the P-impacted regions.
Temporal Variability of Biogeochemical Processes and Indicators
Based on results from the spatial sampling, two sites were established in each of the reference and impacted areas for bimonthly sampling over 12 months to examine temporal variability. One site in the NE-impacted region had slightly elevated TP values (based on the spatial sampling), but retained the native Cladium vegetation community. In January 2001, a large portion of the marsh burned, including two reference and the two impacted sites in the southwest. Monitoring began in March 2001, after vegetation regrowth had begun. The hydrologic condition drastically changed during the year of temporal study. Low rainfall caused a complete drawdown of the marsh to occur from March to August in 2001. Anaerobic conditions were not established until July 2001 in the NE and SW regions and September 2001 in the NW region.
Physiochemical parameters including pH, bulk density, loss on ignition (LOI), TN, and TC in the 0-10 cm soil layer across all three regions showed no site-specific or regional differences. The temporal data showed that the TP content in the NE and SW regions was greater than the reference concentrations taken in the NW region, irrespective of plant species. Distinct variation in the detrital TP concentrations between the six sites was observed, but did not coincide with the designations of impacted and nonimpacted regions, nor were they related to the dominant vegetation species. Comparison of the C:N:P ratios indicates P limitation in the system at three of the sites, the two NW sites and the Cladium sp. NE site. There was significant difference in the N:P ratios within the NE region between the two sites (Typha sp. and Cladium sp.). The total organic P (TPo) concentration accounted for approximately 85 percent of the TP at the 0-10 cm soil layer. Analysis of the TPo and total inorganic P (TPi) concentrations showed a separation of the sites into two distinct groups with the NE Typha and SW sites having greater concentrations of both TPo and TPi at both the detritus and soil levels when compared with NE Cladium sp. and NW sites. Microbial biomass nutrient pools varied among the three sites depending on sampling date. For soil, MBC was higher in the early phase of temporal study (in March and June) for all three sites. For both detritus and soil, LOC was higher in the early phase of the study for all three sites. Aerobic and anaerobic respiration for detritus showed a general increase toward the end of temporal study. On the other hand, the anaerobic CO2 production rate was higher in the early phase of the study for soil in all three sites.
Microbial biomass N was significantly higher in the SW and NW regions than the NE region for the soil and detritus layers. Significantly lower MBN was present for detritus associated with Cladium communities from both the NW and NE regions. For extractable NH4+, a seasonal effect was observed as well as a significant difference in seasonal patterns between the three regions within both the soil and detrital layers. For the detritus layer, a significant difference in PMN rates was found over time with the NE and SW regions having higher mineralization rates than the NW region. Extractable NH4+ concentrations were found to be significantly higher in the NE and SW regions within the detritus layer, corresponding to PMN rates. PMN rates and extractable NH4+ concentrations were not significantly different within the soil layer between the impacted and unimpacted regions and therefore, apparently not influenced by total P concentrations. PMN rates within the detritus layer were significantly lower for NW (Cladium) vegetation than other vegetative communities. This corresponds to the low PMN rates in the NW region, reflecting a higher C/N ratio or lower decomposition rate of the Cladium substrate. No significant difference between vegetative communities was found for PMN within the soil. For the detritus layer, total P influenced PMN rates and extractable NH4+ concentrations.
The relative proportion of MBP in the soil was significantly greater in the NW region when compared to the NE and SW regions. The elevated concentrations of MBP in the nonimpacted area could be related to the dominant vegetation, Panicum sp., which has thick, fine-root masses. The distribution of detrital MBP demonstrated no significant differences between any of the sites, but the MBP proportions suggest a trend among similar vegetation communities. The distribution of the forms of P followed regional TP trends at the 0-10 cm soil layer; however, the detrital TP distribution was more clearly linked to vegetative communities than to the soil TP concentrations. The organic P fraction of the soil showed fluctuations over time in the MBP, LOP, FAP, HAP, and residual Po. Microbial biomass changed by seasonality with maxima occurring in the summer (June and July) and winter (December and January) months.
