Landscape Ecotoxicology and Risk Assessment
The program is an attempt to evaluate systems at a large regional scale with a variety of habitats, landforms and multiple stressors. IETC has two ongoing programs:
- The use of metapopulation and patch models to evaluate the effects of toxicants upon patches of various sizes, distributions and migration pathways.
- Ecological Risk Assessment of the impacts of multiple stressors upon fish, benthic and phytoplankton community structure with the Willamette-McKenzie Rivers of Oregon and Codorous Creek in eastern Pennsylvania. This three year project is being funded by the National Council for Air and Stream Improvement.
The use of metapopulation and patch modeling to understand the influence of landscape upon ecotoxicological effects.
Betinna John, Wayne G. Landis, Lori Lenart, L., Louis R. Macovsky, L.R., John F. McLaughlin, Julann A. Spromberg
One of the classical problems of environmental risk assessment is the extrapolation of concentration-response data to risks in patchy environments. Spromberg, Johns, and Landis have modified the metapopulation models of J. Wu and colleagues to incorporate interaction with a toxicant or other stressor. The initial research demonstrated the importance of initial conditions, distance between patches, and configuration of the landscape. Multiple outcomes were possible from the same initial conditions. This preliminary work led to the Action at a Distance Hypothesis. The hypothesis states that impacts upon one patch can affect distant patches not coming into contact with a stressor. Macovsky confirmed this finding using a constructed metapopulation of the flour beetle, Tribolium castaneum. Mortality in patch-1 of a linear metapopulation alters the age structure of the population in the patch farthest from the initial stressor. This finding confirms the Action at a Distance Hypothesis for this experimental metapopulation. More recent work by Lenart and Landis expanded the models to examine the alterations in impacts due to different patterns of concentration-response curves, patch configuration, and whether a source or sink patch is dosed. McLaughlin is expanding this work to large-scale systems using a variety of techniques that detect patterns in nonlinear dynamics. The Action at a Distance Hypothesis offers a powerful tool for understanding the potential impacts and risks at a landscape level. Next steps include further experimental verification, confirmation of the effect in the field, and methods for incorporating these dynamics into ecological risk assessment.
Patch Dynamics and Ecological Context
Wu and Loucks (1995) presented hierarchical patch dynamics as a new paradigm blending the constructs of non-equilibrium dynamics, heterogeneity, and scale into a comprehensive understanding of ecological dynamics. This emergent paradigm has important implications for understanding the local and regional impacts of stressors. In the past several years our research group has incorporated a segment of this paradigm, patch dynamics (Wu et al. 1993), and added a chemical stressor (Spromberg et al. 1998).
Chemical contamination occurs in heterogeneous environments and the distribution has been found to be patchy (Clifford et al. 1995). Using a template from Wu et al. (1991, 1993), we have developed a computer model of a generic animal metapopulation that has contaminated patches (Spromberg et al. 1998) . The model was written using Stella (Figure 1) The single species model is discrete, deterministic, and uses a stochastic function to simulate exposure of the organism to the xenobiotic. Running a series of simulations, we have come to four principal conclusions. First, low levels of toxicity in one patch can significantly impact populations of other landscape patches that have no contact with a toxicant. Also implied in this statement is that sites connected by migration to a contaminated site are not reference sites. Second, the arrangement of the patches is critical to the dynamics of the system and the overall impact of a toxicant. Third, effects of the toxicant persist long after the degradation of the material and these effects are exhibited in the population dynamics. Fourth, several discrete outcomes are possible with the same initial conditions. Outcomes may range from the population reaching carrying capacity to eventual extinction.
Figure 1. A three patch metapopulation model in Stella (Spromberg et al 1998)
One of the questions not addressed in earlier modeling efforts was the placement of the contaminated site in the context of the landscape. To examine this aspect we ran a series of simulations with a linear arrangement of patches (Fig. 2) and a degradable toxicant. The initial population sizes were 200, 50, and 50 in patches 1, 2, and 3, respectively. The contaminant at the LD100 was placed in patch 1, then patch 2, and then patch 3 with 50 runs performed in each state to evaluate the frequency of outcomes. In a second series of simulations, patch 2 contained the largest population, 200, and the two end patches started with a population size of 50 each
Figure 2. Arrangements and initial population sizes of the patch models. In Case 1 the source patch on the end is dosed. In Case 2, the middle patch is dosed and is also the source patch.
