Grantee Research Project Results
Final Report: Interactive Effects of Climate Change, Wetlands, and Dissolved Organic Matter on UV Damage to Aquatic Foodwebs
EPA Grant Number: R829643Title: Interactive Effects of Climate Change, Wetlands, and Dissolved Organic Matter on UV Damage to Aquatic Foodwebs
Investigators: Bridgham, Scott D. , Shmagin, Boris A. , Johnston, Carol A. , Lamberti, Gary A. , Maurice, Patricia A. , Frost, Paul C , Lodge, David M.
Institution: University of Oregon , Natural Resources Research Institute , Trent University , University of Notre Dame
EPA Project Officer: Packard, Benjamin H
Project Period: June 24, 2002 through June 23, 2005 (Extended to June 23, 2006)
Project Amount: $897,307
RFA: Assessing the Consequences of Global Change for Aquatic Ecosystems: Climate, Land Use, and UV Radiation (2001) RFA Text | Recipients Lists
Research Category: Climate Change , Ecological Indicators/Assessment/Restoration , Water , Aquatic Ecosystems
Objective:
Project Overview
Understanding the factors controlling ultraviolet radiation (UVR) flux into aquatic ecosystems is critical given its deleterious effects on many ecological processes. UVR is strongly attenuated in aquatic ecosystems by dissolved organic matter (DOM), and thus we hypothesized that landscape controls over DOM would ultimately control UVR exposure to, and subsequent effects on, stream biota. Previous research suggests that the quantity and quality of DOM at the landscape scale is controlled primarily by: (1) vegetation community and soil type, with wetland area being of particular significance; (2) flow paths through soil; (3) the discharge regimes of rivers and streams; and (4) within-stream DOM degradation and production mechanisms. Climate change will likely affect each of these DOM control factors in complex ways. While a significant amount of previous research has focused on the separate roles of DOM and UVR in aquatic ecosystems, much less is known about the interactive effects of climate change, landscape, DOM, UVR, and aquatic foodwebs.
The overarching objective of this project was to enhance the understanding of how UVR affects foodweb structure in streams and rivers through its complex interactions with DOM, landscape characteristics, and climate in a northern forested watershed (Figure 1).
Figure 1. Proposed Interaction of Landscape and Land-Use, Climate Change, DOM, and UVR and its Effect on Aquatic Foodwebs
More specifically, we had five main objectives:
- Determine the extent to which UVR exposure in streams is controlled by DOM concentration and chemistry.
- Determine the response of stream foodwebs to the interactions among UVR intensity and DOM concentration and type.
- Determine landscape controls over DOM concentration and chemistry (and, hence, UVR).
- Determine how in-stream processing of DOM through biodegradation and photodegradation varies spatially and controls over that spatial variation.
- Determine how various climate change scenarios will affect discharge and, thus, DOM concentration and UVR exposure.
Overall Experiment Design
We tested our central hypothesis and the linkages at a variety of spatial scales ranging from artificial streams and laboratory experiments to an entire watershed, as was appropriate to address the project’s five main objectives. For our landscape analyses that related to Objectives 1 and 3, we sampled 60 catchments within the larger 3,460 km2 Ontonagon River watershed in northern Michigan and Wisconsin (Figure 2) for DOM concentration and chemistry and related water chemistry variables in September 2003. Based upon this initial sampling, we sampled 35 catchments at 11 additional times in all seasons during the next 2 years. We also measured discharge at each location at most time points. Additionally, we sampled soil throughout the basin to obtain carbon and nitrogen content. We assembled a large GIS database of land cover and use, stream characteristics, wetland characteristics, soil type, and surficial geology. We then used these rich data in a variety of multivariate analyses to describe landscape controls over DOM concentration and chemistry to fulfill Objective 3 (Frost, et al., 2006a; Larson, et al., 2007; Johnston, et al., in press; Bridgham, et al., in preparation). In a subset of these sites, we determined UVR penetration with depth in the water column and related it to DOM concentration and chemistry and other water quality variables to fulfill Objective 1 (Frost, et al., 2005; Frost, et al., 2006b).
