Grantee Research Project Results

2003 Progress Report: Assessing the Interactive Effects of Landscape, Climate, and UV Radiation on River Ecosystems: Modeling Transparency to UVR and the Response of Biota

EPA Grant Number: R829642
Title: Assessing the Interactive Effects of Landscape, Climate, and UV Radiation on River Ecosystems: Modeling Transparency to UVR and the Response of Biota
Investigators: Morris, Donald P. , Williamson, Craig E. , Pazzaglia, Frank J. , Weisman, Richard N. , Hargreaves, Bruce R.
Institution: Lehigh University
EPA Project Officer: Packard, Benjamin H
Project Period: July 30, 2002 through July 29, 2006
Project Period Covered by this Report: July 30, 2003 through July 29, 2004
Project Amount: $825,850
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:

Climate influences the transparency of lotic ecosystems through rain-mediated transfer of ultraviolet (UV)-attenuating substances from watershed to water and solar radiation-mediated photochemical reactions of some of these substances. Land cover also affects the transfer of water along with UV-attenuating substances into streams and rivers. Forests control erosion of sediments; wetlands release dissolved organic matter; and human development of land for agriculture, roads, and buildings tends to increase storm runoff at the expense of groundwater recharge. Two factors control the biotic effects of UV radiation (UVR) in aquatic ecosystems: exposure to UVR and physiological resistance mechanisms.

The objective of this research project is to determine how current watershed and river properties (including land use and land cover) interact with climate and solar radiation to determine current UV exposure and how living organisms have adapted to survive this UV exposure. We also will establish the tolerance or susceptibility of macroinvertebrates to UVR exposure and how this may be modified by temperature and oxygen. Only by understanding present interactions under a range of conditions can we hope to predict the response of aquatic ecosystems to future change, including anticipated increases in extreme weather conditions and increases in ultraviolet-B (UV-B) radiation associated with ozone destruction in the stratosphere.

Progress Summary:

This document provides a summary of our progress through the first full year of field studies. Progress in six main project areas is described. Each of these projects is coordinated by one or more of the principal investigators and includes the work of graduate students and undergraduate interns.. These project areas include: (1) fluvial geomorphology; (2) investigations of small streams and watersheds and the influence of riparian canopy in determining incident UVR; (3) susceptibility of macroinvertebrates to UVR; and (4) photolability of chromophoric dissolved organic carbon (DOC) and its possible influence on water column transparency.

Geomorphology Project

Geomorphic research in this study is dedicated to describing and quantifying those physical and hydrological characteristics of watersheds that contribute to river discharge characteristics and water quality. A core hypothesis being tested is that land use plays an important, if not the most important, role in stream discharge and water quality. Our research is ever mindful of trying to construct the physical framework for understand UV transparency in the fluvial environment and its effect on stream ecology as a function of colored dissolved organic matter (CDOM). A picture of channel forms within the Lehigh Watershed emerged from geomorphic research in 2002-2003. For 2003-2004, research migrated towards quantifying stream response to variable discharge as a means of understanding how the various channel forms were shaped.

Experimental Design. Five fourth-order drainage basins, four of which are located in the Lehigh Watershed, have been selected for our analysis. The four drainages within the Lehigh Watershed are the Jordan Creek, Little Lehigh Creek, Saucony Creek, and Saucon Creek. All of these basins have variable amounts of anthropogenic disturbances and all have experienced some degree of land use change practices from agriculture to urbanization. Of these, Saucony Creek has experienced the least amount of urban growth, and Little Lehigh Creek has experienced the greatest urban growth. Saucon Creek, Saucony Creek, and Little Lehigh Creek are basins underlain primarily by carbonates with headwaters rising into the crystalline rocks of South Mountain. The Jordan Creek, in contrast, is underlain almost exclusively by shale and slate and, therefore, provides a measure of control on the effect of variable rock type on basin hydrology and channel form. Standing Stone Creek, located in the Ridge and Valley of central Pennsylvania, is a basin that has never experienced agriculture, and only experienced a limited amount of logging more than a century ago. It stands as a measure of control for the channels of undisturbed basins.

Similar data collection methods were applied for all channels studied. Stream sections, approximately 75 to 90 m in length, were chosen so that they were not artificially constrained or altered by straightening or dredging. Centralized base stations were identified to minimize the need to relocate surveying instruments (i.e., make foresights) by selecting sites where the entire reach could be surveyed from one or two locations. The sections also were chosen along stretches where vegetation was not prohibitive to surveying. Channels lined with grass, trees, agricultural crops, shrubs approximately 2 m or less in height, or a mixture of the above, were not obstructions to surveying. This arbitrary selection based on vegetation and ease of surveying introduces the possibility of a biased collection of sites that are not representative of the rest of the stream. The possibility of skewness in the data, however, will be tested by randomly selecting additional sites along the surveyed streams and comparing those results to the initial data set.

