&EFA
Jmted Sti
Enviroitr1
Agenr'
itection
•
•T,
orkshop Abstracts
•
oint Progress Review for
I.S. EPA STAR Grants:
Regional-Scale Stressor-Response
^^ * *, I •
odels and Consequences of Global
hange for Aauiatic Ecosyste
^*^ '-:•'. '.s\ *rt . . ^
ovember 3-4, 2005
Arlington, VA
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Introduction
Annual Review of EPA STAR Research on:
(1) Regional Scale Stressor Response Models for Environmental Decision-making
(2) Consequences of Global Change for Aquatic Ecosystems: Climate, Land Use, and W Radiation
The mission of the U.S. Environmental Protection Agency (EPA) is to protect, sustain and restore the
health of ecosystems and communities. To support this mission, EPA's Office of Research and Development
(ORD) implements a coordinated research strategy comprised of in-house research conducted at its own
Laboratories and Centers and extramural research sponsored by EPA ORD's Science To Achieve Results
(STAR) Program. STAR awards are made to the academic community and non-profit research organizations
through a highly competitive program of independently peer-reviewed proposals solicited by the ORD's
National Center for Environmental Research (NCER).
This document summarizes research sponsored by two STAR solicitations: (1) Developing Regional-
Scale Stressor-Response Models for Use in Environmental Decision-making, and (2) Assessing the
Consequences of Global Change for Aquatic Ecosystems: Climate, Land Use, and UV Radiation. This research
was funded by the STAR Ecological Research Program1 and the STAR Global Change Research Program2.
The content of this document reflects presentations made by Principal Investigators at the annual progress
review meeting held in Arlington, Virginia, November 3-4, 2005.
The 2002 solicitation for Developing Regional-Scale Stressor-Response Models for Use in Environmental
Decision-making sought proposals for the development of regional scale models that could be used to
investigate, simulate and predict interactions of multiple stressors on the health of aquatic ecosystems. The
research objective was to develop the scientific information needed to facilitate state and local implementation
of integrated ecosystem management practices. This document summarizes a portfolio of 11 separate modeling
studies sponsored under this solicitation. The stressors to ecosystems include non-point sources of nutrients
and sediments, alterations in stream flow due to development or climate variability, invasive species, and
habitat alteration. Modeling techniques include innovative Bayesian statistical models, new approaches to
coupled physical models, and individual-based models. These multiple stressor-response models are being
developed for selected river systems, coastal estuaries, and lakes, and they are being implemented at scales
from site-level to entire watersheds. Although this research is still in progress, in many instances the
investigators have already begun working with resource managers who may ultimately benefit from the
creation of these new modeling tools.
The 2001 solicitation Assessing the Consequences of Global Change for Aquatic Ecosystems: Climate,
Land Use, and UV Radiation sought proposals that contributed to the scientific literature and developed
decision support tools designed to quantitatively and qualitatively relate watershed characteristics to aquatic
ecosystem health and water quality. Principal investigators presented final research results that elucidate: (1)
the impacts of climate change and land use in relation to ultraviolet radiation levels in Rocky Mountain
streams, and the Lehigh and Ontonagon Rivers; and (2) how climate variability, and past, current and future
land use practices influence the ecological structure of vegetation and animal communities relying on playa
lakes in the Southern Plains.
For further background on these topics, go to the following Web sites to read the two Requests for
Applications (RFAs) that resulted in the award of the grants described in this report:
http://es.epa.gov/ncer/rfa/archive/grants/02/02regstressor.html
http://es.epa.gov/ncer/rfa/archive/grants/01/global01.html
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
For more information about the STAR Global Change Program for ecosystems research, contact Bernice
L. Smith at 202-343-9766 (Smith.Bernicel@epa.gov). For information about the STAR Ecological Research
Program, contact Iris Goodman at 202-343-9854 (Goodman.Iris(g),epa.gov). Finally, for more information on
all research areas in EPA's STAR program, please go to http://www.epa.gov/ncer.
The research described in this report has not been subjected to the Agency's required peer review and
policy review, and does not necessarily reflect the views of the Agency. Therefore, no official endorsement
should be inferred. Any opinions, findings, conclusions, or recommendations expressed in this report are not
necessarily those of EPA, but rather those of the investigators who presented their research and other workshop
participants.
The primary focus of EPA ORD's Ecological Research Program is to:
(1) Assess the condition of the nation's ecosystems.
(2) Diagnose the causes of impairments to these ecosystems.
(3) Forecast how ecosystems respond to existing and emerging ecological stressors.
(4) Establish methods to set priorities for protecting and restoring impaired ecosystems.
2
The primary focus of EPA ORD's Global Change Research Program is to improve society's ability to respond and
adapt to future consequences of global change. This entails:
(1) Improving the scientific capabilities and basis for projecting and evaluating effects and vulnerabilities of global
change;
(2) Conducting assessments of the ecological, human health and socioeconomic risks and opportunities presented
by global change; and
(3) Assessing management and adaptation options to improve society's ability to effectively respond to the risks
and opportunities presented by global change.
Assessments of the effects of global change on aquatic ecosystems (freshwater and coastal) and their services in the
context of other stressors and human dimensions is the focus of the Ecosystem Focus Area of EPA's Global Change
Research Program. Linked aquatic-terrestrial ecosystems also are considered to inform the impact of global change
on aquatic ecosystem functioning and services. EPA's Global Change Research Program is consistent with the U.S.
Climate Change Science Program (CCSP) and contributes to the many goals set forth in the CCSP Strategic Plan.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for
Aquatic Ecosystems
Sheraton National Hotel
900 S. Orme Street
Arlington, VA 22204
November 3-4, 2005
AGENDA
This is a Joint Workshop and Progress Review for U.S. EPA STAR Grants on the topics:
(1) Regional-Scale Stressor-Response Models for Aquatic Ecosystems
(2) Effects of Climate, Land Use and UV Radiation on Aquatic Ecosystems
*Posters of PI research will be on display both Thursday and Friday
Thursday, November 3, 2005
7:30 - 8:30 a.m. Registration and Poster Set-Up
8:30 - 9:00 a.m. Welcome and Introduction
Becki Clark, Director Environmental Science Division, U.S. EPA, NCER
Joel Scheraga, Ph.D., National Program Director, EPA Global Change
Research Program, U.S. EPA
Iris Goodman, Ecological Research Program Manager, U.S. EPA, NCER
Session I Theme: Regional-Scale Physical Models for Environmental Decision-Making
9:00 - 9:40 a.m. Development of Coupled Physical and Ecological Models for Stress-Response
Simulations of the Apalachicola Bay
Mark Harwell, Ph.D./Hongquing Wang, Ph.D., Florida A&M University
9:40 - 10:20 a.m. Developing Regional-Scale Stressor Response Models for Managing
Eutrophication in Coastal Marine Ecosystem
Robert Howarth, Ph.D., Cornell University
10:20 - 10:35 a.m. Break (Pis are available to discuss their posters)
10:35 - 11:15 a.m. A Shallow-Water Coastal Habitat Model for Regional-Scale Evaluation of
Management Decisions in the Virginia Province
Charles Gallegos, Ph.D., Smithsonian Institution
11:15-11:55 a.m. Development of a Regional-Scale Model for the Management ofMultiple-
Stressors in the Lake Erie Ecosystem
Joseph Koonce, Ph.D., Case Western Reserve University
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
12:00 noon - 1:15 p.m. Lunch
Session II Theme: Regional-Scale Statistical Models for Environmental Decision-Making
1:15 - 1:55 p.m. Adaptive Implementation Modeling and Monitoring for TMDL Refinement
Kenneth Reckhow, Ph.D., Duke University
1:55 - 2:40 p.m. Bayesian Methods for Regional-Scale Stressor Response Models
Conrad Lamon, Ph.D., Duke University
2:40 - 2:55 p.m. Break (Pis are available to discuss their posters)
2:55 - 3:35 p.m. Developing a Risk Propagation Model for Estimating Ecological Responses of
Streams to Anthropogenic Watershed Stressors and Stream Modification
Vladimir Novotny, Ph.D./Elias Manolakos, Ph.D., Northeastern University
3:35 - 4:15 p.m. Developing Relations Among Human Activities, Stressors, and Stream
Ecosystem Responses and Linkage in Integrated Regional, Multi-Stressor
Models
Jan Stevenson, Ph.D., Michigan State University
Friday, November 4, 2005
8:30 a.m. Introduction
Bernice Smith, Ph.D., Global Change (Ecosystems Research) Program Manager,
U.S. EPA,NCER
Session III Theme: Regional-Scale Population Models for Environmental Decision-Making
8:30 - 9:10 a.m. Effects of Multiple Stressors on Aquatic Communities in the Prairie Pothole
Region
Patrick Schoff, Ph.D., University of Minnesota-Duluth
9:10 - 9:50 a.m. Application of Individual-Based Fish Models to Regional Decision-Making
Roland Lamberson, Ph.D., Humboldt State University
9:50 - 10:20 a.m. Stressor-Response Modeling of the Interactive Effects of Climate Change and
Land Use Patterns on the Alteration of Coastal Marine Systems by Invasive
Species
Robert Whitlatch, Ph.D., University of Connecticut
10:20 - 10:35 a.m. Break (Pis are available to discuss their posters)
Session IV Theme: Climate, Land Use, and UV Radiation on Aquatic Ecosystems
10:35 - 11:05 a.m. Interactive Effects of Climate Change, Wetlands, and Dissolved Organic
Matter on UV Damage to Aquatic Foodwebs
Scott Bridgham, Ph.D., University of Oregon
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
11:05 - 11:45 a.m. The Influence of Climate-Induced Alterations in Dissolved Organic Matter on
Metal Toxicity and UVRadiation in Rocky Mountain Streams
William Clements, Ph.D., Colorado State University
11:45 a.m. - 1:00 p.m. Lunch
1:00 - 1:40 p.m. Interactions Among Climate, Humans, and Play a Wetlands on the Southern
High Plain
Scott McMurry, Ph.D., Texas Tech University
1:40 - 2:20 p.m. Assessing the Interactive Effects of Land Use, Climate, and UV Radiation on
River Ecosystems: Modeling Transparency and the Response of Biota to UVR.
Bruce Hargreaves, Ph.D., Lehigh University
2:20 - 2:30 p.m. Break (Pis are available to discuss their posters)
2:30 - 4:00 p.m. Discussion Session
This informal discussion session provides an opportunity for STAR grantees and
staff of EPA Program Offices, Labs, and Centers to explore opportunities to further
advance the research topics presented. The group will choose the topics to be
discussed; preliminary suggestions include: (1) ways to share data, ideas, or
modeling methods, and (2) ways to synthesize and communicate research results.