The outcomes of the simulations are presented in Table 1. When patch 1 was dosed, the four outcomes were: (1) In 50 percent of the simulations only the population in patch 2 survived and it was at the minimal viable population. (2) In 26 percent of the runs the populations in patches 1 and 2 survived at the MVP. (3) In 10 percent of the simulations all three populations survived at the MVP. (4) In only 14 percent of the runs all three populations reached carrying capacity. Placing the toxicant in nonsource patches 2 and 3 in the middle and at the far end with population densities at 200, 50, and 50 resulted in all populations reaching carrying capacity.
In another series of simulations the arrangement was a population of 50 in patch 1, 200 in patch 2, and 50 in patch 3 (Figure 2). The middle patch (the source) was dosed. Only three possible outcomes arose. (1) In 56 percent of the runs the populations in patches 1 and 3 existed at the MVP. (2) In 28 percent of the cases all three patches reached MVP. (3) In 16 percent of the cases all three populations reached carrying capacity. As with the first set of simulations applying the contaminant to nonsource populations (in this case at the ends of the landscape), all patches reached carrying capacity. In both series of simulations the alteration in possible outcomes and outcome frequency depended upon the location of the contaminated patch in the context of that specific landscape arrangement.
Table 1. Frequency of outcomes from different landscapes. In each case the patch was dosed with an LD100. Distance between patches is two units. mvp=minimal viable population, cc=carrying capacity. If not explicitly mentioned the population in that patch is extinct.
Outcome by patch number
Case 1. End dosed (source patch)
Case 2. Middle dosed (source patch)
1, 2 mvp
1, 2, 3 mvp
1, 2, 3 cc
1, 3 mvp
Source patch not dosed
The findings of these modeling efforts have led Spromberg et al. (1998) to hypothesize "Action at a Distance." The basic premise is that information about a stressor event can be transmitted to other patches by the migration of organisms and resources and the resultant patterns can be expressed at a variety of biochemical, organismal, and ecological layers. Action at a Distance does not require the direct contamination of a patch or of any organism that resides in a patch. The patterns in the resulting dynamics can be varied and nonlinear.
Partial experimental confirmation of Action at a Distance has been provided recently by the research of graduate student Louis Macovsky in our laboratory at the Institute of Environmental Toxicology and Chemistry. This study created a novel laboratory metapopulation model of a single insect species, Tribolium castaneum. Arranged linearly, habitat patches were linked by density-dependent dispersal of the adult morph. Patches were monitored for the indirect effects on population demographics beyond the patch that received a simulated adulticide over the period of approximately one and one half egg-to-adult cycles. It was demonstrated that indirect effects do occur in patches beyond the patch directly impacted. The indirect effects were dose-related and correlated with distance from the directly disturbed patch.
- Control sites or reference structures do not exist if migration occurs. Each structure has a unique etiology and place in the landscape.
- Effects at the molecular and organismal level, such as toxicity, can have impacts upon metapopulation structure and landscape. There is a connection between molecular biology and the shape of a landscape.
- Context in the landscape is critical to the prediction of the impacts of stressors in the environment. The same stressor can have very different impacts depending upon its location in the landscape. In certain situations, several different outcomes can result for the same initial conditions. These factors are:
a) Which is dosed: a source or a sink patch.
b) Distance to other patches and the initial numbers.
c) Relative arrangement of the dosed patch to other patches.
Given these findings we must reconsider our goals in ecological risk assessment. Our risk assessment/risk management goal is now to predict the trajectory relative to a desired ecological state taking into context the landscape characteristics. Site specificity is critical in making accurate predictions of potential outcomes.
Spromberg, J. A., B. M. John and W. G. Landis. 1998. Metapopulation dynamics: indirect effects and multiple discrete outcomes in ecological risk assessment. Environ. Toxicol. Chem. 17:1640-1649
For further information contact: Dr. Wayne G. Landis, Director, Institute of Environmental Toxicology and Chemistry, Huxley College of Environmental Studies Western Washington University, Bellingham, WA 98225-9180, T: 360-650-6136, email firstname.lastname@example.org.
