Grantee Research Project Results
Final Report: Testing Watershed Classifications Relevant to Bioassessment, Conservation Planning, and Watershed Restoration
EPA Grant Number: R830594Title: Testing Watershed Classifications Relevant to Bioassessment, Conservation Planning, and Watershed Restoration
Investigators: Hawkins, Charles P. , Tarboton, David G. , Stevenson, R. Jan , Higgins, Jonathan , Lammert Khoury, Mary , Baker, Michelle , Cao, Yong
Institution: Utah State University , Nature Conservancy, The , Michigan State University
Current Institution: Utah State University , Michigan State University , Nature Conservancy, The
EPA Project Officer: Packard, Benjamin H
Project Period: January 1, 2003 through December 31, 2005 (Extended to November 30, 2006)
Project Amount: $853,515
RFA: Development of Watershed Classification Systems for Diagnosis of Biological Impairment in Watersheds (2002) RFA Text | Recipients Lists
Research Category: Watersheds , Water
Objective:
The primary objective of our research project was to test the effectiveness of a systematic approach for developing watershed classification schemes useful for environmental assessment and monitoring of aquatic ecosystems. To do so, we needed to identify the specific watershed classification schemes of greatest utility for biological assessment, conservation planning, and the diagnosis of anthropogenic stressors for stream ecosystems in the western United States. To address this objective, we attempted to answer four questions:
- How effectively do classifications derived from single types of watershed and reach attributes perform in partitioning naturally occurring biotic variation?
- Can sequential application of classifications based on different types of watershed attributes provide insight regarding the stressors affecting aquatic ecosystems?
- Can a watershed classification derived from a multivariate analysis of the joint variation in different types of watershed attributes achieve greater effectiveness in partitioning biotic variation among watersheds than classifications based on single factors?
- To what degree can we infer aspects of ecosystem function from watershed classifications that predict biotic structure (i.e., composition)?
Addressing these questions required that we pursue several tasks related to understanding and quantifying the natural physical and chemical variability among western watersheds in addition to quantifying how stream biota respond to stressors.
Summary/Accomplishments (Outputs/Outcomes):
1) Development of rapid watershed delineation tools.
Our study required that we be able to characterize the watersheds of each study stream, which meant deriving the watershed boundaries for hundreds of catchments. To efficiently create such data, we developed a Windows-based application (The Multi-Watershed Delineation Tool) to delineate multiple watersheds in batch mode (Hill, et al., 2007; Chinnayakanahalli, et al., 2007a). This tool uses functionalities of TauDEM and ArcObjects (http://edndoc.esri.com/arcobjects/8.3/) Exit to quickly delineate a large number of watersheds. This tool also is capable of computing DEM-derived watershed and channel network attributes such as watershed area; statistical summaries of various watershed features such as elevations, precipitation, and temperature; stream network summaries; slope of each stream segment; and contributing area at upstream and downstream ends of each stream segment.
We have adopted the TauDEM terrain analysis software to delineate and derive watershed attributes for the approximately 800 gauged, reference watersheds and 112 impaired watersheds. With the availability of improved Digital Elevation Models (DEM), digital watershed delineation tools like TauDEM (Tarboton and Ames, 2001, http://hydrology.usu.edu/taudem/taudem4.0/index.html) Exit and ArcHydro (Maidment, 2002, http://www.crwr.utexas.edu/gis/archydrobook/Archydro.htm) Exit produce reliable watershed boundaries. They also can be used to compute various DEM-derived watershed attributes like relief, drainage area, and watershed shape. Nevertheless, these tools are limited by DEM grid size, do not separately delineate nested watersheds, require outlets to be precisely located on streams and do not calculate all the attributes of interest. The Multi-Watershed Delineation Tool solves all of these issues.
2) Classification of watershed geology and predictive models for water chemistry.
Watershed geology is a primary driver of several stream attributes including stream bed substrate character and water chemistry. We reclassified state geology coverages to provide a classification more relevant to stream processes (Olson, 2008; Olson and Hawkins, 2003, 2005). We now can predict important aspects of base flow water chemistry as well as the potential of watersheds to produce specific types of stream bed substrates. We tested the new classification through our continuing collaboration with state water quality agencies in the Western United States. For example, the new classification was used in Wyoming, a state with particularly heterogeneous geology to refine predictions of the invertebrate biota expected at Wyoming streams. These predictions allowed the Wyoming Department of Environmental Quality to more accurately assess biological impairment and the likely reasons causing it (Harget, et al., 2007). We also have used the classification in the State of Idaho to predict water alkalinity, conductivity, nutrient levels at base flow, and the watershed rock hardness. These models allowed us to develop expected conditions in support of a state-wide periphyton index (Cao, et al., 2007a).
3) Classification of watersheds based on thermal regime and prediction of water temperature based on watershed and climate variables.
