Jump to main content.


An Integrated Exposure and Effects Model for Probabilistic Risk Assessment for Estimating Risks to Aquatic Organisms from Pesticides

K. Gallagher, J. Lin, D. Randall, D. Young, D. Farrar, T. Barry, Ian Kennedy, T. Bargar, L.W. Burns, I.M. Sunzenaeur

On this Page


Introduction

image with poster title, authors and images of a crop duster and aquatic life

Introduction

Top of page


Revising the Ecological Assessment Process in USEPA Office of Pesticide Programs (OPP)

Top of page


Objectives of Refined Ecological Assessment Approach

Top of page


Four Levels of Refinement

Top of page


Background: Current Deterministic Level 1 Process

Top of page


New Level I: Refined Deterministic Screen

Top of page


Integrated Exposure and Effects Model for a Level 2 Probabilistic Assessment for Aquatic Organisms

Consists of Three Modules:

  1. Effects Module: Uses the FULL concentration-response curves

  2. Exposure Module: Use the FULL range of the selected exposure distribution

    • Selected model input parameters can be entered as distributions

  3. Effects and Exposure analyses are combined yielding predictions of likelihood and magnitude of effect

Top of page


Aquatic Conceptual Model Flow Diagram

aquatic conceptual model flow diagram

Aquatic Conceptual Model Flow Diagram

Top of page


New Additions to Water Body

Top of page


Variable Volume Water Body

image of water body with runoff input, evaporation, infiltration, actual and maximum water depths

Variable Volume Water Body

Top of page


Draft Revised Level 2 Model Cover Page

image of cover page with options to go to Instructions, Background Information, Model Menu and Exit

Draft Revised Level 2 Model Cover Page

Top of page


Model Background

Background for the Aquatic RRA Model (Version 2.0)

The RRA Model estimates the joint probability acute risk function for an exposed aquatic population (i.e., determines the distribution of individual acute risks, based on the distribution of exposure data and distribution of toxic effects data) and the statistical uncertainty in the estimate of the joint probability function. This is accomplished using a 2-dimensional (2D) Monte Carlo Analysis program written in C by Timothy Barry for the Pilot Aquatic Risk Model, Version 1.0, and translated to MatLab. The basic exposure algorithm is

exposure = Inverse CDF (probability | parameters)

where probability is uniformly distributed, probability ∼ U(0,1), and parameters are the parameters of the fitted dose distribution. Inversion of the dose (or concentration) distribution is used to accommodate several distributions. Given an estimate of exposure, the risk is then calculated using the log-probit dose-response function

risk = Normsdist(intercept+slope*log(exposure))

where Normsdist is the Excel workbook function for the normal cumulative distribution function and intercept and slope are the parameters of the log-probit dose-response function. In order to run this model, characterizations of the uncertainty in each of the distributional parameters of the exposure distribution and dose-response function are needed. These are generally developed from the statistical uncertainty in the parameter estimates (e.g., derived from maximum likelihood) or based on expert judgment.

Top of page


User Directions

Instructions for Conducting a 2D Simulation

  1. An attempt has been made to make this user interface "user friendly" and relatively self explanatory. To run the 2D simulation model you will need to enter scenario-specific application data (e.g., chemical, crop, application rate, and location), which is referenced as Exposure Scenario Data in this program, and species-specific acute effects data (i.e., Effects Data). The specific Exposure Scenario data and Effects Data required to run the 2D model are described in the appropriate steps given below. There are instances where default values can be used rather than empirical data and this is identified where available. The program has been written to clearly document in the output the Exposure Scenario Data and Effects Data that was used to perform the simulation.

  2. You will need to enter a complete Exposure Scenario and Effects Data File for the RRA simulation to run. To start entering the data, return to the Main Menu and select Model Menu.

  3. The Model Menu page provides options for entering Exposure Scenario and Effects Data for a new project or opening an existing file (*.txt) and editing it as needed.

    1. Effects Input Data -- Your have three choices for the type of acute toxicity data required.
         (i)

IN YELLOW CELLS ONLY, Enter Best Estimates and Standard Errors for Exposure and Dose-Response Parameters, as well as their correlations, based on calculations made externally.
NOTE: Do not adjust values in green cells; Crystal Ball selects these during simulation.

This model is based on fitted chemical exposure distributions; the parameters' best estimates, their standard errors and correlation are required. Exposure data are based on pesticide usage on specific agricultural crops. the dose-response parameters are based on a log-probit fit to toxicity test data. This spreadsheet assumes that the individual parameters are normally distributed, with the specified central tendency (best estimate) and dispersion (stardard error).

3. In the Crop numeric cell (C15), change the fitted exposure distribution type for the inversion needed

  1. Clear cell contents if fitted exposure distribution is different from previous run.
  2. Use function wizard. select excel or user defined function as needed for particular fit inversion.
  3. Click on cells for shape (D6) and scale (D7) parameters as appropriate to assign function. Probability = cell C14

Directions are also given in cell comment on 2D page.

Top of page


Exposure Module (Excerpt showing some input parameters)

image of model screen with two input parameters, soil aerobic metabolism and aquatic aerobic metabolism

Exposure Module (Excerpt showing some input parameters)
Based on data available, user can input a 90th% value of a small data set, a multiple of an individual datum, or fit a distribution

Top of page


Effects Module (Data Entry Page)

image of data entry page for sheepshead minnow with columns for specification, dose, #exposed, #effected, LC50 and slope

Effects Module (Data Entry Page)
Based on data available, user can evaluate raw toxicity test data, use pre-analyzed data for LC50 and Slope, or use LC50 only

Top of page


Species Sensitivity Distribution

image of screen with data from analysis of individual species, user selection options for distribution for LC50 and slope and graphical display of Probability vs. LC50

Species Sensitivity Distribution

Top of page


Model Output Example

image of output example showing probabilistic risk estimates as a table and graphs for various species

Model Output Example

Top of page


Work In Progress

Top of page


Publications | Glossary | A-Z Index | Jobs


Local Navigation


Jump to main content.