To start off the discussion, we will have three speakers give brief remarks (i.e., 5 -
7 minutes):
Rochelle Araujo, Ph.D., U.S. EPA, NERL, ADE; Charles Noss, Ph.D., U.S. EPA,
ORD; and Noha Gaber, Ph.D., U.S. EPA, NCER
4:00-4:10 p.m. Closing Remarks
Bernice Smith, Ph.D., Global Change (Ecosystems Research) Program Manager,
U.S. EPA, NCER
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Regional-Scale Physical Models
for Environmental Decision-Making
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Developing Regional-Scale Stressor-Response Models for Use in
Environmental Decision-Making in Apalachicola Bay
Mark A. Harwell, Ping Hsieh, Wenrui Huang, Elijah Johnson, Katherine Milla,
Hongqing Wang, Kevin Dillon, Glynnis Bugna, and John H. Gentile
Environmental Cooperative Science Center (ECSC),
Florida A&M University (FAMU), Tallahassee, FL
Abstract
Florida's Apalachicola Bay is the "last great bay" that is relatively pristine in the United States. The Bay is
one of the nation's major producers of American oysters (Crassostrea virginicd). It is significantly influenced
by freshwater flows from the Apalachicola, Chattahoochee, and Flint River system (ACF), which drains
approximately 60,000 km2 of Georgia, Alabama, and Florida. Of particular concern are the present and
anticipated reductions below the historical freshwater flows from the ACF system, particularly for urban use in
Atlanta and center-pivot irrigation agriculture in Georgia. The valued ecosystem components (VECs) of the
Apalachicola Bay ecosystem include oysters, recreational fisheries, salt marshes, and associated aesthetic,
endangered, and recreational species of birds, fish, and invertebrates. Stressors include changes in salinity and
turbidity, sea-level rise, nutrient inputs, tropical storms and hurricanes, and habitat alteration. Our objectives
are to develop models that can be used to evaluate the stress-responses of the Bay to these natural and
anthropogenic stressors, and provide scientific support to the environmental decision-making process
following the U.S. Environmental Protection Agency (EPA) ecological risk assessment framework.
We have been developing a coupled physical and ecological model system to accomplish our objectives.
The physical models include a 3-D hydrodynamic model (modified from the Princeton Ocean Model [POM]),
a river flow model (MODBRNCH), and a water quality model (EPA WASP 6) to simulate the current,
transport, salinity, sediment, and nutrient regimes of the Bay. These physical models are being coupled
through a Geographic Information System (GIS) framework to ecological models (i.e., salt marsh, oyster
population, habitat suitability and landscape metrics) to simulate effects of stressors on the VECs.
Hyperspectral remote sensing data, acquired under separate funding from National Oceanic and Atmospheric
Administration (NOAA), provide information for model calibration and stress responses.
To date, we have calibrated and validated a 3-D Apalachicola Bay hydrodynamical model to examine the
effects of freshwater flow, tide, and current on estuarine salinity. We have developed a salt marsh soil salinity
model to examine impacts of climate, tidal forcing, soil, vegetation, and topography on soil salinity
distribution along different elevations in the Atlantic and Gulf of Mexico coastal region. We continued to
acquire spatially explicit databases for the Bay, including weather data, topography, and nutrient and salinity
data for 2002-2004. We are now developing an oyster population model for Apalachicola Bay, and acquiring
extensive existing databases for oysters in the Bay. Meanwhile, technical difficulties have developed in
coupling of the hydrodynamical model with a water quality model, and in calibration of the MODBRNCH
model. As a result, we are exploring alternate river and water quality modules for our coupled model system.
The next steps are to: (1) complete the linkage of the 3D hydrodynamic model to the water quality
module; (2) complete the implementation and calibration of a river flow regime model; (3) continue the
acquisition of data and implementation in the GIS database; (4) complete the development and calibration of
the salt marsh ecological/hydrological model; (5) continue developing and coupling the oyster population
model with hydrodynamic model; and (6) conduct a demonstration ecological risk assessment through
development of the test scenarios related to climate change and ACF water management.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Hongqing Wang, who presented for Dr. Mark Harwell
A One participant asked what unit was used for the suspended sediments concentrations in the model. Dr.
Wang replied that it was milligrams/liter. These measurements will be checked against real data to
determine if they are correct. The participant commented that he was surprised to see sea grasses in the
system with the depth of suspended sediments and asked if there were water clarity measurements. Dr.
Wang responded that there was monitoring for a number of water quality parameters.
A One participant stated that he was interested in the use of the Princeton Model and the application to the
bay and asked if the project looked at how the grid size of the model affected the predictions of tidal flow.
Dr. Wang responded that he was not the hydrologist on the project, but the grid size they are using should
be suitable with some variation.
A Iris Goodman asked if Dr. Wang has any sense of the relative contribution of tidal and bay influences
versus the water management in the uplands draining into the bay. Dr. Wang responded that if the river
location changes, it will affect the salinity and oyster population in the bay.
A One participant stated that the bay has a strong west to east salinity gradient and 90 percent of the fresh
water is coming from the Apalachicola River, and asked why there is low salinity in the western part of the
bay? Dr. Wang responded that most of the ocean water is coming from the eastern part and the ocean
current is east to west. The fresh water comes from the river mouth and reduces the salinity.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Developing Regional-Scale Stressor Response Models for Managing
Eutrophication in Coastal Marine Ecosystem
Robert Howarth
Cornell University, Ithaca, NY
Abstract
Our goals are to: (1) develop a regional-scale model for analyzing nutrient inputs to coastal ecosystems;
(2) develop a model-based classification scheme for the comparative analysis of the sensitivity of coastal
ecosystems to these nutrient inputs; and (3) develop quantitative approaches for evaluating how other stressors
such as climate change, land-use change, and sediment fluxes interact with nutrient inputs to affect coastal
ecosystems.
The project has two interacting parts. First, we are developing and testing the Regional Nutrient
Management Model (ReNuMa), a model designed to be used by managers to evaluate sources of nutrient and
sediment fluxes from regions and large watersheds to coastal marine ecosystems, and to be responsive to
watershed management practices. We are refining and modifying this model to increase its effectiveness as a
tool to investigate the interacting effects of climate variability, climate change, and land-use change on fluxes
of water and nutrients from regions and watersheds. Second, we have developed a nutrient-phytoplankton-
zooplankton (NPZ) model of estuarine response to hydrologic forcing and nitrogen loading. We are using this
model with available data sets (e.g., National Oceanic and Atmospheric Administration and Land-Ocean
Interactions in the Coastal Zone) on physical and ecological aspects of coastal marine ecosystems towards
evaluating the sensitivity of coastal marine ecosystems to nutrient enrichment, and to predict how climate
change and other stressors interact with this sensitivity.
Progress to date includes: (1) additional analysis of interactions between hydrology and nitrogen fluxes in
16 Northeastern U.S. watersheds; and (2) development and preliminary testing of a watershed-scale regional
nutrient management model (ReNuMa) and the NPZ model of estuarine response to nitrogen loading, which
we are using to develop a process-based estuarine classification scheme. Analysis of the relationship between
climate and Net Anthropogenic Nitrogen Inputs (NANI) to Northeastern U.S. watersheds suggests that 75-80
percent of nitrogen inputs are retained or denitrified in the landscape; those watersheds with higher
precipitation or discharge export a greater fraction of N inputs in streamflow, possibly due to reduced
denitrification. The ReNuMa model shows good agreement between observations and annual simulations of
both streamflow and dissolved inorganic nitrogen at the regional scale, though better parameterizations may be
required for individual watersheds. The NPZ model suggests that differences between estuarine responses to
point source and non-point sources of nitrogen may result due to the differences in coupling with hydrology.
Nutrients are now the largest pollution problem in the United States. This set of tools will allow
environmental managers to set priorities for targets in nutrient reduction, by source of nutrient, and among
multiple watersheds in the context of relative benefit to be achieved in coastal water quality. It also will allow
managers to explore scenarios for how land-use change and climate change may interact with plans for
reducing nutrient pollution. This project will fulfill two high priority recommendations of the National
Research Council (NRC) (2000) report on coastal nutrient pollution.
The major objectives for the next year include: (1) further development of the watershed model and
preliminary testing in the Upper Susquehanna and other Northeastern U.S. watersheds; (2) further development
and analysis of the dynamics of the NPZ estuarine response model and comparison with available estuarine
datasets; and (3) coupling of the terrestrial and estuarine response models.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Robert Howarth
A One participant asked why the coastal plain area was not shown on the graph showing the distribution of
the 16 watersheds. Dr. Howarth responded they are using the U.S. Geological Survey (USGS) monitoring
data for the river flux estimates and there are no gauging stations in that area. There are no USGS water
quality data further downstream in any of the watersheds closer to the coast. The water that is upstream of
the gauging stations is relatively rural. The highly industrialized, heavily populated area, which also is
contributing to nitrogen to the coast, is not included in the analysis. The participant asked if correction
factors were added. Dr. Howarth responded that in the statistical mass balance approach they are only
dealing with the watersheds. He is fairly confident that they can understand, at least in a temperate zone,
what is going on with human alteration of nitrogen cycling in relatively rural areas. He is not sure that
when they get into hilly, urbanized areas that the parameters are valid. They have a separate effort funded
by the Hudson River Foundation where they are trying to do mass balances for the lower Hudson River
estuary in the New York City Harbor area. They are using the City of New York water quality data and
plotting nutrient concentrations versus salinity to determine statistically nutrient input sources versus the
city. That will give them at least one urban area in the next few years.
A One participant asked if Dr. Howarth had addressed the role of hydrologic variability in terms of whether
the wetness of the watersheds is really affecting the amount of nitrogen deposition. Dr. Howarth responded
that they are just starting to work on it. They have looked at long-term averages for sinks. There also is a
short-term effect of short-term storage and loss, and they will be looking at the historical data set for this
data. For the Northeast watersheds, they have total nitrogen fluxes on an annual basis from the 1980s to
the present. They also have nitrate data for a subset of about seven or eight of the watersheds, some of
which go back to 1919.
A One participant asked if they expected more contributions from climate change on nitrogen flux to rivers
in the watershed level compared to human inputs such as fertilizer use. Dr. Howarth responded that
climate change affects the nitrogen deposition. On average, most of the nitrogen is not reaching the coast;
it is going into the landscape where it is denitrified, and he thinks that it is climatically sensitive. The
problem will be aggravated where climate change leads to wetter environments. There are places where
climate change will not make it wetter. Most of the model projections say that the Mississippi River Basin
will be dryer.
A One participant commented that there is much data on nitrification. He asked why there is variability that
is not understood in nitrification and what the knowledge gap was? Dr. Howarth responded that there is a
tremendous amount of data measuring denitrification in well-ends and in low-order streams. There are hot
spots for denitrification. When you try to scale it beyond the plot scale, there are problems. The issue is of
the water resonance time and whether the water that is draining off the landscape is really going through or
under wetlands, or if it goes through the wetlands so quickly that it does not really process for
denitrification. It is very hard to scale to an entire catchment basin to show how effective the wetlands are
as a sink.