Ecological Risk Assessment of the impacts of multiple stressors upon fish, benthic and phytoplankton community structure with the Willamette-McKenzie Rivers of Oregon and Codorous Creek in eastern Pennsylvania.
The estimation of regional risks due to multiple stressors is a frontier in environmental toxicology and risk assessment. We are conducting a regional scale ecological risk assessment due to multiple stressors in the Willamette Valley, Oregon. We have broken the McKenzie and Willamette watersheds into 12 risk regions and have mapped the locations of the point sources and are in the process of incorporating land use data. We will present our ranking criteria for the stressors and preliminary habitat evaluation. We are using an approach to ecological risk assessment developed by Wiegers, Landis, and colleagues to combine multiple stressors and receptors in a regional context. The model involves the development of risk matrices that combine diverse stressors and habitats within the region with numerical ranks. The model can be employed to estimate risk at both regional and subregional levels, determine the contribution to risk by each stressor, and predict uncertainty. The Willamette and McKenzie river watersheds cover a large area and the region includes multiple stressors. The Willamette River drains an extensive agricultural area and forests of both the Coastal and Cascades mountains. The river also receives effluents from paper mills and urban wastewater treatment facilities. A major tributary of the Willamette is the McKenzie River. The McKenzie watershed, which extends into the Cascade Mountains, is extensively forested. Stressors in this watershed include alterations in the landscape due to the harvest of trees, the infrastructure required for the logging, modification of the river or stream banks, and inputs due to contamination by localized urban and non-point sources. Concurrent with this risk assessment is a multi-year sampling program to characterize fish, macroinvertebrate and periphyton communities that will be used to test the hypotheses of the regional ecological risk assessment.
This study will use our baseline model for differentiating between anthropogenic and natural alterations to aquatic systems for two freshwater and a terrestrial site. We developed the model in an ecological assessment of Port Valdez, AK. Port Valdez has a variety of anthropogenic stressors including fish hatcheries, effluent from the pipeline, and tanker traffic (Landis and Wiegers 1997, Weigers et al 1997, Weigers et al in press 1998). Natural stressors include the aftermath of the Alaskan Earthquake, sediment deposition from spring glacier melts, and ice scouring of the mudflats. Such an assessment dealing with a multitude of stressors brings with it unique characteristics. Analyses of diverse scales, a variety of land forms and landscape types, and the integration of a diverse group of distinctly different stresses are necessary (Suter 1993). Stresses occur at distinct locations that affect interactions and outcomes. Organisms migrate depending on life stage and season, which alters their exposure to stressors. Sources of stresses also vary by season and with changes in human activity and agricultural and industrial processes.
To examine the utility of the methodology we are using two distinct study sites: Codorus Creek in eastern Pennsylvania, the Willamette and McKenzie Rivers in western Oregon, and the Olympic National Park in Washington State.
All of our proposed study sites include a variety of anthropogenic stressors, effluents from paper manufacture, municipal discharges, agricultural run-off, in addition to natural events such as flooding. Our study sites are also quite different in character. Our goal is to use a common approach that can distinguish a variety of impacts and that is also specific to a particular location. Such a method is our recently developed relative risk approach to environmental assessment.
The methods used here conform to the risk assessment three-phase approach: problem formulation, analysis, and risk characterization. Problem formulation is information gathering and includes a definition of assessment endpoints. Information is processed to give an estimate of risk during the analysis phase and then interpreted in the risk characterization phase.
We initiate our risk assessments by asking three questions:
- What are the physical and biological characteristics of the study sites?
- What are the components to be protected in these sites?
- What impacts are known to have occurred in the environment?
A variety of sources of information will be utilized. Baseline investigation of the study sites will provide information about seasonal fluctuations, circulation patterns, habitat types, and plant and animal populations. We will also examine the various discharge permits, determine if data regarding stressors are available, request data when pertinent, and examine the literature to determine the range of stressors possible from each source.