Figure 2. Map of Ontonagon River Watershed in Northern Michigan, USA. Smaller order streams (< 3) are not shown but were included in the study (Frost, et al., 2006a).
To fulfill Objective 2, we constructed a large artificial stream facility at the University of Notre Dame Environmental Research Center, at the southern edge of the Ontonagon basin, to examine experimentally how DOM and UVR interact in controlling foodweb structure in streams. We added DOM with natural UVR exposure and with UV-B removed and examined the effects on stream periphyton and algal communities (Frost, et al., 2007). We also used this facility for several other experiments to examine the effects of DOM concentration and chemistry on stream biota (Larson, 2006, Frost, et al., 2007)).
We performed a variety of experiments in the artificial streams, in the lab, and in situ to examine the effects of photodegradation and biodegradation on DOM concentration and chemistry to fulfill Objective 4 (Young, et al., 2004; Young, et al., 2005; Docherty, et al., 2006; Larson, 2006; Cherrier, et al., in preparation).
To fulfill Objective 5, we modeled discharge in the larger branches of the Ontonagon River and examined climatic and land cover controls over discharge in the basin (Wu, et al., 2006, Wu and Johnston, 2007; Wu and Johnston, in press). We also related DOM to discharge in the 35 repeatedly sampled catchments (Bridgham, et al., in preparation). However, as most of these catchments are not continuously gauged and thus we could not adequately parameterize the hydrologic model for them, we were somewhat limited in our ability to relate climate-driven changes in hydrology to DOM concentration (and hence UVR exposure) in smaller streams. Furthermore, to put the Ontonagon basin within a larger geographic context, long-term stream flow records (1956–1988) from 32 U.S. Geological Survey (USGS) gauging stations within the Great Lakes Basin were analyzed using multivariate statistical techniques (Johnston and Shmagin, in preparation).
Summary/Accomplishments (Outputs/Outcomes):
Landscape Controls over UVR Exposure in Streams
We examined the attenuation of UV-B and UV-A radiation flux and its environmental control directly using spectrometry in 32 streams in the Ontonagon River watershed in 2003 and 2004 (Frost, et al., 2005). Additionally, variation in UV-B and UV-A radiation within and among streams was examined using plastic dosimetry strips along longitudinal transects in seven streams in this same watershed in 2004 (Frost, et al., 2006b). Plastic dosimetry strips indicate a cumulative UV dose by changes in their absorbance characteristics and, deployed over 1–2 days, allow for integrated UV measurements at different depths within a stream and under different forest canopy types.
Both experiments demonstrated strong effects of DOM concentration on attenuation of UV-B and UV-A and the depth at which 1% of the incident UVR remained (Figure 3). Shading, as determined by canopy cover and stream width, also had strong effects on UVR exposure (Frost, et al., 2005; Frost, et al., 2006b). We used this information to develop a model to predict UVR exposure in streams in this region (Figure 4).
Figure 3. Relationships Between UVR Attenuation (Kd and the Depth at Which 1% of Incident UVR Radiation Remains) and Dissolved Organic Carbon (DOC) Concentrations in Streams in the Ontonagon River Watershed in Northern Michigan (Frost, et al., 2005). Kd for UV-A and UV-B was calculated as the slope of the regression between natural log-transformed irradiance and stream depth.
Figure 4. Modeled UV-B Flux in Streams as a Function of Water Depth, DOC, and Forest Canopy Cover as Determined by a Stream Ultraviolet Model (Frost, et al., 2005)
Interestingly, it appears that DOM is more effective at attenuating UV-B in streams than in other aquatic ecosystems, likely reflecting its largely terrestrial origin, short residence time, and limited photodegradation and biodegradation (Frost, et al., 2005). Supporting this interpretation, we found that streams originating from lake outflows had lower DOM concentrations and higher UV-B exposure than streams without lake inputs (Larson, et al., 2007).
Overall, our results indicate that the high DOM concentrations in the Ontonagon River watershed, and in other similar systems, strongly limit UVR exposure to aquatic biota in the vast majority of streams and rivers.