A Iris Goodman commented that EPA deals with the problem of how to best combine insights gained from
monitoring and insights gained from modeling. Certainly, both have their limitations and there is no right
answer.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
A Shallow-Water Coastal Habitat Model for Regional Scale Evaluation
of Management Decisions in the Chesapeake Region
Charles L. Gallegos, Donald E. Wetter, Thomas E. Jordan, Patrick J. Neale, and J. Patrick Megonigal
Smithsonian Environmental Research Center, Edgewater, MD
Abstract
Management decisions to protect estuaries are being made in the context of unprecedented environmental
changes including rising concentration of atmospheric C02, increased ultraviolet (UV) radiation, especially the
damaging UV-B, and changes in land use patterns. Interactions between altered rainfall regimes and changes in
land use patterns will have consequences for the delivery of sediments and nutrients to estuaries. Projecting the
effectiveness of management actions must proceed on the basis of predictions from mathematical models,
because experimental manipulations cannot be made on the relevant scales.
Our modeling efforts focus on shallow tributary embayments and small tidal creeks of Chesapeake Bay,
because the ecological importance of shallow systems far exceeds their volumetric contribution to the bay.
The end points for our model will be those indicators being used as delisting criteria for Chesapeake Bay,
namely, chlorophyll, water clarity (diffuse attenuation coefficient), and dissolved oxygen.
Progress has been made on development of a three-compartment subestuary model. The model domain
segments with different volumetric dimension, lying in between boundaries at the mouth and at the head of the
subestuary. The watershed is considered to be an upstream boundary where multiple stressors, such as
inorganic nutrients, sediments, and dissolved organic matter, are discharged. The downstream boundary is
considered to be the Chesapeake Bay, or a major tributary, such as the Potomac or James River. Each
subestuary segment has multiple ecological components (26 state variables), including nutrients, size-
fractionated phytoplankton (3 size classes) and zooplankton (3 size classes), and detritus. Nitrogen and
phosphorus are tracked separately to allow for spatial or temporal changes in the limiting nutrient (double-
currency system), and light propagation through water column is computed based on empirical bio-optical
algorithms.
One project objective is to analyze how the geographic variability in physical structure and human use
among the linked watershed-subestuary systems of Chesapeake Bay affects estuarine responses to multiple
stressors. The descriptive portion of this task has been completed. We have identified 128 Chesapeake Bay
systems that fit our local watershed-subestuary paradigm and have captured the boundaries of these systems in
a Geographic Information System (GIS) database. Subestuary areas range from 0.1-101 km and their
associated local watershed areas range from 6-1664 km2, with the National Land Cover Database land cover
percentages ranging from 6-81 percent forest, 1-64 percent cropland, 2-38 percent grassland, and 0.3-89
percent developed land. We also analyzed digital shoreline and bathymetric data to calculate a number of
subestuary metrics, including subestuary area and water volume, mouth width and area of vertical profile,
proportion of shallow water ([ 2 m) area, elongation ratio, fractal dimension, the ratio of subestuary perimeter
to subestuary area, and ratio of local watershed area to subestuary volume.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Charles Gallegos
A One participant asked if there was any progress in measuring water clarity or transparency. Dr. Gallegos
responded that there was a lot of progress in terms of the basic frame of the model. The model will be able
to do a good job of predicting the diffuse attenuation coefficient from CDOM, particulate matter, and
chlorophyll.
A One participant asked how they were handling the nitrogen versus phosphorus limitation. Dr. Gallegos
responded that it will be from concentrations in the delivery water plus what is coming out of the
sediments in relation to phytoplankton needs and demands. For phytoplankton, they are running an
internal pool type of model. It is one of the things that will be examined in the generalized sensitivity
analysis to determine if that level of detail really is needed to predict the limiting nutrient. It is being
handled primarily by the sources.
A One participant asked how sensitive the model is to including the different size classes. Dr. Gallegos
responded that he did not know how sensitive the model was to the different size structures. He wants to
start out with something detailed and see how far down he can simplify it. One of the things that might
survive is the effect on water clarity, because that is one place where size makes a difference.
A One participant asked whether taxonomic composition, which may play a major role in the potential for
sinks, will be addressed in the model. Dr. Gallegos responded that if they handle it at all, it will be very
parameterized in terms of sinking rates versus time from what they know about the changes in the
taxonomic composition. He does not think that attempts to model that explicitly have been very successful.
A Iris Goodman commented that 18 metrics have been developed on the 128 systems. She asked if these
metrics would be sufficient to parameterize the model in all the needed ways or would they need to add
more parameters. She was interested in the notion of how they are working their way from these
aggregated metrics to making these predictions spatially at all of these small estuaries. Dr. Gallegos
responded that they will examine whatever data they can obtain. There is a lot of data for the Road River.
About 30 of these systems have Chesapeake Bay program stations, and they should be able to obtain some
aggregate measures. There also are a growing number of these systems that have Maryland Department of
Natural Resources Eyes on the Bay stations that have probes recording at 15-minute intervals. They will
be using data from these types of stations to develop relationships between exchange and other metrics.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Development of a Regional-Scale Model for the Management of
Multiplestressors in the Lake Erie Ecosystem
Joseph F. Koonce and Benjamin F. Hobbs
Case Western Reserve University, Cleveland, OH; Johns Hopkins University, Baltimore, MD
Abstract
The objective of this research is to develop a regional-scale, stressor-response model for the management
of the Lake Erie ecosystem. Stressors addressed include effects of land use changes and Total Maximum Daily
Load (TMDL) targets for nutrients, habitat alteration, and natural flow regime modification at the scale of
individual watersheds, coupled with whole lake ecosystem effects of invasion of exotic species and fisheries
exploitation. Model predictions focus on effects of stressors on production and abundance of Lake Erie fish
populations as indicators of the health of the Lake Erie ecosystem and will be incorporated into a multi-
objective decision making tool for use by Lake Erie water quality and fisheries managers along with other
resource planners. The research approach involves joining multi-level modeling with multi-objective risk
decision tools. The research plan focuses on: (1) linking changes in watershed habitat and nutrient loading
regimes proposed for the TMDL process to Lake Erie ecosystem health; (2) quantifying uncertainties in model
predictions and determining the effects of uncertainties on management decisions; (3) evaluating interaction of
stressors, particularly focusing on cross-scale additivity of stressors; (4) developing tools to evaluate ecological
risk of land-use changes in watersheds of the Lake Erie ecosystem; and (5) identifying and evaluating critical
break-points in ecosystem integrity of the Lake Erie ecosystem and of its integrated management. Highlights
of research of the second year of the project (June 1, 2004 to May 31, 2005) included:
A Completion of field work and testing of models for establishing a habitat supply inventory for the entire
Lake Erie watershed. Products include one MS Thesis, two manuscripts submitted for publication, and one
manuscript in preparation. Work also has involved developing both ESRI and open source
implementations of Geographic Information System (GIS) data analysis layers for use with public groups
and managers.
A Using models (IHACRES and SWAT) parameterized in task 1.2, work has continued to use models to
develop a functional (hydraulic transport) representation of land cover effects on flow and nutrient
loading. The product of this work is currently an M.S. Thesis, which is nearing completion.
A Assembly of a component-based DEVS modeling and simulation framework to perform cross-scale
analysis of the interaction of stressors. The products are a new software package and a number of
presentations. Work is coordinated with the Everglades ATLAS model. Products include a software
library, a modeling framework that incorporates a DEVS modeling and simulation platform, and model
repository for assembly and execution of hierarchical models.
A Development and testing of a decision analysis framework to explore the tradeoffs associated with dam
removal from selected tributaries in the Lake Erie ecosystem.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Joseph Koonce
A One participant commented that the model is incredibly detailed and that it seems that the uncertainty
analysis will be a daunting task. The participant asked if Dr. Koonce anticipated fully characterizing the
error terms in the model parameters and the model specifications. Dr. Koonce responded that there is no
question that there is pattern in the parameter space of these complicated models. The challenge is to have
a framework in place so that they can easily generate the quantities of data that they need to characterize
the patterns so they can begin to understand how the uncertainties actually work. A variety of parameter
combinations give the same output predictions. There are other computational tools becoming available to
do functional analysis. They are hoping to apply them in the parameter space. There are some very nice
algorithms to begin to look for these patterns. In some of the preliminary work they have done in
disaggregating fish population into age groups and into cohorts, they know that there is structure that there
is structure in the parameter space.
A One participant commented that they have been trying to develop good models for predicting phosphorus
loading, and asked what tools they are using and how well they work. Dr. Koonce responded that the
International Joint Commission is collecting monitoring information on loadings. Unfortunately, with the
changes in the Great Lakes water quality agreement in 1988, some of the monitoring schemes begin to
break down. Dave Dolan of the University of Wisconsin at Green Bay is continuing to maintain the
databases on river flows, sewage treatment, and effluents so that they are tracking the loading. The
problem with the loading calculation is that all phosphorus in not equal. It may be the case that with the
reduction in the sewage treatment plants of detergent phosphorus, there are more resistant forms of
particulate phosphorus as the primary component of the loading. There is not a lot of good information on
the composition of the loading. There are fairly good estimates of the seasonality and loading. In the Lake
Erie ecosystem, there are four dominant tributaries in the connecting channel that makes it easier to
investigate.
A Iris Goodman asked which break points in the ecosystem functions they were focusing on that might have
the best results. Dr. Koonce responded that almost all of the break points that they know exist in the
system have to do with the fish community composition. When predator-prey abundance tends to achieve
certain threshold levels, the chance events that come with the strong year class of walleye, for example,
can send the system into a very long-term predator-prey ossolation. This happened in 1988. The fisheries
preference is to have a lot of large body predators in the systems. The management always is tending to
put the system at risk.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Regional-Scale Statistical Models
for Environmental Decision-Making
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Adaptive Implementation Modeling and Monitoring
forTMDL Refinement
Kenneth H. Reckhow
Duke University, Durham, NC
Abstract
The primary objectives of this project are to: (1) develop an adaptive implementation modeling and
monitoring strategy (AIMMS) for Total Maximum Daily Load (TMDL) improvement; and (2) apply and
evaluate AIMMS on the Neuse Estuary nitrogen TMDL in North Carolina. For this study, the models in
AIMMS are the NeuBERN Bayes network estuary model linked with a Bayesian version of the U.S.
Geological Survey (USGS) Neuse SPARROW model; AIMMS allows us to analytically integrate TMDL
modeling with post-implementation monitoring to refine and improve the TMDL over time.
After the completion of the Bayesian analysis of SPARROW, we focused on two fronts. First, we started a
series of studies on the use of a simple Bayesian approach for updating nutrient loadings and nutrient
concentrations. Second, we started to use the Neuse River Digital Watershed (from the NSF-CUAHSI
program) to provide annual nutrient loading data and revise our Bayesian SPARROW model to update model
parameters as well as estimate nitrogen loadings when new data are available. In this report, we describe
progress made in these two areas.
