691 (KAP-1117-561805) Ecology and Ecological Risk Assessment: why spatial patterns and scales matter.
Start time: 1:50 PM
Kapustka, L¹, ¹ ecological planning and toxicology, inc., Corvallis, Oregon, USA
Ecological risk Assessment (EcoRA) began largely as a tool to address the environmental component of legal framework ′to protect human health and the environment.′ The practice generally considered potential toxicity effects to selected receptors in environmental settings. Rarely, however, were ecological processes and relationships assessed explicitly. For most small-scale situations, the emphasis on toxicity apart from ecology has worked well; for larger cases such approaches generally have performed unsatisfactorily. Development of spatially-explicit risk assessment approaches is making it possible to consider multiple species and multiple stressors (typical at most sites) concurrently. These new approaches promote explicit descriptions of ecological relationships including species-specific habitat characterization, metapopulation dynamics, and considerations of biological and physical stressors in addition to traditional chemical stressors. Hypothetical case studies will be used to illustrate the consequences of using landscape-level ecological relationships to obtain provide better approximations of real world dynamics and improved environmental management information.

692 (AKC-1117-638251) Effect of spatial heterogeneity on assessing population-level ecological risks.
Start time: 2:10 PM
Akcakaya, H. R.¹, ¹ Applied Biomathematics, Setauket, NY, USA
Assessing ecological risks at the population-level requires estimating endpoints such as risk of population decline, chance of recovery, and the distribution of future population sizes and growth rates. When the spatial distribution of the individuals, or the threats faced by the species are heterogeneous, such endpoints may depend sensitively on how these spatial heterogeneities are incorporated into the models used in the assessments. Spatial heterogeneities in the environment can be incorporated into (or ignored by) models at various levels of detail. In this talk, the relative importance of these levels of complexity and interactions are reviewed with examples, and the types of models that incorporate spatial structure for population-level ecological risk assessment are discussed. At the most basic level, a model can assume a single panmictic population, or a metapopulation with multiple (sub)populations. At a finer level, a metapopulation model might assume that all populations have identical parameters (such as survival, fecundity, carrying capacity, etc.) or it can incorporate population-specific values for these parameters. At the finest level, the metapopulation model can take location of each population into account, incorporating spatial variation in the interactions among populations (e.g., dispersal and spatial autocorrelation), for example by making these functions of the distance and habitat between the populations. In addition, there may be interactions between temporal and spatial heterogeneity in the form of temporal trends or random changes in spatial structure (for example, due to habitat dynamics, or time course of toxicity). Risk assessment examples from ecotoxicology and conservation will be used to illustrate the importance of spatial heterogeneity at these different levels.

693 (COL-1115-318045) Landscape Scale Assessment of Contaminant Effects on Cavity-nesting Birds.
Start time: 2:30 PM
Colestock, K¹, Gervais, J¹, Fair, J², Ryti, R³, Gonzales, G^{1, 2}, ¹ Utah State University, Logan, UT, USA² Los Alamos National Laboratory, Los Alamos, NM, USA³ Neptune and Company, Inc., Los Alamos, NM, 87544
We studied the effects of contaminants on reproductive parameters of a western bluebird (Sialia mexicana) population nesting across a contaminant gradient at Los Alamos National Laboratory, New Mexico. Bluebirds are regional migrants, so we restricted our investigation to juveniles to explore local contaminant uptake. We evaluated reproductive success in relation to soil contamination levels in the individual home range. Home range size was determined using foraging observations, and the observed home ranges were correlated to measured and interpolated soil contaminant data. Eggs were analyzed for PCBs, heavy metals, and radionuclides and prey remains were collected from study nests to determine the potential contaminant uptake from the home range. We also measured nesting success, fledging success, and fluctuating asymmetry of nestlings. These results are compared to contaminant concentrations and to predicted hazard quotients from a spatial exposure model. This provides a comparison of measured and modeled exposure information and also provides a validation of the modeled effects information with field data.