Response of Stream Biota to Interactions among UVR Intensity and DOM Concentration and Type
We constructed a large artificial stream facility at the University of Notre Dame Environmental Research Center, at the southern edge of the Ontonagon watershed, to examine experimentally how DOM and UVR interact in controlling foodweb structure in streams. The artificial stream facility consists of 24 channels fed with groundwater or lake water with motorized paddles providing current. We conducted two artificial stream experiments in the summer of 2003 and two experiments in the summer of 2004.
Overall, these experiments indicate minimal effects of UV-B on periphyton mass or algal community composition, but increased DOM concentrations caused greater accumulation of periphyton mass (Figure 5), altered its C, N, and P stoichiometry, and changed algal community composition (Larson, 2006; Frost, et al., 2007).
Figure 5. Periphyton Ash-Free Dry Mass (AFDM), Chlorophyll A, and Algal C in Streams Receiving Ambient UV-B, Plus DOM (gray circles, solid line), No UV-B, Plus DOM (gray circles, dashed line), Ambient UV-B, Ambient DOM (open circles, solid line), and No UV-B, Ambient DOM (open circles, dashed line). Error bars represent 1 standard error (s.e.). Treatments with different letters were found significantly different (P < 0.05) from each other with a 2-way ANOVA on the final day of the experiment (Frost, et al., 2007).
We also performed two experiments where we reciprocally transplanted the microbial communities from two to three different aquatic DOM sources. These experiments indicated strong effects of DOM source on microbial community structure and growth and large effects of microbial community structure on DOM biodegradation dynamics (Young, et al., 2004; Young, et al., 2005; Docherty, et al., 2006).
Overall, our results indicate that UVR has minimal effects on stream biota in these streams because of high natural DOM concentrations, but that DOM concentration and type is a primary driver of periphyton and microbial community structure and ecosystem function.
Landscape Controls over DOM Concentration and Chemistry
Given the above results showing the central importance of DOM in controlling UVR exposure to stream biota, and in controlling overall stream structure and function, we spent considerable efforts to understand landscape controls over DOM concentration and chemistry. We sampled 35 catchments in the Ontonagon River basin a dozen times over a 2-year period. Additionally, we did an initial survey of 60 sampling locations within the Ontonagon River basin in September 2003, upon which we based our final selection of the 35 sites that were sampled seasonally. In each sample we measured DOM concentration and chemistry, dissolved and particulate nitrogen and phosphorus, cation concentrations, particulate carbon, total suspended solids, chlorophyll, and stream pH. In addition, stream gauge height was recorded at each visit to a site, and discharge was also determined 4–6 times at each site to develop stage-discharge curves.
The second component to our watershed analysis was to relate DOM concentration and physiochemistry in tributaries of the Ontonagon River to discharge and landscape characteristics with multivariate statistics (multiple regressions, Akaike Information Criterion [AIC], and classification and regression trees [CART]). We compiled an extensive GIS landscape database for each of the 60 catchments, including stream characteristics, wetland abundance and type, upland landscape characteristics, land use, surficial geology, soil carbon and nitrogen, and topography.
This landscape analysis showed a strong effect of wetland area on DOM concentration and physiochemistry, but the area of different types of wetlands may have positive or negative correlations with DOM concentrations (Frost, et al., 2006a; Johnston, et al., in press). However, there are also important additional landscape controls over DOM concentration and physiochemistry, including upland land cover and topography (Frost, et al., 2006a; Bridgham, et al., in preparation). The DOM concentration/discharge relation was widely variable among the catchments but it could be predicted well with percentage development, soil C:N ratios, stream density, and a limited number of surficial geology predictors (Bridgham, et al., in preparation). Many of the streams in this region originate as lake outflows, and such streams have lower DOM and molar absorptivity (UVR adsorbance per unit C) than streams not originating from lakes (Larson, et al., 2007).
Controls Over In-Stream Processing of DOM
We performed two studies that examined the role of microbial community structure and the initial molecular-weight distribution of DOM in DOM biodegradation rates. These studies have resulted in three publications (Young, et al., 2004; Young, et al., 2005; Docherty, et al., 2006).