Information synthesis is usually the motivation for employing Bayesian analysis; thus, Bayesian analysis
serves as perhaps the ideal approach for the analytics of adaptive implementation. The conventional
application of a Bayesian approach emphasizes the combination of prior information (in this case, from the
TMDL forecast model) and a single set of data (post-implementation monitoring data). It is shown, however,
in Bayesian statistics texts that sequential updating using the posterior from the previous step as prior is
equivalent to updating using all of the data together; thus, sequential updating provides a means to investigate
possible temporal patterns in the data, which is attractive for adaptive implementation of a TMDL.
As an example of sequential Bayesian updating for adaptive implementation of a TMDL, a series of
computer programs were developed to automate the process of updating water quality concentration estimation
from model predictions and subsequent monitoring data. These programs use Bayesian analysis results for
(log) normal random variables, and the conjugate family of prior distributions. The process has three steps.
First, a number of procedures were developed for converting TMDL model forecasts of water quality
concentrations to a prior distribution of the underlying concentration distribution parameters. Second, a
program was developed to produce the posterior distribution of the underlying concentration distribution
parameters and the posterior predictive distribution of future observations, based on the pre-TMDL model
forecast (the "prior") and the first year of post-implementation monitoring data (the "sample"). Third, the
"posterior" distribution of the underlying concentration distribution parameters is then converted to a prior
distribution of the same parameters for the next time period, and the process repeats when new data are
available.
To demonstrate this process, a Bayesian SPARROW model-predicted 1992 nitrogen concentration
distribution for the Neuse River Estuary was used to develop a prior distribution of the mean and variance of
log nitrogen concentrations, and the sequentially updated posterior predictive distributions for each subsequent
year. The same process was repeated for the chlorophyll a concentration distribution in the Neuse River
Estuary. The prior distribution for chlorophyll-a was developed using an empirical model (NeuBERN) and the
results from the SPARROW model. Although the prior distribution based on NeuBERN overestimated the
chlorophyll a concentration, the sequentially updated posterior predictive distributions (based on post-
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
implementation measurements) quickly converged to a distribution similar to the observed chlorophyll-a
concentration data.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Kenneth Reckhow
A One participant commented that he thought that much of the Neuse was phosphorus limited. He asked if
that would explain the poor prediction for chlorophyll-a. Dr. Reckhow responded that he agreed. The
challenge is how to conclude definitively that a water body is limited by a particular nutrient. They will
add a phosphorus component to the model.
A One participant asked what approach would be used to reparameterize SPARROW to handle
subcatchments. Dr. Rechhow responded that they will use approaches that allow them to estimate
parameter sets that will capture the covariants. They will not be identifying individual parameters.
A One participant commented that that you are exceeding expectations when you have very low chlorophyll
levels. From his recent experience, there is considerable discussion among stakeholders to establish a
lower criterion. Dr. Reckhow responded that the results he presented are not widely known; people have
not recognized how much things have improved in the Neuse. One of the recommendations in the National
Academy of Sciences study of the TMDL program was for the states to take the triannual review of
standards more seriously.
A Iris Goodman commented that it is a challenge for everyone to develop ways to communicate to people in
the fastest, most expeditious way to get to the answers they are seeking. Dr. Reckhow responded that you
should try to listen to both the monitoring and modeling people, but ultimately the monitoring data may
begin to dominate.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Bayesian Methods for Regional-Scale Stressor Response Models
Conrad Lamon
Duke University, Durham, NC
Abstract
We developed regional-scale eutrophication models for lakes, ponds and reservoirs to investigate the link
between nutrients and chlorophyll-a. The Bayesian TREED (BTREED) model approach allows association of
multiple environmental stressors with a biological response with simple linear models, while identifying
subsets of the data in which the models fit well. Nutrient data for lakes and ponds (water body type = 4) and
reservoirs (water body type = 5) across the United States were obtained from the U.S. Environmental
Protection Agency (EPA) National Nutrient Criteria Database. The nutrient data consist of measurements for
both stressor variables (such as total nitrogen and total phosphorus), and a response variable (chlorophyll-a),
used in the BTREED model. Variables used in subsetting include ecoregion, water body type, month and
categorical variables representing the nitrogen and chlorophyll analytical methods used. Markov chain Monte
Carlo (MCMC) posterior exploration is used to guide a stochastic search through a rich suite of candidate tree
structures to obtain models that better fit the data. The Bayes factor provides a goodness of fit criterion for
comparison of resultant models. We randomly split the data into training and test sets; the training data were
used in model estimation, and the test data were used to evaluate out of sample predictive performance of the
model. An average relative efficiency of 1.02 between the training and test data for the four best fitting
(highest log-likelihood) models suggests good performance in out of sample predictive efficiency relative to
other eutrophication models. We found that the 98,169-observation dataset produced large, complex trees,
making their structure difficult to interpret. Adjustments may be made to priors to control the size/complexity
of resulting TREED models, improving interpretability to some extent. Some increased complexity, however,
stems from the binary tree structure itself. We therefore intend to address these problems on two fronts. First,
we are working to incorporate more predictor variables into our dataset for use in the end node models, in the
hope of reducing the need to explain variability in the chlorophyll-a response with added complexity in the
tree structure. Second, we are in the process of shifting our modeling framework to a more purely Bayesian
Hierarchical one, in which the binary tree structure is absent. Knowledge of the tree structure identified using
the BTREED approach will provide valuable information for use in determination of the hierarchical structure
for use in the new framework.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. E. Conrad Lamon
A One participant commented that he had chaired an NRC committee on coastal eutrophication. They were
asked by Congress and EPA's Office of Water to comment on nutrient criteria development for coastal
waters. The committee had a strong consensus that concentration data were not helpful and that the criteria
should be loadings to estuaries. The response of estuaries is not well correlated with standing
concentrations. Total nitrogen and total phosphorus have a lot nonbiologically active pulls (e.g.,
phosphorus is stored in all sorts of inorganic forms). What managers are managing are the inputs, not the
concentrations. He asked if Dr. Lamon would consider looking at load as one of the variables as opposed
to concentrations for their lake analysis, and wondered what the advantage is to be able to predict
chlorophyll from either total nitrogen or total phosphorus in the data set. Dr. Lamon responded that you
need to know the relationship between nutrients and chlorophyll to reduce chlorophyll. At an annual
average scale, you may get better results with load. They are using the nutrient criteria database and they
do not have load information. They could go to the upstream river locations of each lake and find USGS
data to map and calculate loads, which may be easier and better than using SPARROW estimates.
A One participant commented that he just completed a project where the task was to identify the commonly
measured criteria that are most predictive of the designated use. He consistently found that chlorophyll
was the best predictor of designated use, which suggests that what Dr. Lamon is doing is appropriate. He
also stated that they commented on the National Academy of Sciences study on TMDLs as to where the
criterion should be placed. If you stop with load, there is a good deal of hidden uncertainty between load
and designated use that is missing from the prediction. You move from action, load, concentration, and
then chlorophyll or some biological response in the designated use. If designated use is the driving factor
in the objective, their structural equation modeling strongly suggested that chlorophyll was the appropriate
predictor for eutrophication related impacts.
A Iris Goodman commented that Dr. Lamon's analysis showed the strong importance of method in
controlling the trees. She asked if Dr. Lamon would prepare a 500-word essay that could be distributed to
EPA Program Offices. Dr. Lamon agreed to prepare the essay. She also commented that in looking at
some CART techniques that are not Bayesian based, one of the important aspects of those methods for
users is to be able to do interpretation on the tree structure itself. She urged the group in their
collaborations to begin looking at additional kinds of predictor variables that ultimately are useful for
managers to gain a better understanding of the issues that they need to address. She offered to provide
contacts from the landscape ecology grants who might have suggestions on important predictor variables
that are ultimately useful for interpreting the tree structure.
A One participant asked if the BTREEDs were inverted multiple regressions rather than reverse ANOVA. In
nutrient criteria development, one of the issues is how rigid the breaks are between 2.6 m and 16.3 m
depth. He asked if one would expect a different relationship between loading, concentration, and endpoints
like chlorophyll-ess or biological diversity to break at those specific levels, or if it is just a continuous
gradient between nutrient concentrations and chlorophyll and it happens to be breaking at those points. Dr.
Lamon responded that the breakpoints were not absolute or precise. This is an exploratory type method to
indicate the important structure in the data that can be further developed with a fully Bayesian hierarchical
approach in which a distribution on the split point could be obtained. The method does identify the change
points fairly well. The participant asked if Dr. Lamon had classified the systems based on size but not on
DOC. Dr. Lamon responded that there were fewer DOC than nitrogen or chlorophyll measurements to go
with the phosphorus. The participant commented that the first step is to classify what you predict to find in
the system so natural features are used to classify expected conditions. The next step is to look at how the
systems vary along a phosphorus gradient (e.g., how a valued attribute such as chlorophyll changes along a
phosphorus gradient). Dr. Lamon responded that it is classifying based on the change points and then the
continuous is in the regression model.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
One participant asked how the model could be used to predict the effect of environment changes, such as
climate change. Dr. Lamon responded the model could be used to the extent that climate changes cause
changes in nitrogen or phosphorus concentrations in a lake. Temperature, however, is not included in the
model. If temperature were included for climate change predictions, it would be outside of the range for
past observation, which could be a problem with regression-based methods.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Developing a Risk Propagation Model for Estimating Ecological
Responses of Streams to Anthropogenic Watershed Stresses
and Stream Modifications
Vladimir Novotny1, Elias Manolakos1, Timothy Ehlinger2, Alena Bartosovd2,
Neal O'Reilly3,Ramanitharan Kandiah1, and Ferdinand Hellweger1
1 2
Northeastern University, Boston, MA; University of Wisconsin, Milwaukee, Wl;
3Illinois State Water Survey (University of Illinois), Champaign, IL
Abstract
The goal of this research is the development of a regionalized watershed-scale model to determine aquatic
ecosystem vulnerability to anthropogenic watershed changes, pollutant loads, and stream modifications (such
as impoundments and riverine navigation).
The tasks completed are:
A Development of the database software shell.
A Development of a watershed loading Geographic Information System (GIS) based model.
A Development of Neural Net algorithms for modeling input/output (cause/effect) response of the ecological
system (supervised ANN learning).
A Development of Neural Net unsupervised Self-Organizing Maps (SOM), followed by Canonical
Correspondence Analysis derived (mined) from two large databases.
A Analysis and quantifying the risks expressed as Maximum Species Richness relationships for major
stresses.