Results indicate that most of the microbial parameters were significantly affected by seasonality. Although many of the microbial parameters were significantly affected by time and station individually, many of them did not show significant difference when time and station effects were combined. The 1-year temporal study showed that differences in vegetation type, soil conditions, and hydrology had an impact on the TP of the soil. The majority (>85 percent) of the soil and detritus TP was in the organic form. TP, which is more readily available for microbial uptake and utilization by vegetation, did not show significant seasonal or site variations. The fraction of organic P most significantly affected by seasonal and site effects was LOP, with reduced concentrations during the summer months of June and July correlating with the drought period. The detrital layer was influenced by the dominant type of vegetation present, more than differences in soil TP. Overall, the TP concentrations appear to be diminishing over time, indicating the system is redistributing P and recovering. Variations in soil characteristics as a result of fluctuations in water levels appeared to have a greater effect on the P pools than the two surface fires that occurred in BCMCA prior to sampling. Changes in hydrology appear to have a sustained impact over several sampling periods, dependent on duration of unsaturated conditions. Fire can be a sudden and intense impact to the system; however, it appeared to have a minimum impact on the soil P over the year focused on by the first sampling period after the event.
Diversity and Composition of Prokaryotic Groups Related to C, N, and P Cycling in Wetlands
The objectives of this task were to: (1) investigate the diversity and composition of prokaryotic groups related to C cycling in BCMCA; (2) understand the diversity of indicator groups with respect to nutrient loading in the BCMCA impacted versus nonimpacted sites; and (3) examine the structure-function relationships of syntrophic-methanogenic groups with respect to the nutrient loading in impacted versus nonimpacted sites. Using polymerase chain reaction (PCR)-based cloning and sequencing or analytical techniques such as terminal restriction fragment length polymorphisms (T-RFLPs), shifts in composition of assemblages within the methanogens were investigated. In addition, methane concentrations were monitored weekly in microcosms using soil from impacted and nonimpacted areas of BCMCA. Addition of sulfate inhibited methanogenesis in the NE Typha soils when butyrate was the substrate, whereas for propionate, methanogenesis increased twofold. This reflects the need of syntrophic bacteria for sulfate to oxidize propionate. These results indicate a competition between the sulfate-reducing bacteria (SRBs) and the syntrophs for C donors (propionate and or butyrate). However, in the nonimpacted NW Cladium site, exogenous addition of sulfate completely inhibited methanogenesis, indicating the greater involvement of SRBs there.
Ribosomal DNA clones were grouped into operational taxonomic units based on their restriction fragment length polymorphism (RFLP) patterns. Representative clones having different and similar RFLP patterns then were sequenced, their sequences were compared and aligned, and a phylogenetic tree was constructed. Remarkable differences were obtained in the archaeal RFLP patterns with two different restriction enzymes in the impacted (NE) and the nonimpacted (NW) sites. Commonalities also were seen with soils from impacted sites spiked with propionate and butyrate, but clear differences were obtained in the nonimpacted sites enriched with the same substrate (propionate and butyrate). A high diversity was observed in the bacterial community, and there were few similarities in the impacted and the nonimpacted sites.
Spatial Distribution of Biogeochemical Indicators in Water, Litter, and Soil
Based on the results obtained from the preliminary geostatistical analyses in Task 1, a large-scale network was developed to characterize the spatial distribution of biogeochemical indicators in water, litter, and soil throughout the study site. Duplicate soil cores and detritus samples were obtained from 267 sample locations, and biogeochemical, physicochemical, and microbial parameters were characterized as described.
We used conditional sequential Gaussian simulation (CSGS), a stochastic simulation method, to generate realizations of soil properties to describe the spatial patterns of properties and the range of possible outcomes of realizations. An advantage of the CSGS approach is that all conditional distributions are normal and determined exactly by the mean and estimation variance. One hundred realizations at a pixel resolution of 100 meters were generated using CSGS. To assess the accuracy of TP realizations, we randomly split the dataset into model (67 percent of observations) and an independent validation dataset (33 percent of observations), resulting in 183 model observations and 84 validation observations. The root mean square error was used to evaluate realizations.