To determine the resources that are to be protected in each area we will examine the goals of each state that regulates these riverine systems. Salmonid populations will obviously be important in the Oregon site, while bass and other sportfish will be more important in Pennsylvania. In some states IBI values are considered important variables, other states regulate particular species. Any assessment system should be able to take into account the varying local requirements.
Assessment and Measurement Endpoints
Discussions with risk managers, community interviews, and public meetings will result in the selection of assessment endpoints. Fisheries, tourism, and the community's concern for the quality of their environment influence the emphasis of particular assessment endpoints.
Results of the Problem Formulation: Conceptual Model. Information gathered during the problem formulation phase provides a basis for constructing a conceptual model. Together, all of the risk components (sources, stressors, habitats, receptors, ecological impacts, and receptor responses) and links form the basis of the conceptual risk model. We will likely divide each site into subareas for analysis. Analysis of smaller areas within a region maximizes the chance of detecting effects from anthropogenic input. We will select categories for each of the regional risk components: sources, habitats, and impacts. Source and habitat categories describe anthropogenic and ecological components. Other factors can be nested into these primary ones. In this way all potential stressors, receptors, and receptor responses of the rivers can be included in the assessment.
Once these component categories are established, we will explore links between them. In a regional multiple-stressor assessment, the stressors are linked to sources and the receptors to habitats. Stressors and receptors both present or expected in an area provide the basis for developing risk scenarios. The characterizations describe factors driving exposures and the effects likely to occur with each combination of stressor and receptor. Generalized risk scenarios provide a format from which to develop hypotheses for future quantitative assessments of selected concerns. In a large area, risks vary by location. Applying the ranking approach within each subarea of each river will lead to a comparative analysis of risk.
The analysis phase of the assessment will include two parts: quantitative analyses of specific cases using the site-specific approach and comparative analysis of risks from a regional approach.
The relative risk model is a framework for comparing risks from anthropogenic stressors throughout the study sites and for considering more than one stressor and receptor at a time. We will use our relative risk model to compare sources and habitats in subareas of the study site and to determine which subareas will have a greater chance of impact. Comparisons are based on ranking criteria from information specific to each subarea. Analysis will be specific to each environment. Assessing risk in this manner allows us to establish a relative scale to weigh the likelihood of receptors in a particular habitat being exposed to, and possibly affected by, events in the study sites.
Comparative risk estimates will be based on the following assumptions:
- The greater the size or frequency of a source in a subarea, the greater the potential for exposure to stressors.
- The type and density of receptors present is related to the available habitat.
- The sensitivity of receptors to stressors varies between habitats.
- The severity of effects in subareas depends on relative exposures and the characteristics of the receptors present.
The relative risk model is a system of ranking risk components and filtering each possible combination. This system involves the following three steps:
Each source and habitat will be ranked for each subarea to indicate high, moderate, low, or no risk. Ranks will be assigned using criteria specific to each study site. Criteria will be based on the size and frequency of the source and the amount of available habitat. Ranks will be assigned to each source and habitat type on a 2-point scale from 0 to 6 where 0 indicated lowest potential for exposure and 6 the highest.
- Filter Design
We will use filters to determine the relationship between risk components (sources, habitats, and impacts to assessment endpoints). A filter consists of weighting factors 0 or 1 indicating low or high probability, respectively. We will use two filter types: exposure and effect. The exposure filter screens the source and habitat combinations likely to result in exposures. The effects filter weights those likely to affect a specific assessment endpoint.
The first step in designing the exposure filter for this project will be to determine which stressors are produced by the sources. Two sequential questions about each stressor in relation to specific source&endash;habitat combinations will then be considered:
Will the source release or cause the stressor?
Will the stressor then occur and persist in the habitat?
If the answer to both questions is yes, then a 1 is assigned to the source habitat combination. If the answer to either question is no, then a 0 is assigned.
The design of the effect filter is similar, but a separate filter is made for each assessment endpoint. The first step in this process is to determine what type of effects are important to the specific endpoint. Criteria for the effect filters are:
Will the source release stressors known to cause this particular effect to the endpoint?
Are receptors associated with the endpoint sensitive to the stressor in this habitat?