The workshop presentation features an application of unsupervised Artificial Neural Networks (ANN)
yielding SOM. Indices of Biotic Integrity (IBI) and its metrics for fish were analyzed by ANN, followed by
Canonical Correspondence Analysis (CCA) to extract (mine) knowledge on the impact of watershed and water
body stresses on biotic integrity of receiving waters. Two large databases of measurements of about 50
parameters, including fish numbers from 2000 sites in Maryland, and about the same number of sites and a
greater number of parameters from Ohio were used for the analysis, development and testing of the
methodology. Self-Organizing Mapping by ANN discovered nonlinear clustering of the fish metrics of the
IBIs that was then overlaid with clustering of environmental variables to find correlation between the metrics,
their clusters and individual impact parameters. CCA then quantitatively identified the parameters of the
greatest importance, the Cluster Dominating Parameters. This analysis has wide spread ecological implications
and enables quantitative assessment of the impact of multiple stresses on IBIs and similar multimetric biotic
indices.
Using the information on the top Cluster Dominating Parameters, an input-output ANN model was
developed by supervised learning using the Ohio database. The model was very accurate for data used in
calibration (randomly selected 75% of data, R = 0.99) and adequately accurate for verification with the
remaining 25 percent of data (R= 0.76).
The team members have created a Web site where all reports and other publications or their abstracts will
be available: http://www.coe.neu.edu/environment.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Vladimir Novotny
A One participant asked how much the SOM was picking up the natural variabilities that the State of Ohio
uses in their classification of expectant conditions in the weather to the streams. Dr. Novotny responded
that it has something to do with the ecoregions. The northwest is part of the Lake Erie ecoregion. There
was some relationship to ecoregions. Dr. Manolakos commented that they are training the SOM based on
the metric scores as recorded in the database. They assume that the metric scores take into account the
location of the sites.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Regional-Scale Population Models
for Environmental Decision-Making
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Effects of Multiple Stressors on Aquatic Communities
in the Prairie Pothole Region
Patrick K. Schofj, Lucinda B. Johnson , Glenn G. Guntenspergen , and W. Carter Johnson
University of Minnesota, Duluth, MN; U.S. Geological Survey, Patuxent, MD;
3South Dakota State University, Brookings, SD
Abstract
The Prairie Pothole Region (PPR), forming the northeastern edge of the Great Plains, encompasses a large
area of diverse wetlands that represent crucial aquatic resources for flood control and aquatic and terrestrial
production. Because most of the PPR is subject to mixed use agriculture and open grazing, the wetlands are
routinely exposed to a variety of anthropogenic stressors, such as pesticides, nutrients, and domestic animal
pathogens. In addition, higher latitudes are exposed to increased ultraviolet B radiation (UV-B). Climate
models indicate that the PPR is likely to be severely impacted by climate change through increasing
temperature and reduced precipitation.
The overall goals of the project are to: (1) quantify the relationships among factors directly affected by
climate change (e.g., hydroperiod), differing land use, and amphibian community structure and composition in
the prairie pothole region; (2) quantify the relationships among physical and chemical wetland attributes (e.g.,
hydroperiod, thermal regime, pH, and DOC), UV-B radiation, and land use (including associated pesticide use)
on amphibian organismal and community responses; (3) quantify the effects of multiple stressors (shortened
hydroperiod, increased UV-B radiation, and atrazine exposure) on the health and organismal responses of R.
pipiens; and (4) use regional climate scenarios and hydrologic models in conjunction with empirical data
gathered through field and mesocosm studies to predict potential effects of multiple stressors on prairie pothole
wetlands and their associated amphibian communities.
Data on stressor impacts were analyzed at three spatial scales: landscape (67 sites), wetland (35 sites), and
mesocosm. Sites within the landscape and wetland scale studies were designated as seasonal (SS) or semi-
permanent (SP) and were classified further as crop or grassland land use. Landscape-scale (level I) study sites
are distributed across the U.S. portion of the PPR; wetland-scale (level II) study sites are concentrated in east-
central South Dakota. Mesocosm studies (level III) were conducted at the Oak Lake Field Station of South
Dakota State University.
Eleven amphibian species were observed in total, and species richness per wetland ranged from zero to
five. No species richness differences were observed between row crop and grassland wetlands, however, more
species were observed in Central Tall and Northern Tall Grassland Ecoregion wetlands than in Prairie Coteau,
Northern Short and Northern Mixed Grassland Ecoregion wetlands. Northern Leopard Frog (Rana pipiens)
was represented across the entire PPR, with adults in 44 sites breeding evidence in 29 sites. Logistic regression
indicated probability of R. pipiens presence was significantly influenced by the interaction of hydrology and
land cover (p = 0.02), while R. pipiens breeding in a wetland was significantly influenced by hydrology, but
not by landcover or an interaction of the two treatments (p = 0.08).
Fewer of the wetland scale sites produced numbers of metamorphic R. pipiens necessary for malformation
surveys in 2004 than in 2003. Seven of 13 sites (54%) surveyed produced malformed frogs. In these sites, a
total of 3.5% (26/748) displayed at least one identifiable malformation. As in 2003, hindlimb malformations
constituted a majority (71.4%) of the total. Malformation prevalence was not significantly correlated with
surrounding land use or atrazine concentration. Metamorphic frogs were collected from wetlands representing
a range of land-use types for analysis of gonadal dysmorphogenesis.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Based on a repeated measures ANOVA including both intensive and extensive sites, hydrology and land
cover were significant predictors of several water quality parameters. Seasonal wetlands were significantly
more alkaline than semi-permanent wetlands, with seasonal crop wetlands significantly more alkaline than all
other treatments (p < 0.001). Land cover also influenced specific conductivity (uS) and DOC (ppm), both of
which were significantly higher in wetlands surrounded by grassland than row crop (specific conductivity, p =
0.056 and DOC, p < 0.0001; intensive sites only). As expected, hydrologic regime influenced maximum depth
and water temperature (both day and nighttime); semi-permanent wetlands were deeper and colder than
seasonal wetlands.
UV-B radiation surface levels and attenuation rates (through the water column) were measured in 20 sites
in 2003 and in 34 sites in 2004. UV-B was rapidly attenuated through the water column [Kd (attenuation rate) =
6.24 - 58.93 m"1]. Preliminary analysis indicates that DOC and color, which are the dominant factors in UV-B
attenuation in lakes, may not be the primary influences on UV-B attenuation in PPR wetlands.
Triazine (atrazine and simizine) concentrations assessed in water collected in mid-April, mid-May, and
late June 2004, ranged from non-detectable (< 0.01 ppb) to 7.124 ppb, with concentrations jj, 0.01 ppb present
in 71 percent of wetlands sampled in Survey 1, and in 100 percent of wetlands in Surveys 2 and 3. Triazine
concentrations were higher in wetlands where land use within a 90 m buffer was classified as > 70 percent
crop than in those where grassland comprised the greatest proportion of the buffer ([ 7.124 ppb vs. [ 0.657
ppb, respectively). Semi-permanent wetlands with corn in the 90 m buffer had higher mean atrazine
concentrations, and corn presence was the best overall predictor of wetland atrazine concentration for Surveys
2 and 3 (ANOVA, p = < 0.001-0.025). However, neither land use (crop vs. grassland; ANOVA, p = 0.662) nor
hydrologic regime (ANOVA, p = 0.878) were significant predictors of maximum atrazine concentration in
April-July 2004.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Patrick Schoff
A One participant asked how far away the grasslands wetlands are from where the atrazine is used, and what
the distance is over which the drift would be occurring. Dr. Schoff responded that it is a large drift and it is
reasonable. It always is windy in the Prairie Pothole Region. One of the challenges in the 2003 season was
to conduct UV attenuation measurements because the water surface has to be calm. It would be calm
before 10 a.m. and after 5 p.m. The original presumption of the study was that atrazine would be run-off
from agricultural fields. In this area, it does not appear to be happening.
A One participant asked why the DOC was not related to UV attenuation, and whether it is because the UV
attenuation varied but not the DOC concentration. Dr. Schoff replied that the statement was correct. The
participant commented that in areas with a lot of sun and long residence time, there could much
photobleaching, which could change the transparency. Dr. Schoff responded that there is a lot of wind-
driven mixing that takes place.
A One participant asked how the atrazine is applied. Dr. Schoff replied that much of the atrazine was applied
by air. The participant asked if the sites were small or big fields and whether the grassland and crop land
potholes were next to each other. Dr. Schoff responded that they attempted to locate crop sites that were
embedded within a cornfield and were able to find about 50 percent of the sites within the vicinity of a
cornfield. In North Dakota and South Dakota, most of the corn is grown for forage; therefore, the fields are
smaller and more interspersed. In Iowa and Minnesota, they had a harder time finding crop sites.
A One participant asked if atrazine could be moving through the groundwater. Dr. Schoff responded that it
could. The participant asked if GIS and distances from crop fields could be used to look at potential
controlling factors. Dr. Schoff responded that a long-term study of the hydrology of the potholes has been
set up in the Prairie Coteau ecoregion. He commented that their study is being done under a period of
drought. Some areas are persistently wet and some are persistently dry. He suggested that EPA may wish
to consider establishing a long-term study program to long at a 10-year time scale and capture some of the
natural variations.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Developing Relations Among Human Activities, Stressors,
and Stream Ecosystem Responses and Linkage in Integrated
Regional, Multi-Stressor Models
R. J. Stevenson1, M. J. Wiley2, D. Hyndman1, P. Seelbach2, and B. Pijanowski3
Michigan State University, East Lansing, MI; 2 University of Michigan, Ann Arbor, MI;
3Purdue University, West Lafayette, IN
Abstract
Nutrients, dissolved oxygen (DO), and hydrologic alteration of streams are commonly affected by humans,
and all have profound effects on valued ecological attributes (VEAs). Few models explain relations among
human activities, these stressors, and VEAs with sufficient precision for nutrient criteria, Total Maximum
Daily Loads (TMDLs), and stream management. The objectives of our research are to refine relationships
among human activities, multiple common stressors, and the fisheries and ecological integrity of stream
ecosystems.
A multi-scale strategy is being employed to refine several tools used in water quality assessment. At the
broad spatial scale, stream conditions throughout the southern Michigan region were surveyed and existing
data from a variety of sources were compiled. We refined stream survey methods to relate early morning DO
to human activities, contaminants and habitat alterations, and VEAs by sampling conditions at low flow, more
than 3 days after rain events, and during early morning hours; thus controlling variation in DO as a result of
natural factors. Preliminary results show: nitrogen and phosphorus concentrations were positively related to
agricultural land use in watersheds; algal biomass increased with nutrient concentrations; and early morning
DO concentrations decreased with increasing algal biomass. Interestingly, indirect relationships in the causal
pathway from land use to nutrients, algal biomass, and low DO often were more precise than direct
relationships, which indicates high temporal variation in intermediate factors.