Principle Components Analysis (PCA) was used to transform a number of possibly correlated biogeochemical variables into a smaller number of principal components. The semivariogram of TP was modeled with two basic structures. Specific patterns such as lower TP values in the northern part and higher values in the southern part of the study area prevail on all maps. However, local spatial patterns are slightly different on all realization TP maps. These local deviations show the range of possible outcomes or the uncertainty of predictions based on 267 observations of TP. A crescent-shaped area in the northern part showed TP as low as 340 mg kg-1, which resembled natural TP conditions. The smallest and largest realizations can be interpreted as "best" or "worse" case scenarios of TP predictions, rendering the uncertainty of TP predictions. High-TP predictions are more uncertain compared to low-TP predictions. Realizations of TLON and TC also were generated. A simplified representation of spatial patterns based on PCs facilitates to concentrate the information on the spatial structure rather than mapping all biogeochemical properties. The variation of soil property realizations (TP, TC, and TLON) within classes was considerable. We can conclude that relationships between soil biogeochemical properties change across the BCMCA caused by ecosystem processes (e.g., mineralization, decomposition of vegetation, enzymatic activities, and other). Mapping of the mean, minimum and maximum soil property values within classes facilitated a better understanding of the variation of spatial patterns.
Validation of Predictive Equations Using Independent Measurements
The objective of this portion of the research project was to evaluate the ability of multivariate statistical procedures such as cluster analysis (CA), PCA, canonical discriminant analysis (CDA), and discriminant function analysis (DFA) to quantify the relationships among biogeochemical indicators, biogeochemical processes, vegetation types, and ecological impacts. Results from CA indicate that biogeochemical characteristics appear to be more consistent within the SW and NW regions, although a variety of soil biogeochemical characteristics exist in the NE region. PCA of this data set suggests that four or five PCs are sufficient (rather than the 28 original variables) to describe the variability in the BCMCA data. CDA provides good discrimination between the unimpacted region, NW, and the impacted regions, NE and SW. The impacted regions exhibit high values of moisture content, LOI, and TPo,but the unimpacted area shows low concentrations of these values. At least three canonical variables determined based on vegetation are necessary to discriminate between the six vegetation groups sufficiently. The results indicate that the vegetation distribution shifts from native vegetation types (Cladium and Panicum) in the unimpacted region, NW, to invasive vegetation types (Hyacinth, Mixed, and Typha) in the impacted regions, NE and SW. When analyzed by vegetation community, the impacted regions exhibit high values of moisture content, TPo, and bicarbonate extractable total P and low values of LOI, TP, and microbial P. DFA can be expected to accurately classify new observations into vegetation types at least 75 percent of the time, into regions at least 85 percent of the time, and into clusters at least 90 percent of the time.
Twelve wetlands were selected in Florida and Georgia to test some of the predictive relationships developed. At each site, a minimum of three stations were sampled as previously described, and soil biogeochemical parameters were determined. Wetlands differed significantly by ecoregion based on basic chemical characterization. Differences in soil chemistry appeared to drive microbial nutrient cycling, exemplified by APA levels. To isolate and identify factors influencing predictive relationships between biogeochemical factors, modeling was limited to wetland communities of similar vegetation communities. Relationships between the abiotic environment (independent variables) and biological responses (dependent variables) were analyzed to validate the results across ecosystems. A clustering method using an independent dataset was developed to place the soil chemical measures in groups that reflect the sampling location. We applied a combination of stepwise discrimination and CDA (stepwise canonical discrimination) to determine which particular combinations of chemical characteristics are influential in generating (abiotic) and predicting (biotic) the multidimensional groups. CA on soil abiotic parameters resulted in two consistent clusters representing nutrient-impacted and nonimpacted sites and some overlap between intermediate and nonimpacted sites. In terms of abiotic characteristics, the impacted area is differentiated by high TC, P forms, and low TN from intermediate sites and the nonimpacted area; the intermediate and the nonimpacted areas are distinguished by shifts in the P dynamics and a substantial depletion in NH4-N at intermediate relative to unimpacted. The resulting set of biotic linear combinations of measures was therefore selected to be applied as indicators of nutrient enrichment to BCMCA. Results indicated a distinct separation of the unimpacted areas from the impacted areas. The interpretation of the results reflects the diversity of vegetation communities in BCMCA in that the primary differentiation is P availability; a secondary factor is the predominant vegetation types. The effect of Typha sp. on the soil biogeochemistry was highlighted by the increase in overall N and P turnover rates. Using this analytical approach, microbial biomass does not play a significant role in describing the variability in BCMCA.
General Conclusions
• The impacted areas (NE and SW) had significantly higher TP than the unimpacted area (NW) for both detritus and soil layers.
• Microbial activities varied with both nutrient level and vegetation community.