We completed a long-term DOM biodegradation experiment that addressed three questions:
- How do DOM concentration and quality interact to affect DOM biodegradation rates?
- Does low nutrient availability constrain DOM biodegradation in some streams?
- Is the effect of photodegradation on DOM biodegradation rates dependent on the DOM source?
We filtered water from six streams with different DOM concentrations and chemistry characteristics and added a composite microbial community from all six sites to each sample. We additionally had three treatments: (1) + nutrients; (2) no nutrients; and (3) photodegraded and then biodegraded, with five replicates of each stream-treatment combination. Response variables included short-term DOM biodegradation by determining CO2 production and bacterial production over 72 hours and long-term DOM biodegradation by measuring the change in DOM concentration until it stabilized, with the remaining fraction indicating the recalcitrant portion of the DOM. We are still in the final process of analyzing the results, but overall they indicate that most stream DOM is highly recalcitrant and not biodegradable even after prolonged decomposition, biodegradation of DOM is not nutrient limited, and prior photodegradation substantially enhances biodegradation.
In two other experiments, we examined the relative importance of photodegradation, biodegradation, and periphyton on the concentration of DOM and its photoreactions (Larson 2006). Overall, these experiments indicated that all three mechanisms, and their interactions, are important in the consumption of stream DOM.
Effects of Various Climate Change Scenarios on Discharge
The hydrology of the Ontonagon River is greatly influenced by the climate-sensitive phenomenon of snow melting (Wu, et al., 2006; Wu and Johnston, 2007). Long-term stream flow data from USGS gauging stations and discharge data collected for this project consistently showed that peak monthly discharge occurred during April snowmelt events, when the accumulated snow pack rapidly released water into streams. Because April discharge is typically three to six times that of flow during the rest of the year, snowmelt events dominate total annual discharge. The high discharge that occurred during snowmelt in April 2003 caused a convergence of DOC concentrations in which normally low DOC streams increased in concentration and normally high DOC streams decreased in concentration (Johnston, et al., in press).
Hydrologic response to climate change was evaluated by creating two versions of the Soil and Water Assessment Tool (SWAT) model for the 901 km2 South Branch Ontonagon River watershed: one calibrated for average climatic conditions (1969–70) and the other calibrated for drought conditions (1948–49) (Wu and Johnston, 2007). Snow melting parameters were altered by drought conditions, increasing snow melting rate and forcing peak flow timing to occur earlier in the spring. The hydrology of the South Branch Ontonagon River was also influenced by the climate-sensitive phenomenon of evapotranspiration. Monthly evapotranspiration in this forest-dominated watershed equals or exceeds monthly precipitation for five months of the year, making evapotranspiration an important factor in the annual water budget. Our calibration results imply that drought conditions reduce the capacity of upper soil layers to meet soil evaporative needs, altering plant water uptake strategies under varying climatic conditions. When the two versions of the SWAT model were applied to a validation data set (1950–65), the drought-calibrated version performed much better than did the average-calibrated version because of the dry conditions that prevailed during most of the validation period (Wu and Johnston, 2007).
The SWAT model was also used to compare the hydrology of adjacent catchments dominated by wetlands and lakes (Middle Branch Ontonagon River) versus forest (East Branch Ontonagon River) (Wu and Johnston, in press). The wetland/lake dominated catchment exhibited substantially different temporal flow trends than did the adjacent forested wetland, with lower peak flows in April and higher baseflows during summer months (Figure 6). These results suggest an important storage function of wetlands and lakes, increasing a watershed’s ability to moderate extreme flows and gradually release water into receiving streams through baseflow. The snow melting algorithm in the SWAT model was essential to adequately simulate flow in this region of high snowfall. Snow melt parameters were not transferable between the two adjacent watersheds, however, and suggested a significant impact of forest canopy and slope direction on snow-pack temperature. Despite different seasonal patterns of monthly stream flow, the similarity in annual runoff efficiency between the two watersheds suggests a similar evapotranspiration demand and geologic structure in the adjacent watersheds, so that the watersheds produce a similar quantity of streamflow through different pathways.