At finer scales, we established master watersheds with either continuous DO and water quality monitoring
or spatially intensive and repeated sampling. These watersheds will be used to parameterize processed based
models. The instrumentation and continuous monitoring of Crane Creek is providing parameters for DO
responses to complex interactions among processes operating at three temporal scales: (1) diurnal variation
with light periods and photosynthesis; (2) daily development between weather-related events of biological
assemblages that regulate diurnal variation in DO; and (3) weekly variation in weather disrupting hydrology of
streams and related biological community development.
A processed-based model relating precipitation and land use to water quantity and quality was refined
using intensive sampling of Cedar Creek. The model shows the importance of groundwater routes of transport
of contaminants to streams in the poorly developed soils with high permeability. These soils are typical of
many watersheds in the glaciated region of the upper Midwest and Great Lakes Region. Nitrate concentrations
in Cedar Creek were predicted much better by a model that included groundwater inputs than traditional
models without it. A 10-year simulation of the Cedar Creek watershed highlighted the significant role land use
plays in influencing the distribution of nutrient concentrations in groundwater and in streams.
Future activities will integrate results of our broad and fine scale field assessments in a suite of statistical,
processed-based, and hybrid models. Many of our statistical models are being used by the Michigan
Department of Environmental Quality to establish nutrient criteria. We will continue to refine our models to
improve understanding of how human activities can be managed to support ecological integrity, aquatic life
use, and fisheries of stream ecosystems.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Application of Individual-Based Fish Models
to Regional Decision-Making
Roland Lamberson and Steven Railsback
Humboldt State University, Humboldt, CA
Abstract
This project's goal is to develop and demonstrate the usefulness of individual-based models (IBMs) of
stream salmonids as a tool for regional decision-making. The specific objectives are to: (1) adapt our IBM that
simulates stream reaches to watershed-level assessment; (2) conduct a demonstration assessment; and (3)
examine uncertainties and sensitivities in the regional assessment.
We are using and enhancing inSTREAM (individual-based stream trout research and assessment model),
an existing IBM that represents how flow regimes and water quality parameters such as turbidity and
temperature affect individual trout and, consequently, trout populations. The first objective is being met by
developing ways to synthesize input for inSTREAM from data that are commonly available for western
streams. The second objective is being met by conducting an example assessment of how changes in turbidity
regime and flow regime, and presence of an exotic predator fish, affect a watershed's trout populations. The
third objective is being met by two studies; one examines parameter uncertainty and sensitivity in inSTREAM
applied to a single reach, and the second examines how sensitive model results are to uncertainty and
variability in the channel shape and hydraulic input used to represent a site.
The preliminary results are: (1) Stream habitat survey data collected by the California Department of Fish
and Game has been combined with other widely available information and stochastic modeling techniques to
synthesize input for inSTREAM. These methods can be used to generate input representing a variety of sites
within a region. (2) Site-specific example assessments show that inSTREAM is useful for predicting trout
population responses to stressors such as flow regime alteration, increased turbidity, and introduced fish.
Collaborative research with U.S. Forest Service scientists is improving our model of a key process: how
turbidity affects fish feeding efficiency. We also used inSTREAM to model interactions between wild trout and
stocked hatchery trout, and how the interactions could change if hatchery practices are changed to enhance
survival of stocked fish. (3) A parameter uncertainty analysis showed that the primary output of inSTREAM,
adult trout biomass, is not highly sensitive to any unexpected parameters and that the model can easily be
calibrated to reproduce the observed size and abundance of yearling and adult trout.
Our results so far confirm that inSTREAM can be, when applied and analyzed carefully, a uniquely
powerful tool for understanding the effects of multiple stressors on fish populations. The model is now in use
by several resource agency and university scientists, and we conducted our first training class in July 2005.
In the final year of the project, we plan to finalize a public release of inSTREAM, including its software, a
complete description of the model, and a guide to using it. These products are now in peer review prior to
publication. A demonstration assessment will be completed using the South Fork Eel River watershed as a site.
The analyses of uncertainties and sensitivities will be completed with studies of how sensitive the model's
ranking of management alternatives is to uncertainty in parameters and in site-specific input data.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Roland Lamberson
A One participant asked if the turbidity took into account watershed properties or just flows. For example,
watershed properties such as logging could have a big impact on that relationship. Dr. Lamberson
responded that turbidity was just an input parameter. They have a model for turbidity that they use in some
cases to generate data.
A Iris Goodman commented that this was an interesting effort with a lot of challenges to try to scale it up to a
regional application. It also could be very useful as a heuristic device in a simulated situation where you
could take a simulated hypothetical system to try to reproduce and understand the most critical
combinations of stressors. Dr. Lamerson responded that one of the major thrusts of the model is to be able
to use it to look at individual behavior characteristics and how they impact the emergent population levels.
They have looked at habitat selection and the interactions between wild and hatchery fish. In developing
the model, they looked at density dependence in fish populations.
A One participant asked how the sensitivity impacts were calculated. Dr. Lamberson responded that they
used professional judgment to establish the reasonable range for the parameter value. They then did
simulations to investigate sensitivity of abundance of adult mature fish.
A One participant asked how important parameters for the indices of fish integrity (e.g., embeddedness,
substrate, gradient) were incorporated into their model. Dr. Lamberson responded that they have detailed
habitat structures. The stream structure is the actual stream structure from a real stream in California.
There is difficulty in reproducing characteristics of a river because of problems in reproducing the
gradients. Their gradients are much deeper; they used digital elevation maps to reproduce the gradients.
Dr. Steve Railsback commented that they did not directly input emeddedness. They look at habitat
parameters that reflect the amount of velocity shelter. If a substrate is coarse and large, there are places
where fish can sit behind it and reduce its swimming speed, which reduces how much energy it needs to
grow. If it is a highly embedded substrate, there are no velocity shelters. If the fish wants to sit in the
current and feed, then it has to swim against the whole velocity of the river. Embeddedness also may affect
food production. They do not try to model growth and production of the food that the fish eat because it
could double or triple the complexity of the model. They do try to calibrate the model to try to reproduce
observed growth rates by the adjusting their food production parameter. Substrate embeddedness would
affect the food production. Channel shape and gradient are in the model but indirectly because they are in
the hydrolic model that predicts the depth and velocity of the cells at different points for a different flow.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Multiple Threats to Marine Biodiversity: The Interactive Effects
of Climate Change and Land Use Patterns on the Alteration
of Coastal Marine Systems by Invasive Species
RobertB. Whitlatch
University of Connecticut, Groton, CT
Abstract
Ecological responses of global warming are expected and coastal ecosystems are particularly vulnerable
because they are less buffered from temperature increase than more oceanic systems. Resulting alterations in
species distribution patterns, coupled with the enhanced vulnerability to invasions by exotic species, can result
in profound changes in communities and ecosystems. Added to the effects of climate change is the intense
human-related use of these areas, which have resulted in a host of problems that often lead to habitat
degradation and reduced ecosystem function and biodiversity. We have been using the southern New England
coastal zone as a model to study the interactive effects of climate change, land use and invaders on ecosystem
function and biodiversity. The abundance of invaders has increased during the past 20 years and there has been
a decline in the numbers of resident species. Recent invaders are almost exclusively found in coastal systems
that possess reduced biodiveristy and that are most heavily impacted by human activities. In addition, there are
strong within-habitat inverse relationships between resident diversity and invader diversity, suggesting that
variations in local environmental conditions that influence local biodiversity are important components in
response of coastal ecosystems to invasion susceptibility.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Whitlatch
A Iris Goodman asked Dr. Whitlatch to discuss the implications on a functional basis of the increased
cuttlefish invaders. Dr. Whitlatch replied that this most recent invader is causing a lot of concern in the
shellfish industry. It has been reported that 40 square miles of Georgia's banks are covered by this species,
which influences the scallop fishery. The fishing activities of the fishers are acting to disperse the
organisms over a wider area. It has serious potential economic implications.
A One participant asked if it was a gross generalization to say that we might have some positive effect on the
invasive species if we improve water quality. Dr. Whitlatch responded that water quality is correlated with
it. The chlorophyll data for Long Island Sound is not very good. These organisms are all suspension
feeders. It could be the interaction of temperature and productivity that is actually facilitating more
recruitment (i.e., more eggs and sperm). As you increase the water quality of very polluted areas, you start
to see invaders coming in. This also is true for shipworms. Docks that were built around the turn of the
century were not affected by shipworms, but when the water quality is increased there is a shipworm
problem; as a result, the docks are falling down. There are positives and negatives for cleanup of the
coastal zones.
A One participant asked if there might be threshold effects in terms of the relationship between loss of
biodiversity and invasives. Dr. Whitlatch responded that they have seen other types of threshold effects in
their system. Predators of recently settled invaders are vital in controlling invaders. When those predators
are removed or not present there is an abundance of invaders.
A One participant commented that Dr. Whitlatch had shown a strong negative relationship between
biodiversity and invaders, and asked if it was expected to see that relationship in other systems if space
was not such a strong major limiting resource. Dr. Whitlatch responded that if you do the same experiment
and use a native species as the invader, you see exactly the same effect. It is not the fact that the native
species are somehow collectively acting together to reduce the effects of a new species invading the
environment. They are collectively reducing the amount of available space at any one particular time. The
more species you have, the less space you have because of variations and fluctuations of their life
histories. In soft sediment systems, they have tried to do similar experiments but there is no effect.
Animals are not as space limited in soft sediment systems; they live in a three-dimensional sediment
column, and they can partition their space both horizontally and vertically.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Climate, Land Use and UV Radiation
on Aquatic Ecosystems
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Interactive Effects of Climate Change, Wetlands, and Dissolved
Organic Matter on UV Damage to Aquatic Foodwebs
in the Ontonagon Watershed, Northern Michigan
Scott D. Bridgham1, Gary. A. Lamberti2, Carol A. Johnston3, Paul C. Frost2, James H. Larson2,
Patricia A. Maurice2, David A. Lodge2, Kathryn C. Young2, Boris A. Shmagin3, andKangsheng Wu3
University of Oregon, Eugene, OR; University of Notre Dame, Notre Dame, IN;
South Dakota State University, Brookings, SD
Abstract
Understanding the factors controlling ultraviolet radiation (UVR) flux into aquatic ecosystems is critical
given its direct and indirect effects on many ecological processes. The strongest attenuator of light and UVR in
aquatic ecosystems is dissolved organic matter (DOM), which previous studies suggest is controlled at the
landscape scale primarily by wetland area and the discharge regimes of rivers and streams. Climate change
may reduce DOM concentrations in aquatic ecosystems, thereby exacerbating UVR effects, by changing the
amount and flow paths of DOM from upland and wetland ecosystems. We hypothesize that the linkages
among wetland area and type, DOM, and climate will be the most important factors determining the amount of
UVR damage to aquatic ecosystems at the landscape scale. Our objectives are to: (1) relate DOM
concentration and chemistry in various tributaries of a relatively pristine watershed in the Lake Superior
drainage basin (Ontonagon River in northern Michigan) to wetland and upland landscape characteristics and
discharge via multivariate analysis; (2) determine interactions among UVR intensity and DOM chemistry,
photodegradation, and biodegradation; and (3) determine the response of stream foodwebs to the interactions
between UVR intensity and DOM concentration and type.