If the answer to both questions is yes, then a 1 is assigned. If the answer to either is no, then a 0 is assigned.
- Integrating Ranks and Filters
To integrate factors potentially causing risks in each site, we will convert the ordinal data in the form of ranks into a point system that is then combined mathematically. Integration occurs in two steps. The first involves integrating the source and habitat ranks by multiplication. This allows us to show a source in a high-ranking habitat as more important than the same source in a low-ranking habitat. The filters refine the resulting values to account for the likelihood of an exposure or effect occurring with each particular combination of source and habitat. The second integrative step is to add all the points received for a particular component (subarea, source, or habitat) and to compare the final tally.
Final risk scores (RS) are calculated for each subarea by multiplying ranks by the appropriate weighting factor (Wjk) as indicated below.
RS = Sij x Hik x Wjk, (1)
i = the subarea series
j = the source series (discharges ... shoreline activity)
k = the habitat series (rapids ... stream mouths)
Sij = rank chosen for the sources between subareas
Hik = rank chosen for the habitats between subareas
Wjk = weighting factor established by the exposure or effect filter.
The results form a matrix of risk scores for the relative exposure or effects associated with a source and habitat in each subarea. The potential risk resulting from a specific source [Eq. (2)] and occurring within a specific habitat [Eq. (3)] is summarized for each subarea by adding the related scores,
RSsource = Â(Sij x Hik x Wjk) for j = 1 to 8, (2)
RShabitat = Â(Sij x Hik x Wjk) for k = 1 to 8. (3)
This matrix provides the relative contribution of each stressor, as measured by an increase in risk to the aquatic environment. Such an ability is the primary goal of this research program.
One of the advantages of the relative risk procedure is that it produces testable hypotheses concerning which areas are impacted and the outputs of various sources. In the course of our developmental program we will test our predictions against the field data being collected. Especially useful will be the confirmation of impacts and the contributions of sources using data not included in the original assessment.
In this study we will address uncertainty (1) in the problem formulation, (2) in the calculation of relative risk, and (3) in the accuracy of relative risk estimates. Uncertainty associated with the problem formulation is mostly qualitative. The calculation of relative risk has a quantifiable level of uncertainty. We have designed a sensitivity analysis approach to ascertain the possible variance within this mathematical model and will use this approach for this project. The third type of uncertainty will be explored through the confirmatory analysis used to quantify or describe specific risks in each study site.
The sensitivity of the relative risk model depends on its ability to distinguish between high and low risk areas. The model operates on input that ranks habitats and sources then filters out the probable exposures or effects. To analyze the sensitivity of our relative risk model, we will incorporate randomly chosen input and examine the results for each subarea. The sensitivity analysis will be based on the premise that when input is randomly chosen, the model results will not discriminate between different subareas. Input that is risk-related, instead of random, will drive the model to detect the high-risk areas.
Testing the Assessment Methodology
We are testing the utility of the relative risk assessment methodology on Codorus Creek in eastern Pennsylvania, the Willamette and McKenzie Rivers in western Oregon, and the Olympic National Park in Washington State.
Willamette River, Oregon. The reach of the Willamette River that we are investigating lies in the Upper Willamette River watershed with the stretch of interest downstream from Eugene through Corvallis. This stretch is composed of characteristic large river habitats, including main channel, side channels, and backwaters formed by meander cutoffs and oxbow lakes; depths in the main channel are up to 10 m with strong currents. The riparian area and floodplain are intensively cultivated for rowcrop and turfgrass production. Downstream from the Eugene-Springfield metropolitan area is the confluence with the McKenzie River, a medium-sized river with a steeper gradient and rapid currents. Upstream anthropogenic impacts include an unbleached kraft paper mill 13 km upstream on the McKenzie River and the Eugene/Springfield wastewater treatment plant on the Willamette River upstream of the confluence with the McKenzie River at River Kilometer (RKm) 178.5.
The town of Harrisburg discharges treated municipal wastewater at RKm 161. At RKm 149 the Long Tom River, which exhibits considerable impact from agriculture, enters the Norwood Island side channel. Two miles downstream from its confluence with the Willamette River, the effluent from a bleached kraft paper mill is discharged.