Figure 6. Seasonal Pattern of Monthly Average Stream Flow During a 16-Year Period From 1950 to 1965 in the Forested (4035000) and Wetland/Lake Dominated (4033000) Watersheds. Error bars denote standard deviation above the mean. An asterisk denotes a significant difference of monthly stream flow between watersheds at P = 0.01 level (Wu and Johnston, in press).
To put the Ontonagon Basin within a larger geographic context, long-term stream flow records (1956–1988) from 32 USGS gauging stations within the Great Lakes Basin were analyzed using multivariate statistical techniques (Johnston and Shmagin, in preparation). Factor analysis and cluster analysis of average annual flow revealed six patterns of river runoff within six distinct regions of the basin. Streams represented by pattern 1, including the Middle Branch Ontonagon River, occurred mostly in the upper peninsula of Michigan. This region exhibited relatively constant flow through the 1960s and 1970s, but decreasing flow during the 1980s. Streams represented by pattern 2 occurred mostly in New York State; these streams had pronounced cycles of low flow in the early 1960s and high flow in the 1970s, returning to average during the 1980s. Streams represented by pattern 3 occurred mostly in the lower peninsula of Michigan and exhibited a continuous increase in flow over the 30-year time period. The remaining three patterns distinguished watersheds in Ohio, Minnesota, and Wisconsin, respectively. These historical results imply that the different flow regions of the Great Lakes Basin will respond differently to future climate change.
Significance of Accomplishments
We have thoroughly addressed each of our original five objectives and detailed the complex interactions among land use and land cover, climate, DOM, and UVR damage to aquatic foodwebs. The streams in the northern Great Lakes region typically have very high concentrations of colored DOM, which is an effective attenuator of UVR. Canopy cover also effectively reduces UVR exposure to streams. Consequently, we conclude that the biota in the streams in this region likely experience minimal UVR exposure, except in streams that are shallow, have low DOM, and have a relatively open canopy. However, even moderate amounts of DOM greatly reduce UVR exposure to stream biota.
While UVR exposure appears to be of minor significance to stream biota in the northern Great Lakes region, DOM is extremely important in structuring stream foodwebs and in ecosystem services such as productivity. Besides its effect on the attenuation of visible light and UVR, DOM acts as a carbon and nutrient subsidy. At a landscape scale, DOM concentration and chemistry are controlled by a complex set of factors that include vegetation, soil, and surficial geology, but our findings reinforce those of previous studies that wetlands are a particularly important landscape control over DOM. To our knowledge, this is the first study to demonstrate that different kinds of wetlands can have positive or negative effects on DOM concentration and properties. The direct effects of future climate change on stream biota will be enhanced to the extent that they affect the landscape factors that control the production, transport, and cycling of DOM. In particular, changes in wetland area and type and hydrology may be particularly important controls over future DOM dynamics in aquatic ecosystems.
Users of Data, Relevance to Science, and Management of Resources
Our results provide essential background information to resource managers, policy makers, and scientists on the relative importance of UVR on aquatic foodwebs in the northern Great Lakes region and in similar regions in the U.S. We also demonstrate the central importance of DOM in the structure and function of stream ecosystems in this region, and land use, land cover, hydrological, and climate controls over DOM concentration and chemistry. We have primarily disseminated our results through frequent presentations at national meetings and in the peer-reviewed literature. We have already published 14 peer-reviewed publications from this project and will publish 5 to 10 additional publications from the data gathered as part of this project.