694 (GOO-1117-754469) Spatially Explicit Ecological Risk Assessment for Pesticides: Applications of the GeoSpatial Exposure Model (GeoSEM).
Start time: 2:50 PM
Thayer, W¹, Goodrum, P¹, Negley, T¹, ¹ Syracuse Research Corporation, Syracuse, NY, 13212
Spatially explicit exposure models can be used in ecological risk assessments to link exposure concentration data with information and assumptions regarding a receptor's habitat use. The spreadsheet model developed by Hope (2005, HERA 11:1-27) provides a framework for investigating alternative exposure scenarios for receptors that exhibit one of different foraging strategies: territorial, nesting, and wide ranging. The same rules for receptor movement behavior were incorporated into the Geospatial Exposure Model (GeoSEM). Case studies were developed to illustrate the application of the spatially explicit model to ecological risk assessments at a scale relevant to agro-environments in which terrestrial receptors (small mammals and birds) may be exposed to pesticide residues both within and adjacent to treated fields. Concentration-time profiles of pesticide residues in potential exposure media (water, soil, forage items) were estimated with fate and transport models typically used by U.S. EPA Office of Pesticides (e.g., PRZM v.3.12, EXAMS v.2.98.04.06, TREX v.1.13). Exposure factors such as ingestion rates, field metabolic rate, and habitat use preferences were obtained from the open literature and Agency guidance. A methodology is proposed to account for multiple chemical stressors for the case in which pesticides that share a common mode of action are applied during the same time of year within the same habitat use area. Population-level effects are estimated by repeating simulations of individuals until population density estimates are matched. Results of simulations with small mammals are linked to a simple food web model for a predatory bird species. Using sensitivity analysis, this case study highlights the relative importance of risk assessment assumptions, including habitat use preferences, foraging strategies, and accounting for multiple chemical stressors. The tool can also be used to inform risk management options that may be most effective at reducing risks, including decisions involving pesticide use practices and land use practices.

(58058) COFFEE BREAK.
Start time: 3:10 PM

695 (LAN-1117-842913) Invasive species ecological risk assessment methodology at landscape scales.
Start time: 3:50 PM
Landis, W¹, Colnar, A¹, Kaminski, A¹, Kushima, G¹, Seebach, A¹, ¹ Institute of Environmental Toxicology, Western Washington University, Bellingham, Washington, United States
The invasion by a novel species is a phenomenon that is influenced by the characteristics of the invasive and also the characteristics of the receiving landscape. We have now conducted regional scale ecological risk assessments for four invasive species, The risk assessment for the EGC and Sargassum was for the Northwest Washington Coast, the Asian Oyster assessment area was the Chesapeake Bay of Maryland and Virginia, and the Nun Moth assessment region was the terrestrial area within the Mid-Atlantic States. The method was derived from the relative risk model with particular attention to the incorporation of the Hierarchical Patch Dynamics Model (HPDM). The four sets of assessment had common patterns. First, because of the variety of endpoints some can be adversely affected at the same time other endpoints are enhanced. Contaminants can be important disturbances but in the context of altering the landscape patterns. Uncertainty can be clearly documented including the use Monte Carlo techniques. A common difficulty was the lack of suitable habitat data at appropriate scale and grain for the assessment of a particular invasive. Although some of the species are renown invasives, there is still a species specific lack of basic ecological information on the process of invasion. This lack of information includes the effects of toxicants, even for the herbicide or pesticide used for control. However, it does appear that a common approach for the evaluation of risk due to invasion does work for a variety of species and habitats.

696 (BUR-1117-831866) Programming Perspectives on Wildlife Exposure Modeling: SEEM and FR.
Start time: 4:10 PM
Burmistrov, D¹, Wickwire, WT¹, Stackelberg, K von¹, Menzie, CA¹, Johnson, MS², Bridges, TS³, Dortch, MS³, ¹ Menzie-Cura & Associates, Inc., Winchester, MA, USA² Health Effects Research Program, US Army Center for Health Promotion and Preventive Medicine, Aberdeen Proving Ground, MD, USA³ United States Engineer Research and Development Center, Us Army Corps of Engineers, Vicksburg, MS, USA
Increasingly sophisticated analytical tools offer the risk assessment community the ability to improve the realism of wildlife exposure estimates. New models offer to account for the intersection of spatially distributed contamination and wildlife habitat and to record the impact that the intersection has on exposure estimates, e.g. high quality habitat intersecting with low contaminant levels. The project team is developing two spatially explicit exposure models: the Spatially Explicit Exposure Model (SEEM) for terrestrial wildlife and FishRand (FR) for aquatic wildlife. Both models are currently being integrated into the U.S. Army Risk Assessment Modeling System (ARAMS). Because a model is an interpretation of reality and cannot be a perfect representation of nature, we make a number of simplified assumptions regarding wildlife behavior. Understanding the applied modeling algorithms is important in order to understand the impact of the assumptions and the uncertainties that emerge from the assumptions on the final model outputs. Both SEEM and FR are Monte Carlo models. They use different modeling approaches in recognition of behavior differences between terrestrial and aquatic ecosystems. Both incorporate habitat suitability as a characteristic that increases the likelihood of exposure, and employ movement-based individual foraging. Movement rules are used to capture different foraging strategies. Markov Chain Monte Carlo is used to create a habitat quality weighted, though randomized, movement/foraging pattern for each individual (SEEM) or group of individuals (FR). Both models are powerful tools that provide assessors with the opportunity to explore wildlife exposure from the perspective of populations with exposure impacted by habitat availability and suitability. With an understanding of the underlying mathematics and assumptions, users can avoid the common model application pitfall of applying results without a full understanding of the assumptions used to create the model.