The thermal regime is a critical driver of the distribution of stream organisms; however, estimates of natural water temperatures are lacking for most streams. We have developed multiple linear regression models to predict naturally occurring stream water temperature from long-term climatic data, watershed characteristics, and channel attributes (Hill, 2008). We are using these models to characterize both natural distributions of stream biota, as well as the likely response of biota to thermal alteration.
4) Classification of watersheds based on flow regime and predicting key flow-regime variables based on watershed and climate characteristics.
Flow regime is another critical element influencing the distribution and abundance of stream biota. Because only a fraction of U.S. streams are gauged, we need methods for characterizing the natural flow regime of un-gauged basins. We examined a wide range of variables that have been used to characterize flow regimes and chose a subset that was considered highly relevant to stream biota. Based on a western-wide data set (800 sites), we classified natural flow regimes into 4, 6, and 8 different regimes and developed statistical models that predict flow-regime type from watershed size, watershed morphology, climate, and other factors (Chinnayakanahalli, 2008; Chinnayakanahalli, et al., 2006a,b, 2007b). Among the several modeling techniques we explored, the results obtained from Random Forests, a non-parametric technique, were most accurate. These models allowed us to characterize biotic response to natural flow regimes and assess the degree to which flow alteration affects biota.
5) Establishing different types of human-disturbance gradients to test whether sequential application of watershed classifications can provide insight regarding the stressors affecting aquatic ecosystems.
We selected and sampled 112 sites to establish five different types of primary disturbance gradients: urbanization, livestock grazing, row-crop agriculture, logging, and flow diversion. We used GIS models and satellite-image data to estimate the proportions of different land uses and road densities in each watershed. Population densities, average energy expenditure, and the intensities of other human activities for each watershed examined also were estimated. Land uses were further summarized by calculation of indices that characterized the amount of watershed/waterway alteration in each watershed (Bowman, et al., 2007).
6) Identifying general indicators and stressor-specific indicators.
We have examined how several indices of biological condition (O/E, EPT [mayflies, stoneflies and caddis flies] richness, percent of individuals as EPT, and dominance) vary along the established disturbance gradients. The responsiveness of these indicators to the same type of disturbances differed significantly from one location to another. Both O/E and EPT richness clearly captured the urbanization gradient in Portland, OR, less so in and around Salt Lake City, UT, and failed to do so in and around Denver, CO. These results imply that we need to develop a better understanding of the natural variability of biotic metrics (Bowman, et al., 2007).
7) Modeling the responses of biotic metrics to natural environmental gradients.
We found that the natural variation in biotic metrics often is so substantial that natural variation can mask real responses to general or specific human caused disturbances. Therefore, we needed to develop ways of adjusting for the effects of natural gradients on these indices. We found that Random Forests models derived from reference sites were particularly successful in quantifying these natural-gradient effects (Hawkins, et al., 2007a).
8) Examining the suitability of specific reference sites for evaluating biotic condition at disturbed sites. Ecoregions and other types of landscape strata are commonly used to classify watersheds and infer biotic characteristics in streams. We found that the reference sites most appropriate for evaluating specific disturbed sites are seldom spatially aggregated. This result highlights the need to model relationships between individual indicators or biotic metrics and environmental variables when developing useful classifications of watersheds.
9) Comparing the responses of macroinvertebrate assemblage composition to different types of disturbances or in different areas.
We used non-metric multidimensional scaling (NMDS) to evaluate how strongly sites that varied in type and degree of watershed disturbance differed from reference sites in terms of their macroinvertebrate assemblages. We found that 1) the responses of stream macroinvertebrates to urbanization were stronger and more directional than the responses to other types of disturbances (logging, grazing, agriculture, and flow diversion); 2) natural regional signals were almost equally strong thus sometimes obscuring disturbance effects; and 3) the responses of macroinvertebrates to the five types of disturbances were rather similar, which illustrates the difficulty in diagnosing the causes of biological impairments based on field sampling data alone.
10) Simulating the responses of macroinvertebrate assemblages to different types of stressors.
As our primary results and other studies suggested, it was difficult to confidently identify the cause of biological impairments based on survey data. Therefore, we adopted a different strategy to deal with this challenge, i.e., simulating biological impairment with known stressors (Cao and Hawkins, 2005; Hawkins, et al., 2007a). Based on a literature review, we compiled tolerance values to various stressors for a large number of macroinvertebrate taxa, such as nutrients, conductivity, temperature, and fine sediments (Yuan and Hawkins, 2004). In our simulation, we related these stressor-specific tolerance values to how the abundance of a taxon at selected reference sites varied with increasing stress. These simulations appear promising for helping us understand how to best tease out stressor-specific responses of stream biota.
11) Quantification of ecosystem functional processes to natural gradients and human-caused stressors.
Water quality managers often implicitly assume that indices based on biotic structure (MMIs, O/E, etc.) provide insight regarding functional alterations in stream ecosystems. We examined rates of leaf litter decomposition across natural environmental gradients as well as stressor gradients and found that structural indices were poor predictors of litter decomposition (Simmons and Hawkins, 2004, 2005, 2007). Furthermore, we found that the range of natural rates of litter decomposition were so high that detection of alterations associated with stressors will require factoring out the effects of natural gradients via modeling in much the way we had to use models to improve the performance of structural indices. The models we developed imply that thermal differences among streams are primarily responsible for differences in litter decomposition.
