2010 Progress Report: Predicting Relative Risk of Invasion by Saltcedar and Mud Snails in River Networks Under Different Scenarios of Climate Change and Dam Operations in the Western United States

EPA Grant Number: R833833
Title: Predicting Relative Risk of Invasion by Saltcedar and Mud Snails in River Networks Under Different Scenarios of Climate Change and Dam Operations in the Western United States
Investigators: Poff, N. LeRoy , Raff, David A. , Shafroth, Patrick B. , Merritt, David M. , Bledsoe, Brian P. , Auble, Gregor T. , Purkey, David , Lytle, David , Friedman, Jonathan
Institution: Colorado State University , Stockholm Environmental Institute , U.S. Forest Service , U.S. Bureau of Reclamation , Oregon State University , United States Geological Survey [USGS]
Current Institution: Colorado State University , Oregon State University , Stockholm Environmental Institute , U.S. Bureau of Reclamation , U.S. Forest Service , United States Geological Survey [USGS]
EPA Project Officer: Packard, Benjamin H
Project Period: July 1, 2008 through June 30, 2012 (Extended to June 30, 2013)
Project Period Covered by this Report: July 1, 2009 through June 30,2010
Project Amount: $599,748
RFA: Ecological Impacts from the Interactions of Climate Change, Land Use Change and Invasive Species: A Joint Research Solicitation - EPA, USDA (2007) RFA Text |  Recipients Lists
Research Category: Global Climate Change , Aquatic Ecosystems , Ecosystems , Climate Change


The project seeks to predict the establishment and spread of invasive species in rivers subject to novel climatic conditions.  Changes in temperature and precipitation are expected to combine with human water needs to alter flow regimes in many watersheds.  Thermal shifts and novel discharge patterns may then influence population and community processes, potentially disfavoring native species while facilitating invasion by harmful non-natives.  The approach consists of linking output from a hydrologic model, driven by downscaled climate scenarios, to biological response models representing invasive population growth as a function of discharge, geomorphic setting and, potentially, community interactions.

Progress Summary:

The project team selected the Upper Green River basin as the primary study area on the basis of its ecological, agricultural and recreational importance, the potential for high-density tamarisk and New Zealand mudsnail invasion, and the presence of major water control infrastructure.  During this reporting period, the group completed construction and calibration of the hydrologic process model for the Upper Green River basin.  Additionally, a geomorphic classification model was applied to the study area, resulting in the identification of polygons with different habitat attributes.  Continuing exploration of biological response model options has led to the identification of three primary candidate approaches.