• Changes in hydrology and duration of unsaturated conditions appear to have significant confounding influence on temporal patterns of microbial and nutrient dynamics.
• Nutrient loading in the impacted regions enriched particular bacterial groups as evidenced by the common RFLP patterns.
• The multivariate statistical analyses presented indicate that soil biogeochemical measurements are indicative of vegetation types and can be used to evaluate ecological integrity in the BCMCA.
• The effect of Typha sp. on the soil biogeochemistry was highlighted by the increase in overall N and P turnover rates.
• The primary differentiation among impacted and nonimpacted areas is P availability.
• Results reflect the diversity of vegetation communities in BCMCA, making the predominant vegetation type an important secondary factor distinguishing impacted and nonimpacted.
• The application of the biotic variables selected for in an independent dataset to BCMCA data resulted in a distinct separation of the unimpacted areas from the impacted areas.
• Microbial biomass does not play a significant role in describing the variability in BCMCA. It is possible to extrapolate predictive variables among independent wetlands within the region.
• Our study provided a shapshot in time, giving insight into the current ecological status of BCMCA. Elevated TP caused by previous adjacent agricultural activities is still present in the topsoil. The transport of P out of the wetland is dependent on biotic variables including vegetation communities, and may take several more decades.
Journal Articles on this Report : 7 Displayed | Download in RIS Format
Other project views: | All 26 publications | 7 publications in selected types | All 7 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Bostic EM, White JR. Soil phosphorous and vegetation influence on wetland phosphorus release after simulated drought. Soil Science Society of America Journal 2007;71:238-244. |
R827641 (Final) |
Exit Exit Exit |
|
Bostic E, White J. Phosphorus and vegetation effects on phosphorus retention in wetland soil after a drawdown/reflood event. Soil Science Society of America Journal. |
R827641 (Final) |
not available |
|
Corstanje R, Reddy KR. Response of biogeochemical indicators to a drawdown and subsequent reflood. Journal of Environmental Quality 2004;33(6):2357-2366. |
R827641 (Final) |
Exit Exit Exit |
|
Corstanje R, Reddy KR, Portier KM. Typha latifolia and Cladium jamaicense litter decay in response to exogenous nutrient enrichment. Aquatic Botany 2006;84(1):70-78. |
R827641 (Final) |
Exit Exit Exit |
|
Corstanje R, Reddy KR. Microbial indicators of nutrient enrichment: a mesocosm study. Soil Science Society of America Journal 2006;70(5):1652-1661. |
R827641 (Final) |
Exit Exit Exit |
|
Grunwald S, Corstanje R, Weinrich BE, Reddy KR. Spatial patterns of labile forms of phosphorus in a subtropical wetland. Journal of Environmental Quality 2006;35(1):378-389. |
R827641 (Final) |
Exit Exit |
|
Prenger JP, Reddy KR. Microbial enzyme activities in a freshwater marsh after cessation of nutrient loading. Soil Science Society of America Journal 2004;68(5):1796-1804. |
R827641 (Final) |
Exit Exit Exit |
Supplemental Keywords:
geostatistical analyses, nutrient cycling, microbial, diversity, landscape scale, water quality, water quality nutrient loads, soil quality, geographic information system, GIS, statistical models, aquatic ecosystem, biogeochemical indicators, ecosystem indicators, eutrophication, nitrogen, nutrient fluxes, nutrient gradiants, nutrient supply, nutrient transport, risk assessment, soil, spatial and temporal patterns, statistical evaluation, vegetation gradients, watersheds, wetland vegetation., RFA, Scientific Discipline, Water, Geographic Area, Ecosystem Protection/Environmental Exposure & Risk, Nutrients, Environmental Chemistry, Ecosystem/Assessment/Indicators, Ecosystem Protection, State, Ecological Risk Assessment, Ecology and Ecosystems, Ecological Indicators, aquatic ecosystem, eutrophication, nutrient supply, nutrient transport, risk assessment, wetlands, wetland vegetation, biogeochemical indicators, watersheds, nutrient gradiants, ecosystem indicators, soil, vegetation gradients, nutrient cycling, water quality, aquatic ecosystems, GIS, nutrient fluxes, spatial and temporal patterns, Florida, nitrogen, statistical evaluationRelevant Websites:
http://wetlands.ifas.ufl.edu/ Exit
Progress 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.