Grantee Research Project Results

2004 Progress Report: Development of Watershed Classification Systems for Diagnosis of Biological Impairment in Watersheds and Their Receiving Water Bodies

EPA Grant Number: R830597
Title: Development of Watershed Classification Systems for Diagnosis of Biological Impairment in Watersheds and Their Receiving Water Bodies
Investigators: Martinko, Ed , Dobson, Jerome E. , Jakubauskas, Mark E. , deNoyelles, Frank J. , Whistler, Jerry L. , Liechti, Paul M. , Huggins, Donald , Thorp, James , Price, Kevin
Institution: University of Kansas
EPA Project Officer: Packard, Benjamin H
Project Period: March 3, 2003 through March 2, 2006
Project Period Covered by this Report: March 3, 2003 through March 2, 2004
Project Amount: $899,683
RFA: Development of Watershed Classification Systems for Diagnosis of Biological Impairment in Watersheds (2002) RFA Text |  Recipients Lists
Research Category: Watersheds , Water

Objective:

The objective of this research project is to produce a watershed classification system for predicting biological impairments and to recommend ecosystem rehabilitation in small watersheds (wadeable) through those supporting great rivers (non-wadeable). The classification system is based on the development and modeling of landscape metrics, both natural and anthropogenic, as indicators of stress on streams and rivers. Existing and newly collected stream sample data, used as system response indicators, are implemented to calibrate and test models. Once a predictive stressor model(s) is produced, several techniques will be explored to classify the model results into classes indicative of biological impairment.

Progress Summary:

Activities in Year 2 of the project have focused on: (1) computation of watershed boundaries for each sample point; (2) extraction of landscape geospatial data for each watershed, including static parameters and remote sensing-derived parameters; (3) correction, manipulation, and analysis of existing wadeable stream data from the four-state Central Plains; and (4) collection of the second year of field data from 40 non-wadeable rivers in the four-state Central Plains region, analysis of some first-year data, and processing samples from the first and second years.

Remote Sensing, Geographic Information Systems (GIS), and Model Development

The primary focus of Year 2 for the Landscape Characterization Group was twofold: (1) completion of the acquisition and generation of geospatial data sets required for development of the watershed assessment model(s), and (2) generation of watersheds based on sampled points. An essential philosophical precept guiding the data acquisition and generation process has been the use of data sets that are either readily available or easily produced by other researchers wishing to replicate our methodologies, as opposed to the production of project-specific, limited-distribution data sets constrained in both geographic extent and attributes. The geospatial data take the form of raster grids (remotely sensed imagery, digital terrain data) and vector coverages (e.g., hydrography, political boundaries). A milestone for the year was the assembly and completion of the landscape data sets (this task was comprised of several smaller, intermediate milestones) and the calculation of the Vegetation Phenology Metrics for the 15-year period 1989-2003.

The major milestone for the year was the automated delineation of more than 2,400 individual watersheds using sample point locations and digital elevation model data. Our intent was to generate accurate upstream watersheds for each field-sample used, whether it came from existing data sets (e.g., U.S Environmental Protection Agency [EPA] Region 7 REMAP 1994-95, TriState Study, state agency data) or from sites sampled specifically for this study. In a previous study, manual methods were used to define watershed boundaries (i.e., subjectively drawing watershed boundaries on topographic maps and manually digitizing them into a GIS coverage). Given the thousands of data points for this study, we deemed this to be impractical. Furthermore, we wished to develop a more objective, ideally faster, method of watershed delineation. Having developed the procedures for automated watershed delineation, another milestone reached was the automation of data extraction of landscape metrics. This was achieved through coding AML scripts in Arc/Info. The raw data to be extracted were evaluated for scale, accuracy, and general data quality (e.g., missing, incomplete, or out-of-range values), and processed to a common geographic projection. Once all watersheds were completely delineated, landscape predictor data were extracted for each watershed as source data for statistical analysis and modeling in Year 3 of the project. The Landscape Group also played a major role in determining water quality data availability and development of guidelines for processing water quality databases (i.e., temporal data issues, intra- and inter-annual mean/median determination, incomplete observations handling, and the use of search windows to match invertebrate and fish data with chemistry data).

Aquatic Components

The aquatic portion of our original proposal involved the acquisition of both existing data from the Central Plains and production of new data from 40 river sites. During the first two summers of our project (2003 and 2004), we collected fish and other organisms from rivers in the Central Plains and began processing samples. At the same time, we worked with the Central Plains Center for Bioassessment (CPCB), a Kansas Biological Survey Program, to get their massive data set (nearly 40 years of data, 2,300 sample sites in four states, and 52,000 sample events) on wadeable streams into a form useable for our model analysis. These data are currently being employed to develop (50% of the data set) and test (the remaining 50%) our watershed-lotic system predictive model. The final step will be to use the data collected from our 40 non-wadeable river sites to test the applicability of the model for larger watersheds. This aspect of our project will occur in Year 3 of the project following analysis of 2003-2004.

Staff from the CPCB completed processing databases obtained from state agencies in the four-state region (Kansas, Iowa, Nebraska, and Missouri). The databases contained, in varying degrees of completeness and detail, chemical and biological data from reference and other streams throughout the four-state region. This work was performed in conjunction with a project that CPCB is doing for the Biocriteria EPA Region 7 workgroup in which CPCB will collate all biological and chemistry data into one database. Dealing with gaps in the biological and chemistry databases for the reference streams was highly time-consuming for CPCB. The four-state stream chemistry database, however, has been completed, and the insect and fish databases should be completed in Year 3 of the project.

