Research
Model-Assisted Design for Ecological Sampling
Collaborators: J-Y Courbois (UW Statistics), R Buchanan (UW Quantitative Ecology & Resource Management), D Stevens (Oregon State Universiy), S Urquhart (Colorado State University)
Support: Oregon State University
An important issue for ecological sampling is that of optimizing designs for population and process inference. A good application is that of designing sampling schemes for sampling river networks. River networks have been successfully modeled as Markov chain processes that allow for correlation running downstream (flow direction) and also upstream (many biological processes such as salmon migration). An effective monitoring design needs to account for such network processes. Using model-assisted designs, we intend to develop sampling strategies that estimate model parameters and, at the same time, address traditional monitoring purposes, tracking biological, chemical, geological responses through time.
Habitat Monitoring of Stream Restoration Sites on Oregon's North Coast
Collaborators: T. Sibley (faculty) and B. Thom (graduate student)
Support: Oregon Wildlife Heritage Foundation
In summer 1995, woody debris in the form of large trees were added to streams with the goal of enhancing salmonid habitat. Initial monitoring of the projects involved evaluation of the stability of the placed wood in all projects completed in 1995 as well as monitoring of channel morphology and habitat in 10 paired reference and treatment stream segments. The goals of the initial monitoring were to evaluate the treatment techniques for their ability to create salmonid habitat, and estimate the potential basin wide effects of similar projects that will be carried out in the future. A majority of the placed wood remained stable through two large (5-year, 100-year) storm events with only 18% of the wood showing noticeable movement. Channel morphology and habitat changes were variable both within and between streams. Mildly significant increases (p <.10) have occurred in the quantity of pools preferred by overwintering coho salmon (Oncorhynchus kisutch), the amount of gravel in riffles, the number of complex pools, and the number of key pieces of woody debris. Overall, habitat has been enhanced for salmonids, especially coho salmon, and many habitat attributes fall within the range of pre-managed conditions. Basin-wide effects from the projects already completed are minimal compared with the large area of streams with a similar size and gradient that could potentially benefit from woody debris additions.
Ecosystem Management Research in the Pacific Northwest: A Regional Research Consortium
Collaborators: C. Simenstad (staff), T. Quinn (faculty), P. Bisson (USFS), R. Edwards, S. Bolton (UW College of Forest Resources), D. Montgomery (UW Geological Sciences), and N. Mode (graduate student)
Support: U.S. Environmental Protection Agency
The University of Washington, Oregon State University, and University of Oregon have created the Pacific NW Ecosystem Research Consortium (PNW-ERC) designed to (1) create a core research program and conceptual framework for ecosystem management research, (2) stimulate and support research on regional issues, and (3) participate in advisory and external review processes to guide research development. We are working to create a regional landscape context for managing ecosystems, using the Washington Coastal Ecoregion and the Willamette River basin as major examples.
Reduction of Seabird Bycatch in Salmon Drift Gillnet Fisheries
Collaborators: E. Melvin (WSG)
Support: Washington Sea Grant Program under the National Oceanic and Atmospheric Administration, U.S. Dep. Commerce
As part of a process to develop fishing gear and fishing techniques that reduce the accidental entanglement of endangered seabirds, we compared entanglement rates and fish catch rates in different types of salmon fishnets, and across different times of day. We used loglinear models to analyze both bird entanglement and fish catch data. Results indicate that seabird entanglement and salmon catch rates vary significantly among fishnet types and times of day. Particular patterns of variation were specific to the type of bird being investigated. For example, Rhinoceros Auklet entanglement rates decreased with increasing depth of visual barriers in the fishnets while Common Murre entanglement rates were more similar to the control nets. Fishnets with acoustic "pingers" were more effective at reducing entanglement rates of Common Murres than that for Rhinoceros Auklets. Bird entanglement rates also varied according to different times of day. These results may lead to fishing policy changes that use visual net barriers, acoustic pingers for fishnets, and daytime-only fishing to reduce endangered seabird bycatch by commercial salmon fishing boats.
Biological Retrospective of Puget Sound
Collaborators: M. Silkey (graduate student, UW Quantitative Ecology and Resource Management)
Support: U.S. Forest Service
Retrospective Study of Aquatic Biological Work Conducted in the Freshwater Systems of Puget Sound
Collaborators: W. Staubitz, R. Black (USGS), and M. Silkey (UW QERM graduate student)
Source: U.S. Geological Service
Overview of aquatic biological work that has been conducted in the freshwater system of the Puget Sound basin.
Sampling of Ranked Sets of Data: An Investigation of Alternative Sample Designs in Environmental Monitoring
Collaborators: B. Nussbaum (EPA), D. Marker (Westat, Inc.), and N. Mode (graduate student)
Support: U.S. Environmental Protection Agency, through the National Center for Research on Statistics and the Environment
Sampling of ranked sets of data (ranked set sampling, or RSS) is a two-phase sampling procedure involving initial, ranking "more data than are needed" (often via a relatively cheap or fast, rough method of measurement), followed by observing (using a more accurate, more expensive method of measurement) a smaller, secondary sample of data. Examples involve assessing sediment contamination or measuring amount of area present for different habitat types in a stream or a wetland. The RSS is a statistical sampling approach that should yield a much more representative sample than simple random sampling (SRS). We propose to examine under what conditions RSS would be an appropriate and cost-effective procedure to use (including more complex designs, such as stratified sampling). This will involve both theoretical examination and applications to existing data sets.