The Willamette River enters Corvallis at RKm 135, and the Mary's River, another agriculturally impacted stream, joins the Willamette River at RKm 132. The Corvallis wastewater treatment plant discharges down river at the lower bounds of the city at RKm 130.8. The Willamette River flows into the Columbia River at RKm 0.
McKenzie River, Oregon. We are evaluating the McKenzie River from its confluence with the Willamette River at Eugene/Springfield upstream to its confluence with the Walterville Canal. This stretch is characteristically broad and shallow with a swifter current due to a higher gradient than the Willamette River. The substrate is largely composed of sand to cobble and boulder-sized particles. Because of the shifting channel, habitat consists of main channel, side channels and backwaters formed by meander cutoffs. The watershed is subject to a variety of uses including intensive rowcrop agriculture, pasture, forest, and commercial and residential development.
Codorus Creek, Pennsylvania. The Codorus Creek watershed, which forms a portion of the Lower Susquehanna River drainage, is located in York County in southeast Pennsylvania. Land use in this watershed is dominated by agriculture with additional demands on water for public use and industrial needs. The main stem of Codorus Creek begins near the Pennsylvania-Maryland border and flows northwest through a hilly, agricultural landscape toward Lake Marberg, a state-owned 530-ha multipurpose impoundment, which provides fishing and boating recreation. Lake Marberg was created by impounding West Branch Codorus Creek, and the discharge from the lake is released from below the thermocline in the reservoir. This causes West Branch Codorus Creek to be cold through the summer, atypical of other streams in this region. The brown trout species, a coldwater stenothermic fish native to Europe, has established a self-supporting population after initial stocking in the portion of the creek just downstream from the dam. Rainbow trout are restocked annually as a put-and-take fishery. The viability of the trout populations is dependent on the release of the deeper, cold water from the reservoir. The West Branch Codorus Creek joins with the main stem 1 km downstream from the dam, resulting in summer cooling of Codorus Creek.
Oil Creek, which drains a more intensely agricultural watershed, flows into Codorus Creek at the town of Menges Mill approximately doubling the discharge. In addition, inorganic nutrient concentrations increase notably beyond the lesser agricultural impacts in the upper Codorus Creek drainage.
Downstream from the confluence of Oil Creek and the main stem of Codorus Creek is a lowhead dam that creates a small impoundment in the Creek at the town of Spring Grove. The Spring Grove paper mill is located there in conjunction with an electric generating station adjacent to the impoundment. Water is drawn from this pond into the mill for wood processing and into the electric generating station for noncontact cooling water. The electric generating station discharges the warmed cooling water back into the impoundment at the dam, which rewarms Codorus Creek to temperatures similar to other streams in the Lower Susquehanna River drainage. The wastewater from the mill passes through a treatment pond system before being discharged into Codorus Creek downstream from the dam, the result being that a short segment of stream occurs between the thermal discharge and the paper mill effluent. In addition, the municipal wastewater from Spring Grove is also received by the mill where it is sent through the treatment ponds. The treatment pond discharge is stained dark brown by wood by-products, particularly polyphenols, formed during the processing of wood materials. This imparts a dark brown color to the entire stream. A dioxin advisory issued in 1990 for the reach downstream of the mill was lifted in 1994. A U.S. Geological Survey gauging station (Station Number 01574500) located at Spring Grove indicates that maximum spring discharges generally range between 14 and 28 m3/s and that summer discharges fall between 1.4 and 2.8 m3/s, which indicates that the Creek is usually wadeable for sampling purposes. The Codorus Creek watershed encompasses 194 km2 at Spring Grove.
Codorus Creek continues through an agriculturally dominated landscape to the city of York. In York, the Creek begins a stretch through which its original shape and course have been highly modified. The Creek is subject to annual spring flooding, which is occasionally intense, with the most significant occurrence 25 yr ago when Hurricane Agnes passed up the Atlantic Coast. The entire region was subject to heavy flooding and, as a result, flood prevention and mitigation measures have been implemented through the city of York and surrounding municipalities. The Creek has been straightened and deepened through channelization, the banks are revetted with crushed rock, levees have been constructed to prevent overflow of the banks, and obstructions to flow have been removed. In biotic terms, the habitat is considerably less complex and highly modified in this segment.