697 (PUR-1117-839862) Use of habitat-contamination spatial correlation to determine when to perform a spatially explicit ecological risk assessment.
Start time: 4:30 PM
Purucker, S¹, Stewart, R¹, Welsh, C¹, ¹ University of Tennessee, Knoxville, TN, USA
Anthropogenic contamination is typically distributed heterogeneously through space. This spatial structure can have differing effects on the cumulative doses of individuals within the environment. These effects are exacerbated when individuals pursue different movement strategies, and movement strategies can be affected by how individuals and species value habitat. Habitat quality is often neglected when ecological risk assessments are performed, though inclusion of a quantitative habitat measure can have a significant effect on the overall exposure estimate. This presentation quantifies the range of the effect, by coupling an exposure model with an HSI model, to examine the interactions between habitat preferences, spatial distribution of contaminants, and foraging behavior. We construct dose distributions to characterize the range of exposure to Pronghorn deer exposed to flouride when foraging on desert sagebrush. The results show the magnitude of the difference between doses when foraging concentrations are positively or negatively correlated to varying degrees with habitat preferences. In addition, it examines the strength of correlation between the HSI data and the contaminant data as a proxy to determine whether HSI data should be incorporated into a spatially-explicit ecological risk assessment.

698 (KIK-1117-837309) Spatially Explicit Risk Assessment: Population Models and Scenarios.
Start time: 4:50 PM
Kiker, G¹, Pandey, V¹, Linkov, I², ¹ University of Florida, Gainesville, FL, USA² Cambridge Environmental, Cambridge, MA, USA
The Questions and Decisions (QnD) modeling framework has been developed as a spatially explicit risk assessment tool which combines both mechanistic modeling and scenario-based, value-laden planning. The RiskTrace software was originally developed as a spatially explicit exposure assessment model that provides a time series estimation of soil and food contamination that receptors may encounter in their daily movements. Combining these two modeling approaches has allowed both individual-based models and metapopulation models to be tested and compared within an adaptive, scenario-based modeling framework. QnD-simulated ecosystems are represented by combinations of component, process and data objects that are constructed through the use of XML-based, input files. This design allows different ecosystem/habitat/organism/chemical combinations to be efficiently formed, simulated and documented. Individual organisms, designed and tested within the RiskTrace model, were implemented within the QnD landscape for a demonstration terrestrial ecosystem. Both metapopulations and single individuals were simulated to explore the important effects of object design on simulation results.

699 (SAM-1117-133201) Developing Spatial Averages for Discontinuous Chemical Contamination.
Start time: 5:10 PM
Samuelian, J.¹, Alsop, W.², ¹ AMEC Earth & Environmental, Portland, ME, USA² AMEC Earth & Environmental, Westford, MA, USA
The determination of the representative soil concentration of a chemical contaminant (Exposure Point Concentration; EPC) is integral to the evaluation of potential ecological or human health risks at chemical waste sites. EPA has developed guidance for using the upper 95th confidence limit (95UCL) of the mean concentration for deterministic risk assessments as well as the supporting software ProUCL. These methods do not account for the spatial representativeness of the samples. Although geostatistical tools, such as splining or kriging, are available that account for the spatial distribution of the chemical results, many of these methods have an underlying assumption that a continuous distribution of chemicals is present at the site. In the case of particulate contamination, such as explosives, the sample results can have a discontinuous distribution at the site. We will demonstrate some alternate approaches to developing an appropriate spatially representative mean and 95UCL for such data sets. Furthermore, we will also discuss how composite samples have been used to develop spatial averages for risk assessments that better represent the average concentrations of chemical contaminant when compared to discrete samples.