190 likes | 346 Views
Using Regional Models to Assess the Relative Effects of Stressors. Lester L. Yuan National Center for Environmental Assessment U.S. Environmental Protection Agency.
E N D
Using Regional Models to Assess the Relative Effects of Stressors Lester L. Yuan National Center for Environmental Assessment U.S. Environmental Protection Agency The views expressed in this presentation are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.
Motivation • Relevant questions for an assessment: • Which stressors are important? • Where should we be concerned about different stressors? • What is the ecological effect of a given stressor? • Relative to other stressors. • Within a regional context.
Which stressors are important? • Important stressors influence valued ecological attributes. • Valued ecological attribute in small streams? Fish index of biotic integrity Macroinvertebrate index of biotic integrity Presence of certain species Biodiversity
Regional Study Area and Data • Mid-Atlantic Highlands Assessment • US EPA Environmental Monitoring and Assessment Program • Data collected 1st – 3rd order streams from 1993 – 1996. • Biological, chemical, and physical habitat data • Indices of biotic integrity (IBI) developed for fish and macroinvertebrates.
Stressor-response model • Principle component analysis of stressor variables • Generalized additive model • Model each response as an arbitrary smooth curve. • Allows for nonlinear relationships. • Identify stressors using a stepwise modeling procedure: • Total phosphorus (PTL) • Nitrate (NO3) • Sulfate (SO4) • Physical habitat (RBP) • Acidity (pH)
Model Results R2 = 0.22
Where should we be concerned about different stressors? • Can we estimate stressors levels in unsampled streams? • Mid-Atlantic data is not dense enough… • Focus study down to a smaller area.
Study Area and Data Western Maryland Mid-Atlantic USA • MARYLAND BIOLOGICAL STREAM SURVEY • Maryland Department of Natural Resources • Stratified, random sampling of 1st – 3rd order streams in Maryland. • Collected biological, chemical, physical habitat data. • Stressors available: NITRATE, SULFATE, and ACIDITY.
Interpolating Stream Stressors • Estimate stressor distributions from sampled data. • Model mean stressor levels using regression models. • Spatially interpolate residuals from regression.
Modeling Stressor Levels (Part I) • Develop models to predict mean values for SO4, NO3, and PH. • Explanatory variables: • Percent Agriculture • Percent Urban • Sampling Year • Catchment Size
Modeling Stressor Levels (Part II) Use spatial statistics to interpolate residual variation.
Model Performance SO4 NO3 R2 = 0.81 R2 = 0.63 PH Predictive power of model is reasonably high! R2 = 0.40
Spatial Distributions of Stressors NITRATE SULFATE PH
How much do individual stressors affect valued ecological attributes? • Combine regional stressor-response relationships with spatial distributions.
Scaled Stressor Maps NITRATE SULFATE PH
Issues • Reconciling differences among studies • Different measurements of qualitative habitat scores. • Different analytical techniques for water chemistry. • Effectiveness of spatial interpolation varies by stressor. • Correlation does not imply causation. • What is SO4 measuring?
Conclusions • Spatial interpolation is promising approach for imputing information about unsampled streams. • Scaling variables appropriately can help interpret data.