The Water Evaluation And Planning (WEAP) system was used to develop a watershed planning tool for the Upper Green River basin.  The WEAP system is an integrated water basin analysis tool that includes hydrologic processes within a water resources modeling framework such that climatic inputs can be used to drive the water planning model.  This allows for the consideration of how changes in precipitation and temperature may impact water supplies (through changes in snowmelt and runoff patterns) and water demands (through changes in crop evapotranspiration).  The tool also allows for the consideration of a variety of scenarios that explore the implications of management interventions through physical changes to the system, such as new reservoirs of pipelines, or social changes, such as policies affecting patterns of water use.
The WEAP model of the Upper Green River was developed at a spatial scale appropriate to simulate major hydrologic flows and exchanges and to evaluate the effects of water management responses.  In general, the model is organized by the many sub-catchments (i.e. tributaries) within the watershed (Figure 1).  These sub-catchments were disaggregated along watershed boundaries and elevation bands to identify inflows to the main reservoirs and to simulate elevation- or temperature-dependent hydrologic processes.  Within each of these banded sub-catchments, we divided the area by land use classifications (barren, forested, shrub, grassland, agriculture, developed) based on their different hydrologic responses (i.e. rainfall runoff) to climatic inputs.  The soil physical properties were assumed uniform across land use types and averaged within each of the banded sub-catchments.
Figure 1. The sub-catchments in the Upper Green River basin showing the main hydrologic pour points (white triangles) used in the Watershed and Evaluation Planning (WEAP) model.
The disaggregation and characterization of sub-catchments was performed by conducting a GIS analysis to delineate sub-catchment boundaries and to characterize the land uses and soil types within those sub-catchments.  A 30 m DEM dataset was used to delineate the main sub-catchments that control surface water inflow to the Upper Green River.  The same DEM layer was used to divide each watershed into banded elevation catchments, using 250 m elevation intervals.  We used the USGS' SSURGO and National Land Cover Dataset to assess the distribution of soils types and land uses, respectively, within each of these banded sub-catchments.  This assessment considered both natural areas in the upper watershed that control runoff and the developed areas that define the water use areas within the basin. 
The calibration of the WEAP model has been conducted incrementally.  We focused first on the unmanaged portion of the upper watershed and calibrated to measured snowpack and streamflow.  First, we adjusted model parameters—freezing and melting temperatures—to assure that the timing and magnitude of snow accumulation and snowmelt was consistent with historical records of NRCS snowpack measurements.  Then, we calibrated to unimpaired streamflow (USGS) by adjusting soil physical (i.e., hydraulic conductivity, soil water capacity) and land use (i.e., crop coefficient, surface roughness) model parameters.  Lastly, we focused on the managed part of the watershed and calibrated the model to observed historical surface water storage and deliveries.
During this reporting period, a semi-automated Geomorphic Valley Classification (GVC) system developed by the Bledsoe group was applied to the study area (Figure 2).  This set of ArcGIS scripts translates a pre-processed 10 m USGS DEM dataset into polygons on the basis of regionally pre-defined valley and hill-slope divisions.  After building a channel network using regionally appropriate flow accumulation criteria, the method distinguishes an initial set of valley units based on geomorphic structure.  The user then reviews this output, correcting errors due to DEM inconsistencies (i.e. stitching problems, banding due to low-relief surfaces, etc.) as necessary.  Following this step, the corrected units are re-aggregated into a final layer consisting of hierarchically ordered surface types, nested with Level 3 Ecoregions (USEPA).  These types facilitate more mechanistic biological response modeling by allowing construction of relationships that account for the factors such as stream power (i.e. the likely biological impact of a given discharge in a particular slope and channel constraint setting).  The location, extent and network orientation of valley and stream settings can be used to impose constraints on biological response models.
Figure 2. A sub-catchment in the Upper Green River basin showing the valley class types (colored lines) generated by the Geomorphic Valley Classification (GVC) model.
Recognizing the value of comparing more than one method of biological response modeling, the Poff group has pursued classification models based on machine learning algorithms (recursive binary partitioning using conditional inference), population matrix projection (in collaboration with Merritt and Lytle) and individual based applications.  Development of all of these options is ongoing, with the conditional inference trees nearing completion, and the matrix and individual based models nearing the parameterization and testing phase.  We expect these tools to be fully operational during the next reporting period.
Quality Assurance: Geospatial data used in developing and implementing the WEAP system for the Upper Green River basin were acquired from federal and state government sources, which disseminate remotely sensed data with appropriate metadata and maintain stringent guidelines on quality assurance/quality control.  Similarly, data on hydrologic attributes used in the WEAP model and data on geomorphic characteristics used in the GVC model were acquired from the U.S. Geological Survey, which also follows strict QAQC procedures.  Outputs generated with the WEAP and GVC models will be ascribed with essential metadata.
Results: This reporting period involved the completion of the hydrologic and geomorphic model components and further development of the biological response model.  Some of the results of the biological response model are in preparation for publication.  Additionally, (see Auerbach and Poff below), the group expects publication of a theoretical simulation model that was an outgrowth of some of the community dynamics problems raised by the project objectives.

Future Activities:

During the next reporting period, the project team expects to conduct meetings to develop the future climate and management scenarios, begin WEAP model runs using these scenarios while integrating the WEAP output with the GVC units, finalize the biological response model components, and perhaps create the preliminary composite projections of invasions in the Upper Green River Basin.

Journal Articles on this Report : 1 Displayed | Download in RIS Format

Other project views: All 25 publications 7 publications in selected types All 5 journal articles
Type Citation Project Document Sources
Journal Article Auerbach DA, Poff NL. Spatiotemporal controls of simulated metacommunity dynamics in dendritic networks. Journal of the North American Benthological Society 2011;30(1):235-251. R833833 (2010)
R833833 (2011)
R833833 (Final)
  • Full-text: Colorado State-Full Text PDF
  • Abstract: Journal of the North American Benthological Society-Abstract
  • Other: ResearchGate-Full Text PDF
  • Supplemental Keywords:

    Niche modeling, hydrologic disturbance, flow regime, river management, river networks, RFA, Scientific Discipline, Air, Ecosystem Protection/Environmental Exposure & Risk, Aquatic Ecosystems & Estuarine Research, Environmental Chemistry, climate change, Air Pollution Effects, Aquatic Ecosystem, Environmental Monitoring, Ecological Risk Assessment, Atmosphere, climatic influence, ecosystem indicators, climate models, aquatic ecosystems, invasive species, coastal ecosystems, global climate models, land and water resources, ecosystem stress, climate variability, Global Climate Change

    Progress and Final Reports:

    Original Abstract
  • 2009 Progress Report
  • 2011 Progress Report
  • 2012 Progress Report
  • Final Report