We have addressed these objectives through a combination of landscape and hydrological analyses;
extensive sampling of water chemistry and UVR attenuation profiles in numerous streams for 2 years; analysis
of the chemistry, photodegradation, and biodegradation of DOM; and experiments quantifying the effects of
DOM concentration and source on UV damage on stream biota in a series of artificial stream experiments.
We sampled a dozen times over a 2-year period water chemistry and determined discharge in 35 sub-
watersheds in the Ontonagon watershed. Additionally, we did an initial survey of 60 sampling locations within
the Ontonagon River watershed in September 2003. Thus, we have developed an extensive spatial and
temporal dataset of water chemistry within this 3600 km2 watershed. The best published landscape correlate
with DOM flux in streams and rivers is with soil carbon:nitrogen (C:N) ratios. As this information was not
available otherwise, we sampled 155 representative soils in the dominant soil types within the watershed to
estimate the spatial coverage of soil C:N ratios within the watershed.
We have measured the spatial and temporal variation in the attenuation of UVR in a variety of streams
within the Ontonagon Watershed and related UVR dosage to stream biota to water depth, water chemistry, and
canopy cover. Additionally, we used plastic dosimetry strips to obtain a better spatial and temporal of UVR
dosage within the watershed.
Our hydrological modeling research has taken a two-pronged approach. First, we have used USGS
gauging station data, land-use data, and factorial analysis to examine landscape controls over seasonal and
annual river discharge at four spatial levels (U.S. conterminous, U.S. Great Lakes basin, Upper Michigan, and
Ontonagon watershed). Second, we recently hired a postdoctoral associate to use a basin-scale model to predict
discharge and seasonal DOM concentrations in all 35 subwatersheds of the Ontonagon that we have been
studying. Substantial progress has been made on parameterizing this model, and when complete we will use
global climate model predictions to examine how climate change will affect DOM concentration and UVR
penetration within the Ontonagon watershed.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
We constructed a large artificial stream facility to examine experimentally how DOM and UVR interact in
controlling food-web structure in streams. A number of different experiments have been completed and are
being prepared for submission for publication.
Our major findings to date are: (1) As predicted, DOM is the major control over UVR dosage to aquatic
biota and in streams and rivers with high DOM concentrations, such as the Ontonagon Watershed, most biota
experience very low UVR exposure. (2) Although wetlands are a significant landscape predictor of DOM
concentration and chemistry in streams, the area of different types of wetlands is important (some even having
a negative correlation with DOM concentration), and other factors are important in predicting DOM
concentrations in the complex glacial geology of our study area. (3) Climate and landscape effects on
discharge and DOM concentration likely will have a much larger effect on aquatic biota in wetland-dominated
watersheds than will UVR effects.
As soon as the soil C and N samples have been analyzed, we will use multivariate statistics to relate DOM
concentration and chemical characteristics to season, discharge, and landscape characteristics in one of the
most complete datasets to have attempted this at a relatively large watershed scale. When the basin-scale
hydrological model is fully parameterized, we will add in our empirical landscape predictors of DOM
dynamics to test how anticipated climate change will affect discharge, DOM concentration and chemistry, and
UVR dosage of biota throughout the Ontonagon Watershed.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Scott Bridgham
A One participant commented that the soil C:N as a control for DOM is going to be a reflection of the
dominant trees and it will probably be responsive to the atmospheric deposition. Dr. Bridgham responded
that one paper that was published had a r2 = .99 (no variation) for the United States. The best correlation is
not just the C:N ratio. In terms of cerol carbons, its value would be as an index of huminification. The
more nitrogen, the lower the ratio; this results in more humified organic matter, which probably would
reflect the amount of DOM.
A One participant asked if the photodegradation of the DOM could be related to the C:N ratio from whatever
forms of nitrogen are present and, therefore, affecting different levels of production molecules such as
hydrogen peroxide. Dr. Bridgham responded that this comment was correct. The huminification would
have everything to do with the chromofores and their effect on light. It was beyond the scope of the study
to look at some of the other photochemistry aspects. Most of their DOM was terrestrially derived.
A One participant commented that in many streams the residence time is too short to have an extensive
photochemical effect, but it is more likely in ponds or lakes where it would be a bigger factor. Dr.
Bridgham responded that many of the streams in their study are quite flat, wetland dominated, and with
low flow. The other factor is high attenuation, so most of the water is not exposed to the sun. They did a
fair amount of photodegradation and biodegradation work to try to understand the controls. There was a
huge effect on DOM chemistry but not much effect on concentration.
A One participant asked Dr. Bridgham to comment on the role of DOM concentration versus the optical
properties of DOM for different wetland types. Dr. Bridgham responded that very little work has been
done in this area. Some studies have looked at different DOM concentrations, with very high DOC
concentrations of 50 to 100 milligrams carbon per liter. There has been very little work done on mineral
soil wetlands, including how the flow pathways would change and the effect of the DOM as it goes
through the mineral soil down to the streams. These transport, biodegradation, and sorption mechanisms
that will have huge effects on both the concentrations and the chemistry. Almost no work has been done in
this area.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
The Influence of Climate-Induced Alterations in Dissolved Organic
Matter on Metal Toxicity and UV Radiation in Rocky Mountain Streams
William Clements1, Jill S. Baron1, Diane M. McKnight2, and Joseph S. Meyer3
Colorado State University, Fort Collins, CO; University of Colorado, Boulder, CO;
3 University of Wyoming, Laramie, WY
Abstract
The primary goal of our research was to investigate the influence of climate-induced changes in hydrology
and dissolved organic material (DOM) on responses of stream ecosystems to the combined stress of ultraviolet
radiation (UVR) and heavy metals. We hypothesized that changes in climate and UVR will alter the quality
and quantity of DOM in Rocky Mountain streams. Because DOM regulates light attenuation and metal
bioavailability in these systems, we predicted that exposure to UVR and metals will increase as a result of
changes in DOM. We integrated climate and hydrologic modeling with an intensive field monitoring and
experimental program to test the hypothesis that reductions in DOM increase bioavailability of metals and
exposure to UV-B (280-320 nm) radiation. We estimated effects of climate-induced alterations in stream
hydrology on DOM using DayCent-Chem, a model that simulates carbon, water, and nutrient flux to streams.
In controlled laboratory experiments with a full-spectrum solar simulator, we irradiated DOM collected from
six field sites for 24 hours and quantified changes in Cu complexation using a Cu2+ selective electrode.
Additionally, we conducted field and microcosm experiments to test the hypothesis that UV-B exposure
impacted benthic communities and that effects varied among locations along a metals gradient.
Analyses using the DayCent-Chem model showed that increasing temperature linearly by 2.5 °C over a 50
year period increased DOC concentrations relative to control runs. Results of our monitoring studies indicated
that across 21 watersheds there was a significant relationship between DOM concentration and stream
discharge. However, the timing of peak discharge and maximum concentrations of heavy metals, DOC and
other physicochemical variables that influence metal toxicity varied among streams. Uptake of metals by the
caddisfly (Trichoptera) Arctopsyche grandis was predicted by the biotic ligand model (BLM), a model that
accounts for physicochemical factors influencing metal bioavailability. Results showed close agreement
between observed and predicted metal levels in these organisms. After 24-hour exposure to simulated sunlight,
DOM absorbance decreased by approximately 17-25 percent, significantly increasing exposure of benthic
communities to UV-B. Photobleaching of DOM also increased the fraction of bioavailable Cu2+, potentially
increasing toxicity to aquatic organisms. Microcosm and field experiments conducted in Colorado and New
Zealand showed significant effects of UV-B on benthic communities. Results indicated that mayflies
(Ephemeroptera) were especially sensitive to UV-B, and that effects differed among locations. Our findings
are significant because they demonstrate that benthic communities in shallow, alpine streams are exposed to
intense UV-B radiation and that climate-induced reductions in DOC are likely to increase both metal
bioavailability and UV-B exposure. In most cases, effects of these two stressors were additive; however, there
were some examples where combined exposure to metals and UV-B were greater than either stressor alone. In
the final phase of our research we are continuing to investigate the influence of hydrology and increased
temperature on DOC concentration and export using the DayCent-Chem model.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. William Clements
A One participant asked if there were data on the contribution of autochthonous versus allochthonous DOC
to the streams. Dr. Clements responded that Dr. Diane McKnight has done work using fluorescence index
on characterizing not just autochthonous versus allochthonous but also a detailed quantitative analysis of
the different types of carbohydrates and carbons. There is a seasonal variation in the relative importance of
autochthonous versus allochthonous. Allochthonous appears to dominant in these alpine streams.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Interactions Among Climate, Humans, and Playa Wetlands
on the Southern Great Plains
Scott McMurry, Lor en Smith, David Willis, W. P. Dayawansa, Clyde Martin,
Ken Dixon, and Chris Theodorakis
Texas Tech University, Lubboch, Texas
Abstract
We measured avion, amphibian, and plant communities in 80 playas split between 2 years. In each year, 20
playas each were sampled in cropland watersheds and native grassland watersheds. Physical features of all
playas were measured, including sediment depth, hydroperiod, basin volume, and playa area. This project was
designed to address the hypothesis that climatic variability and landuse practices (e.g., crop production,
conversion to grasslands) dictate hydroperiod and spatial distribution of wet playas, influencing the ecological
structure of vegetation and animal communities that rely on playa wetlands for many life requisites. All field
data have been collected, and we are in the process of analyzing all results and coordinating field data with
model projections for climate effects on playa hydrology.
Grassland and cropland playas used in this study were in either fine or medium textured soils. Playas in
fine textured soils were typically larger and contained less sediment than those in medium textured soil.
Cropland playas contained more sediment (2-4 times greater) and had greater volume loss than grassland
playas, regardless of soil texture. Hydroperiod was similar between cropland and grassland playas in 2003 (100
vs. 97 days, respectively) but not in 2004 when average hydroperiod length in grassland playas was 76 days
longer than in cropland playas. Hydroperiod differences between years likely reflect differences in annual
precipitation. Total rainfall in 2004 was about three-fold greater than in 2003.
Analysis of the avian and amphibian community data still is in progress. Preliminary results suggest that
overall richness of avian communities is similar between cropland and grassland playas, ranging from six to
eight species in each playa type. Exotic avian species were, however, typically observed in cropland playas to
a greater degree than grassland playas. Species richness was typically two- to three-fold greater in wet playas
than dry playas in both years. Also, avian community richness often was positively related to playa size, but
this was dependent on the hydrologic state of the playa and season (e.g., wet playas in autumn, r2 = 0.497; dry
playas in summer, r2 = 0.443).