Beyond the modified segment downstream from York, the York Municipal Wastewater Treatment Plant is the principal discharge into the Creek. The Plant is permitted to discharge 13 million gallons per day and contributes an average of 20% of the total discharge in this reach. However, effects to the Creek are probably limited due to the advanced nature of treatment used at the Plant. Ammonia reduction is 96%, and phosphorus removal exceeds 85% because a biological nutrient reduction process was implemented in the late 1980s. Chemical analyses to the upstream reaches of the outfall versus downstream indicate minor changes in water chemistry including decreased pH (7.8 vs. 7.4) and increased hardness (153 vs. 167 mg/L) and nitrate (1.49 vs. 3.06 mg/L) (Page 1997). Prior to discharge of the treated wastewater, ultraviolet radiation sterilizers are used to disinfect the effluent rather than using chlorine as a germicidal agent. Thus the effects of the effluent are probably more related to the effluent's addition to the total discharge of the Creek and not to loading of organic material and inorganic nutrients usually associated with municipal wastewater treatment systems.
The research on these areas has been initiated with a sampling program and the initiation of the relative risk ranking process.
Over the summer of 1998 fish, macroinvertebrate, and periphyton samples are being taken and counted for Codorus Creek, Willamette, and McKenzie. This sampling program in being conducted in collaboration with the National Council for Air and Stream Improvement (NCASI) and should provide information to assist in the development and then confirm our risk predictions.
We have obtained data on the location of the sources of stressors on the Willamette and McKenzie Rivers in Oregon. This data has been combined in a GIS format with the information on the various tributaries and other riverine areas in the study area. We have subsequently divided the Willamette and McKenzie systems into a variety of subareas.
Figure 1 shows Subarea E of the Willamette. Incorporated into this map are the various channels within a basin, dams, and areas of potential chemical contamination. Sites of chemical contamination includes industrial facility discharge sites, permitted sites, and sites that report to the U. S. Environmental Protection Agency a Toxic Release Inventory (TRI). Data exists for each of these sites estimating the amount of discharge or release of materials to the environment. As is readily shown in Subarea E, and throughout the study area, the sources of contamination are clustered in space within the subarea. In some areas a mixture of releases may congregate, in others the releases are of the same type.
Figure 1. Subarea E of the Willamette. Marked on the tributaries are the point sources of stressors for the system.
Also being incorporated into the study are the various land use types within the study areas. The Willamette-McKenzie is dominated by agricultural use flanked by forestry. These study area is flanked by two mountain systems, the Cascades to the east and the Coastal Range to the west. Along the Willamette-McKenzie are the urban areas of Springfield, Eugene and Corvallis.
This research has been supported by a grant from the National Council for Air and Stream Improvement. Mr. Matt Luxon produced the GIS maps and Jo-Ann Cavenagh has assisted in the determination of landforms and sediment modeling. Ms. Cavenagh is supported by the SETAC Taylor-Francis Fellowship for 1989.
Landis, W.G. and J.K. Wiegers. 1997. Design considerations and a suggested approach for Regional and comparative ecological Risk Assessment
Lein, C. 1991. Olympic Battleground: The Power Politics of Timber Preservation, Sierra Club Books, San Francisco, CA.
Morrison, P. H. 1990. Ancient Forests on the Olympic Peninsula: Analysis from a Historical Perspective. The Wilderness Society, Washington, D.C.
Page, M. 1997, personal communication , assistant chief operator, York Wastewater Municipal Treatment Plant.
Suter, G. W. 1993. A critique of ecosystem health concepts and indexes. Environ. Toxicol. Chem. 12:1533-1539.
Waring and J. F. Franklin. 1979. Evergreen coniferous forests of the Pacific Northwest. Science204:1380-1386.
Wiegers, J.K., H.M. Feder, W.G. Landis, L.S. Mortensen, D.G. Shaw and V.J. Wilson, 1997. A Regional Multiple-Stressor Ecological Risk Assessment for Port Valdez, Alaska. Institute of Environmental Toxicology and Chemistry, Western Washington University, Bellingham