Journal Articles on this Report : 12 Displayed | Download in RIS Format
Other project views: | All 39 publications | 15 publications in selected types | All 13 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Cao Y, Hawkins CP, Storey AW. A method for measuring the comparability of different sampling methods used in biological surveys: implications for data integration and synthesis. Freshwater Biology 2005;50(6):1105-1115. |
R830594 (2005) R830594 (Final) |
Exit |
|
Cao Y, Hawkins CP. Simulating biological impairment to evaluate the accuracy of ecological indicators. Journal of Applied Ecology 2005;42(5):954-965. |
R830594 (2005) R830594 (Final) |
Exit Exit Exit |
|
Cao Y, Hawkins CP, Larsen DP, Van Sickle J. Effects of sample standardization on mean species detectabilities and estimates of relative differences in species richness among assemblages. American Naturalist 2007;170(3):381-395. |
R830594 (Final) |
Exit |
|
Cao Y, Hawkins CP, Olson J, Kosterman MA. Modeling natural environmental gradients improves the accuracy and precision of diatom-based indicators. Journal of the North American Benthological Society 2007;26(3):566-585. |
R830594 (Final) |
Exit Exit |
|
Carlisle DM, Hawkins CP, Meador MR, Potapova M, Falcone J. Biological assessments of Appalachian streams based on predictive models for fish, macroinvertebrate, and diatom assemblages. Journal of the North American Benthological Society 2008;27(1):16-37. |
R830594 (Final) |
Exit Exit Exit |
|
de Zwart D, Dyer SD, Posthuma L, Hawkins CP. Predictive models attribute effects on fish assemblages to toxicity and habitat alteration. Ecological Applications 2006;16(4):1295-1310. |
R830594 (2005) R830594 (Final) |
Exit |
|
Hargett EG, ZumBerge JR, Hawkins CP, Olson JR. Development of a RIVPACS-type predictive model for bioassessment of wadeable streams in Wyoming. Ecological Indicators 2007;7(4):807-826. |
R830594 (Final) |
Exit Exit Exit |
|
Ostermiller JD, Hawkins CP. Effects of sampling error on bioassessments of stream ecosystems: application to RIVPACS-type models. Journal of the North American Benthological Society 2004;23(2):363-382. |
R830594 (2004) R830594 (2005) R830594 (Final) |
Exit |
|
Rehn AC, Ode PR, Hawkins CP. Comparisons of targeted-riffle and reach-wide benthic macroinvertebrate samples: implications for data sharing in stream-condition assessments. Journal of the North American Benthological Society 2007;26(2):332-348. |
R830594 (Final) |
Exit |
|
Van Sickle J, Hawkins CP, Larsen DP, Herlihy AT. A null model for the expected macroinvertebrate assemblage in streams. Journal of the North American Benthological Society 2005;24(1):178-191. |
R830594 (2005) R830594 (Final) |
Exit |
|
Van Sickle J, Huff DD, Hawkins CP. Selecting discriminant function models for predicting the expected richness of aquatic macroinvertebrates. Freshwater Biology 2006;51(2):359-372. |
R830594 (2005) R830594 (Final) |
Exit Exit |
|
Van Sickle J, Larsen DP, Hawkins CP. Exclusion of rare taxa affects performance of the O/E index in bioassessments. Journal of the North American Benthological Society 2007;26(2):319-331. |
R830594 (Final) |
Exit |
Supplemental Keywords:
EPA regions, CA, California, OR, Oregon, WA, Washington, ID, Idaho, NV, Nevada, AZ, Arizona, NM, New Mexico, CO, Colorado, WY, Wyoming, MT, Montana, UT, Utah, watershed classification, indicators, bioassessment, restoration, conservation, diagnostics, modeling, anthropogenic stressors, multivariate analysis, aquatic ecosystems, digital elevation model, RFA, Scientific Discipline, INTERNATIONAL COOPERATION, ECOSYSTEMS, Water, Ecosystem Protection/Environmental Exposure & Risk, Aquatic Ecosystems & Estuarine Research, Water & Watershed, Aquatic Ecosystem, Water Quality Monitoring, Monitoring/Modeling, Terrestrial Ecosystems, Environmental Monitoring, Ecological Risk Assessment, Biology, Watersheds, anthropogenic stress, bioassessment, anthropogenic processes, ecosystem monitoring, watershed management, biodiversity, conservation, diagnostic indicators, ecosystem indicators, biota diversity, aquatic ecosystems, bioindicators, watershed sustainablility, water quality, biological indicators, ecosystem stress, watershed assessment, conservation planning, ecosystem response, aquatic biota, restoration planning, watershed restorationProgress 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.