Average amphibian community richness also was similar between cropland and grassland playas, with
about three to four species observed on average in each playa type in 2003 and 2004. Cropland playas,
however, were more likely to have no species (2003) or only one species (2004) compared to grassland playas,
which always had at least two species. Overall, species richness was strongly related to hydroperiod in playas,
with richness dropping sharply when hydroperiod fell below 100 days.
Current modeled predictions of hydroperiod and water volume in cropland playas show marked reductions
over a 50-year time horizon, depending on the use of soil management practices such as furrow dikes or buffer
strips. Indeed, although hydroperiod may remain stable during the first 16 years of cultivation (no furrow
dikes), the rate of reduction in hydroperiod is rapid once it begins. For example, assuming a maximum of 123
wet days from May 1 to August 31, a typical cropland playa will have only 100 wet days after 26 years,
leveling off at about 35 wet days after 39 years. In comparison, a typical native grassland playa is projected to
lose only about 5 wet days after 50 years under the same conditions. These projections are dependent on
cultivation strategies, crop type, soil texture, and climate, but suggest the potential for rapid deterioration of
cropland playas subject to cultivation pressure, and are supported by empirical data from our study and others.
We are in the process of modeling playa hydrology given stochastic temperature and rainfall data, and
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
assessing the influence of different crop types, watershed slopes, soil textures, and mitigation strategies, such
as furrow dikes and buffer strips, on hydroperiod and water volume.
Although preliminary, our data support the rapid reduction in playa hydroperiod (and increased rates of
water loss) as cropland playas fill with sediment. These changes in hydrology will have effects on playa
function as related to biotic community composition. Indeed, the effect of cultivation and subsequent sediment
loading on the hydrology of cropland playas is already occurring as evidenced by data from our study and
others. Plant, avian, and amphibian community composition are subject to these alterations in cropland playas.
For example, our data suggest that as hydroperiod length falls below 100 days, amphibian richness suffers
from the loss of species with longer larval periods (often the rarer species in the community).
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Scott McMurry
A One participant asked what the restoration potential was for the playa sediments. Dr. McMurry responded
that it could be done; however, there are questions about what the playa will look like and how it will
function. No one knows, but it will be better than it is when it is fossilized. The participant asked if the
vegetation regrows quickly. Dr. McMurry responded that no one knows. The first thing that needs to be
done is to put in protection for the playas that are left. The only reason that there are playas left in native
grasslands is because they exist in areas that a hard to farm. There are no incentives to protect them. It may
become economically feasible to protect them in the future because it costs a tremendous amount of
money to irrigate cotton fields by pumping out of the aquifer. It is somewhere between $3 to every $1 for
pumping out of the ground versus pumping out of a playa. A lot of farmers pump out of the playa. The
problem is that no one knows if the playa will be full of water in any given year. When the playa fills up
with sediment, then it will have water fewer times and for shorter periods of time.
A One participant commented that he had heard that cotton was the most heavily pesticide applied crop that
can be grown. Dr. McMurry responded that cotton gets pesticides, herbicides, insecticides, and defoliants.
The participant commented that the amphibians living in these playas are fairly terrestrial, so they are
being exposed both in the fields and in the playas. He asked if there was any indication of that type of
stressor on richness or diversity. Dr. McMurry responded that over the past 2 years they have been
conducting a side project to collect sediment in water. Last summer, they monitored 12 playas and
collected water samples and tadpoles each week for a month. They are in the process of analyzing the data.
Other studies that have been conducted have shown that amphibians from cropland playas weigh less and
have smaller body size, and the populations tend to be denser. There are alterations in physiological and
immune function endpoints. There are effects happening that may have something to do with the
differences in the availability of food, differences in the per capita food amount, or other stressors related
to hydroperiod. As part of another study, they have differences between crop and grass playas with respect
to the rate that the water is lost. The rate of water loss out of crop playas is significantly faster than in grass
playas.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Assessing the Consequences of Global Change for Aquatic
Ecosystems: Climate, Land Use, and UV Radiation
Bruce R. Hargreaves, Donald P. Morris, Frank J. Pazzaglia, Craig E. Williamson,
Richard N. Weisman, and Stephen C. Peters
Lehigh University, Bethlehem, PA
Abstract
Our primary objective in this study is to determine how current stream and landscape properties in the
Lehigh River watershed influence the aquatic ecosystem response to variations in climate and solar ultraviolet
radiation (UVR) and how past changes in land use have influenced stream processes. Our findings will
contribute to understanding future responses of aquatic ecosystems to environmental changes, including
anticipated increases in extreme weather conditions and increases in UV-B associated with ozone destruction
in the stratosphere.
Our approach is to examine watershed and stream properties and processes on different scales of space and
time to learn about functional relationships that influence UV impact. Biotic responses to UVR were tested by
placing stream macroinvertebrates in a laboratory solar simulator, and by manipulating UV-B in situ for stream
trophic groups. For stream-water shed interactions, we sampled stream water periodically across the entire
watershed during base flow and storm flow conditions and also established stations at several headwater
streams with automated equipment for sampling and monitoring rainfall, canopy throughfall, and stream water.
Our samples were analyzed for properties that influence UV attenuation. We also made in situ and laboratory
measurements of photobleaching rates and biolability of stream colored dissolved organic matter (CDOM).
The role of the stream canopy on UVR exposure was examined with two complementary approaches: direct
measurement of the canopy UVR transparency for specific stream reaches, and proxy measurements for
predicting UV transparency (fisheye camera images from below the canopy and satellite images from above).
We also digitized aerial photographs that record changes in land use and stream channel morphology since the
1940s to relate current storm flow patterns measured with in situ flow recorders.
In stream experiments with macroinvertebrates, the abundance of chironomids was enhanced during
exposure to UVR, in spite of their sensitivity to UV-B. The optical quality and source of DOC differed by
region: streams in Lehigh Valley region were enriched in algal CDOM whereas Pocono streams were
dominated by wetland soil CDOM. On average, 14 percent of stream dissolved organic carbon (DOC) was
consumed by bacteria during 2-day microcosm experiments, but this number increased to more than 40 percent
when the water was photobleached by the equivalent of 2 days of sunlight. Forests have lower DOC export
than streams during baseflow, but during storms the canopy is a major contributor to stream CDOM in forests.
Turbidity and CDOM change on different times scales during storm runoff. We developed a method to relate
laboratory water quality measurements directly to in situ UV attenuation. Stream DOC is correlated with
percent area of wetlands in the watershed and in the absence of wetlands is correlated with percent area of
farmland. At any given site, stream DOC is inversely correlated with specific conductance, a reflectance of the
lower concentration of DOC in ion-rich groundwater.
In our previous work in lakes, we observed large variations among groups of organisms in their sensitivity
to UVR and that DOC was a strong determinant of UV attenuation. We also expected suspended particles to
have a bigger impact on UV attenuation in streams compared to lakes, especially during storm runoff. We also
knew that UVR exposure of streams would be reduced in forested areas because of shading by the canopy. We
now have UV-specific information on these factors for stream ecosystems in our region that can be tested for
generality in other regions.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Our next steps include: (1) exploring the observed patterns for stream-watershed interactions in other
parts of the watershed; and (2) determining whether shrubs and crop vegetation also contribute to CDOM in
streams during storm runoff. We plan to integrate measurements of biolability and photolability with estimates
of canopy UVR transparency to estimate the turnover of stream DOC (and corresponding oxygen demand)
across the watershed.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for Aquatic Ecosystems
Question-and-Answer Session with Dr. Bruce Hargreaves
A One participant questioned an outlier data point on the graph for canopy transmittance for diffuse
irradiance. Dr. Hargreaves stated that everything is equivalent. Direct light and diffuse light change
together. Diffuse light varies primarily on the openness of the canopy, whereas direct light depends on
where the canopy opening is versus where the sun is, the time of day, and which way the steam is oriented.
He could have said in his presentation that there is a high variability in the direct transmittance. If things
are equivalent, you get a positive relationship, but there is a great potential for direct light to be all over the
map. In reference to the outlier, that kind of data is not useful without a detailed mapping of where the
gaps are in the canopy.
A One participant commented that r2 is meaningful in terms of statistical significant when you have bivariant
normality (i.e., you have a normal distribution in any direction). The reason that the r2 is equal to 0.8716 is
largely due to the point on the right. The r2 does not have any meaning when virtually all of the data points
are clustered in one spot. Dr. Hargreaves agreed. They find that relationship when they look at their
wetlands. There is variation around the regression but they also find a trend. The data shown were a
subsampling of their analysis. Dr. Hargreaves stated that he would have to fill in the gaps for his data if he
were to compare to the r2 of 0.1825 that Dr. Bridgham reported that had a more even distribution.
The Office of Research and Development's National Center for Environmental Research
-------
Joint Progress Review for U.S. EPA STAR Grants: Regional-Scale
Stressor-Response Models and Consequences of Global Change for
Aquatic Ecosystems
Sheraton National Hotel
900 S Orme Street
Arlington, VA 22204
November 3-4, 2005
Participants List
Rochelle Araujo
U.S. Environmental Protection Agency
Michele Aston
U.S. Environmental Protection Agency
Scott Bridgham
University of Oregon
Rebecca Clark
U.S. Environmental Protection Agency
William Clements
Colorado State University
Noha Gaber
U.S. Environmental Protection Agency
Charles Gallegos
Smithsonian Institution
Iris Goodman
U.S. Environmental Protection Agency
Michael Haire
U.S. Environmental Protection Agency
Bruce Hargreaves
Lehigh University
Robert Howarth
Cornell University
David Hyndman
Michigan State University
Tim Icke
U.S. Environmental Protection Agency
Thomas Johnson
U.S. Environmental Protection Agency
Susan Julius
U.S. Environmental Protection Agency
Joseph Koonce
Case Western Reserve University
Roland Lamberson
Humboldt State University
E. Conrad Lamon III
Duke University
Xuyong Li
Smithsonian Institution
Elias Manolakos
Northeastern University
Scott McMurry
Texas Tech University
Charles Noss
U.S. Environmental Protection Agency
Vladimir Novotny
Northeastern University
Pasky Pascual
U.S. Environmental Protection Agency
Steve Railsback
Humboldt State University
Kenneth Reckhow
Duke University
-------
Joe Roman
American Association for the Advancement of
Science - EPA Fellow
U.S. Environmental Protection Agency
Joel Scheraga
U.S. Environmental Protection Agency
Rita Schoeny
U.S. Environmental Protection Agency
Patrick Schoff
University of Minnesota at Duluth
Michael Slimak
U.S. Environmental Protection Agency
Bernice Smith
U.S. Environmental Protection Agency
R. Jan Stevenson
Michigan State University
Dennis Swaney
Cornell University
Rosaura Vega
U.S. Environmental Protection Agency
Hongqing Wang
Florida A&M University
Don Weller
Smithsonian Institution
Robert Whitlatch
University of Connecticut
Katie Wolff
U.S. Environmental Protection Agency
------- |