UPRLIMET: Upstream regional LiDAR model for extent of trout

Project Description

This waterfall is a physical barrier that prevents fish from moving further up North Fork Ecola Creek in the Oregon Coast Range. USDA Forest Service photo.

Land managers need tools to accurately identify the upstream boundary beyond which no fish are found (referred to as the upper distribution) because fish-bearing streams receive more protection than streams without fish. To meet this need, scientists with the Pacific Northwest Research Station developed a model framework to identify discrete, upper limit points above which all stream reaches are classified as fishless. They then developed the UPRLIMET data dashboard to visually display the data used in the UPRLIMET model, enabling users to see fish predictions and the upper extent of fish points in western Oregon streams.

Development of the tool, which is a logistic regression model, was informed by stream data on fish occupancy and geophysical aspects of the landscape. The research team also used novel LiDAR data (an optical remote-sensing technique that uses laser light to densely sample the surface of the Earth) and calibrated their model with observations from about 100 sites across federal, state, and private land ownerships in western Oregon. Four factors rose to the top in terms of predicting trout presence: (1) upstream channel length above the uppermost fish, (2) drainage area, (3) slope—also known as gradient, and (4) elevation.

Historically, the factors affecting the distribution of stream fish were evaluated at a local scale. UPRLIMET takes advantage of advances in remote-sensing technology and the availability of large-scale spatial data. The tool can provide policymakers with the information they need to design regulations in a transparent and equitable fashion.

Click to view UPRLIMET ArcGIS StoryMap in fullscreen.*

*The use of trade or firm names is for reader information and does not imply endorsement by the U.S. Department of Agriculture of any product or service.

Purpose and Scope

Accurate information about the distribution of fish in streams is crucial because unmapped or misclassified streams often are not protected under existing regulations. Credible, spatially explicit information can help resolve conflict when there are competing management interests.

The model was validated by using fish observations in western Oregon. However, the research team suggests it is reasonable to infer findings from UPRLIMET across regions with similar climatic and landscape conditions, which includes northwestern California, western Washington, and southwestern British Columbia.

Methods

In 2017, we sampled populations of coastal cutthroat trout (Oncorhynchus clarkii clarkii) that were randomly selected in historical surveys in 1999 and 2000 from across western Oregon’s ecoregions and were found above barriers. Sampling these trout populations indicated a starting point for the current assessment, allowing us to assess the uppermost fish occurrence and identify habitat barriers for populations across western Oregon on private, state, and federal lands.

The spatial domain for this study was limited to the 383 USGS 12-digit hydrological unit code (HUC12) subwatersheds in western Oregon having LiDAR-derived digital elevation models and associated LiDAR-derived hydrography in the National Hydrography Dataset.

We compared 26 models to predict upper distribution limits of trout in streams. We used machine learning, logistic regression, a sophisticated nested spatial cross-validation routine to evaluate predictive performance while accounting for spatial autocorrelation, and a stopping rule. The best predictive performance came from a two-stage model that uses a logistic regression algorithm calibrated to observations of coastal cutthroat trout occurrence and variables representing hydro-topographic aspects of the landscape. We predicted trout presence along reaches throughout a stream network and included a stopping rule to identify a discrete upper limit point above which all stream reaches are classified as fishless.

Implementation

UPRLIMET has the potential to facilitate collaboration by providing spatially consistent maps of fish distribution across the region. This could make a major contribution to the implementation of the revised rules and prescriptions developed under Oregon’s Private Forest Accord. Signed into law in February 2022, the accord is expected to profoundly change how millions of acres of private forestland are managed in Oregon. The ability to accurately identify the upper distribution of fish is a key need for implementing the agreement.

Key Findings

• The team’s results suggest that there is no clear advantage to using a more complex model (i.e., one based on more environmental predictor variables than UPRLIMET), as evidenced by the superior predictive performance of the four variables used in UPRLIMET.
• Stream size accounts for the top two variables in the model, which suggests that it is the major driver of the upper distribution limit of fish. The probability of trout presence increases with increasing upstream stream length and upstream drainage area. This finding suggests that downstream stream reaches are more likely to have fish.
• UPRLIMET predicted more fish on private lands than on land managed by the state, U.S. Forest Service, or U.S. Bureau of Land Management, highlighting the importance of using transparent, spatially explicit maps across a region and working across ownerships when developing management plans for fish and forests.