During Year 2 of the project, we sampled 40 river sites over an 11-week period in the four-state Central Plains. These rivers included samples in the mainstream of the Arkansas, Kansas, and Missouri Rivers and many tributaries flowing into them. Sampling included biota (fish, benthic invertebrates, zooplankton, and phytoplankton) and chemical analysis (onsite measurements from a Hydrolab meter along with U.S. Geological Survey information). Sampling has been concluded. We also have counted and identified to the species level 100 percent of the fish samples (~75,000 fish) from Years 1-2 of the project and have generated a basic Index of Biotic Integrity (IBI) for the fish. Although this portion of our project has gone well, we are still considerably behind in identification of our invertebrate samples because of a shortage of funds to hire personnel to sort and identify samples as rapidly as we would like. This resulted from the unexpectedly high cost for field equipment, supplies, and travel costs during the first summer’s research. Thus far, we have processed 95 percent of the samples (full samples rather than subsamples) and have identified 75 percent of the benthic invertebrate samples (again, full samples) to family. We also have identified and counted 15 percent of the zooplankton samples to genus.

Results of our aquatic project have resulted in presentations at two national/international science conferences, with at least two more presentations planned for the summer of 2005. Results from the project have contributed to two papers in press in international journals and to one paper that will be submitted to an international journal in June 2005.

Non-Wadeable Stream Data. Our watershed-stream model is being developed based on data from wadeable streams. The obvious question is whether this model can be scaled up to predict characteristics of aquatic condition for non-wadeable streams. To ascertain the effectiveness of the model for larger watersheds, we collected data over two summers from 40 medium to great river sites in the four-state Central Plains on abundance and diversity of fish, zoobenthos (soft sediments, hard sediment, macrophyte, and snag habitats), zooplankton, and algae. This has resulted in identification of approximately 75,000 fish. About 85 percent of the zoobenthos samples from Year 1 of the project (480 samples that have been picked completely rather than subsampled and are being identified to family) have been counted and identified and the remainder will be processed by the end of the summer 2005. A portion of the zooplankton samples were identified for Year 1 and used in a paper recently accepted for publication in Freshwater Biology. The remainder of Year 1 zooplankton should be identified and counted in Year 3 of the project. During the final year, most of the 2003 zoobenthos data will be analyzed, incorporated into our model, and published.

We have begun collecting extensive data sets upon which our watershed model will be based, and publications from this model will primarily be generated in Year 3 of the project and in the first year after the project is completed. Elements of the aquatic analysis from this project were presented in seminars to the University of Ferrara (Italy) and to the annual meeting of the North American Benthological Society in 2004 and 2005 (two papers).

Future Activities:

Remote Sensing, GIS, and Model Development

During Year 3 of the project, we will continue with the activities specified in our original proposal. Model development will be the primary focus of , subdivided into several areas: (1) exploration of landscape-water quality relationships occurring at a macro scale across all ecoregions; (2) exploration of landscape-water quality relationships on a within-ecoregional-specific basis; (3) examination of the effect of watershed area and landscape scale on the strength of observed relationships to elucidate scale-specific effects; (4) development of means for assessing impairment and vulnerability based on model development conducted in 1 and 2 (above), using both relativistic and calibrated approaches; and (5) extension of model development and testing to non-wadeable streams by means of testing whole-watershed, localized-floodplain, and modularized-watershed modeling approaches.

Aquatic Components

During Year 3 of the project, we will continue with the activities specified in our original proposal. Processing and analysis of the previously collected biotic data from wadeable streams in the four-state Central Plains region will be completed. We will focus on incorporating these wadeable stream data into our watershed model along with landscape predictor variables derived from remote sensing and GIS data. The remaining benthic invertebrate samples collected from the 40 non-wadeable rivers in Year 1 will be analyzed by the end of summer 2005. We will refine metrics and IBI’s for our river zoobenthos data to go with our fish metrics developed in Year 2 of the project.

Supplemental Keywords:

aquatic response variables, benthic invertebrates, geographic information systems, Iowa, Kansas, landscape characterization, landscape indicators, large rivers, lotic ecosystems, Midwest, Missouri, NDVI, Nebraska, phytoplankton, potamoplankton, Region VII, response variables, vegetation phenological metrics, wadeable streams, watershed analysis, zooplankton,, RFA, Ecosystem Protection/Environmental Exposure & Risk, Scientific Discipline, INTERNATIONAL COOPERATION, ECOSYSTEMS, Water, Aquatic Ecosystem, Aquatic Ecosystems & Estuarine Research, Ecological Risk Assessment, Terrestrial Ecosystems, Watersheds, Monitoring/Modeling, Water & Watershed, Environmental Monitoring, watershed management, biodiversity, restoration planning, biological impairment, water quality, diagnostic indicators, watershed assessment, bioassessment, biota diversity, watershed restoration, ecosystem indicators, ecosystem monitoring, ecosystem response, ecosystem stress, aquatic biota, conservation planning, bioindicators, watershed classification, aquatic ecosystems, anthropogenic stress, biological indicators, anthropogenic processes, conservation

Relevant Websites:

http://www.kbs.ku.edu/epawatershd/index.htm Exit

Progress and Final Reports:

Original Abstract

2003

Final