Journal Articles on this Report : 13 Displayed | Download in RIS Format
Other project views: | All 56 publications | 14 publications in selected types | All 13 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Docherty KM, Young KC, Maurice PA, Bridgham SD. Dissolved organic matter concentration and quality influences upon structure and function of freshwater microbial communities. Microbial Ecology 2006;52(3):378-388. |
R829643 (Final) |
Exit |
|
Frost PC, Larson JH, Kinsman LE, Lamberti GA, Bridgham SD. Attenuation of ultraviolet radiation in streams of northern Michigan. Journal of the North American Benthological Society 2005;24(2):246-255. |
R829643 (2003) R829643 (2004) R829643 (Final) |
Exit Exit |
|
Frost PC, Mack A, Larson JH, Bridgham SD, Lamberti GA. Environmental controls of UV-B radiation in forested streams of northern Michigan. Photochemistry and Photobiology 2006;82(3):781-786. |
R829643 (Final) |
Exit |
|
Frost PC, Larson JH, Johnston CA, Young KC, Maurice PA, Lamberti GA, Bridgham SD. Landscape predictors of stream dissolved organic matter concentration and physicochemistry in a Lake Superior river watershed. Aquatic Sciences – Research Across Boundaries 2006;68(1):40-51. |
R829643 (2003) R829643 (2004) R829643 (Final) |
Exit |
|
Frost PC, Cherrier CT, Larson JH, Bridgham S, Lamberti GA. Effects of dissolved organic matter and ultraviolet radiation on the accrual, stoichiometry, and algal taxonomy of stream periphyton. Freshwater Biology 2007;52(2):319-330. |
R829643 (Final) |
Exit Exit |
|
Johnston CA, Shmagin BA, Frost PC, Cherrier C, Larson JH, Lamberti GA, Bridgham SD. Wetland types and wetland maps differ in ability to predict dissolved organic carbon concentrations in streams. Science of the Total Environment 2008;404(2-3):326-334. |
R829643 (Final) |
Exit Exit Exit |
|
Larson JH, Frost PC, Lodge DM, Lamberti GA. Photodegradation of dissolved organic matter in forested streams of the northern Great Lakes region. Journal of the North American Benthological Society 2007;26(3):416-425. |
R829643 (Final) |
Exit Exit |
|
Larson JH, Frost PC, Zheng Z, Johnston CA, Bridgham SD, Lodge DM, Lamberti GA. Effects of upstream lakes on dissolved organic matter in streams. Limnology and Oceanography 2007;52(1):60-69. |
R829643 (Final) |
Exit Exit |
|
Larson JH, Frost PC, Lamberti GA. Variable toxicity of ionic liquid-forming chemicals to Lemna minor and the influence of dissolved organic matter. Environmental Toxicology and Chemistry 2008;27(3):676-681. |
R829643 (Final) |
|
|
Wu K, Johnston CA. Hydrologic response to climatic variability in a Great Lakes Watershed: a case study with the SWAT model. Journal of Hydrology 2007;337(1-2):187-199. |
R829643 (Final) |
Exit Exit Exit |
|
Wu K, Johnston CA. Hydrologic comparison between a forested and a wetland/lake dominated watershed using SWAT. Hydrological Processes 2008;22(10)1431-1442. |
R829643 (Final) |
Exit |
|
Young KC, Maurice PA, Docherty KM, Bridgham SD. Bacterial degradation of dissolved organic matter from two northern Michigan streams. Geomicrobiology Journal 2004;21(8):521-528. |
R829643 (2003) R829643 (Final) |
Exit Exit |
|
Young KC, Docherty KM, Maurice PA, Bridgham SD. Degradation of surface-water dissolved organic matter: influences of DOM chemical characteristics and microbial populations. Hydrobiologia 2005;539(1):1-11. |
R829643 (2003) R829643 (2004) R829643 (Final) |
Exit |
Supplemental Keywords:
water, watersheds, groundwater, land, soil, sediments, global climate, ecological effects, organism, stressor, organics, ecosystem, scaling, terrestrial, aquatic, environmental chemistry, biology, ecology, hydrology, geology, limnology, monitoring, surveys, Great Lakes, Midwest, Michigan, MI, EPA Region 5, RFA, Scientific Discipline, Air, Geographic Area, Water, Hydrology, Water & Watershed, climate change, State, Atmospheric Sciences, Ecological Risk Assessment, EPA Region, Watersheds, water resources, dissolved organic matter, anthropogenic processes, wetlands, environmental monitoring, global change, regional hydrologic vulnerability, aquatic food web, hydrologic models, climate models, UV radiation, vulnerability assessment, aquatic ecosystems, watershed sustainablility, Lake Superior, water quality, land and water resources, Region 5, aquatic ecology, climate variability, Global Climate Change, land use, vegetation models, ecological researchProgress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.