A Topcon Total Station GTS-211D was used after site selection to collect the geodetic data. Nondifferential corrected Universal Transverse Mercator coordinates of the surveying station were collected with a Garmin GPS 12 unit, following standard surveying procedures. The Topcon Total Station GTS-211D was oriented to geographic north using waypoints on the Garmin GPS 12 unit. The number of survey points taken varied, but averaged around 400 measurements per site. Completed surveys were uploaded and converted into a map in the field to ensure that each contained sufficient spatial coverage before the Topcorn Total Station GTS-211D was disassembled. The raw data points were converted into an XYZ grid and turned into a map using Golden Software Surfer (v. 6.03). Nine surveys were completed in the Saucon Creek Watershed, five on Little Lehigh Creek, and three on Jordan Creek.

The relationship between stream discharge and upstream drainage area is investigated through the use of pressure transducers installed at various points of a stream, with the streams selected to test the effects of decreased agriculture and increased urbanization. The transducers used for this project are Leveloggers, manufactured by Solinst Inc., and when installed in the bed of a stream, record both the hydrostatic and atmospheric pressures. After correcting for the changes in atmospheric pressure, the record of hydrostatic pressure can be converted into changes in water depth. A rating curve will be developed at each site to calculate discharge from stage height, but the Leveloggers will be installed first because this relationship can be determined while the pressure sensors are collecting data. The already determined stage then can be used to back-calculate discharge once the rating curve is developed.

Data. Research in 2003-2004 established an anomalous pool-riffle spacing pattern for streams in the Lehigh Watershed. Typically, adjacent pools are spaced 5 to 7 times the width of the channel. In the Lehigh Watershed, the surveyed channel showed an irregular pool-riffle spacing that averaged more than 10 times the width of the channel. There is a general sense from qualitative observations that stream widths have been increasing over the past few decades as farming practices are replaced by increased urbanization. The effect of this land use change is to reduce the supply of fines to the channel, while increasing the peak of the flood discharges. Both the reduction of fines and the increase in peak discharges have led to unstable channel banks and channel widening. If channel widening is occurring, then the pool-riffle spacing is even more out of equilibrium with expected channel widths than previously assumed.

In this context, most of the summer and fall of 2004 has been devoted to establishing the quantitative basis of channel width and pool-riffle spacing results. Pressure sensors deployed at the surveyed channel cross sections experienced one of the wettest summers and early falls on record. As such, robust rating curves have been constructed. The results, although preliminary, clearly illustrate the difficulty in applying an accepted erosion law when, in fact, the channel shaping discharges clearly do not scale with the drainage area This nonlinear scaling behavior also may partially explain the anomalous channel metric characteristics such as pool-riffle spacing. At this time, we do not have an explanation as to why the nonlinearity of the peak flows for both Saucon Creek and Little Lehigh Creek have different signs. Saucon Creek discharge scales as a function of the square root of the area, where as the Lehigh Creek discharge scales as a square. Given the similarity in the basin geology and shape, it is possible that the quantitative fingerprint of different levels of urbanization is captured in these Q-A data.

Watershed Dissolved Organic Carbon (DOC) Modeling and Stream Canopy Modeling

One objective was to refine and test models for variation in DOC concentration and quality across the watershed to see if specific conductance, land cover, and other terrain features could predict variations in UV attenuation underwater. The approach was to construct a pair of automated sampling systems that could collect baseline samples daily and be triggered automatically to collect frequent samples during storm events.

A pair of nearby headwater streams was selected to compare a primarily forested area with one in which the land cover was primarily farmland. ISCO 6712C samplers were equipped with batteries, rain gages, and stream stage sensors in weather-shielding enclosures and deployed along with YSI DataSondes that measured stream water quality parameters at the same interval (15 minutes) as the stream stage sensors. Funnels and plastic bottles were deployed at open and forested regions at each site to collect rain and canopy throughfall to measure the relative contribution of the canopy to stream DOC. By comparing baseflow and storm events and by analyzing DOC, CDOM, and ionic composition in addition to routine parameters (pH, temperature, specific conductance, turbidity), we anticipate that we can validate a multicompartment model to explain changes in water quality based on hydrologic flowpath and differences in land cover.

Another objective was to refine measurements of UV transmittance and riparian canopy properties along small streams in the watershed to model UV penetration of the canopy from GIS and remote sensing data. We evaluated the previous year’s data and determined several sources of error in the photographic images (narrow field of view, instead of hemispherical, in canopy photographs and automatic exposure left important parts of the higher canopy overexposed so these parts were counted as sky). We also determined that diffuse versus direct irradiance could be measured by a simple shading technique and the added information would allow better modeling of our results. We modified our digital camera system by building a self-powered (rechargeable battery + interface circuit) USB extension cable to allow the camera to be used mid-stream while the recording notebook computer stayed on the stream bank. We revisited a number of sites that were used the previous year and developed several sites on campus for testing our protocols. We also calibrated a UV sensor that is permanently mounted on campus to provide a local resource when our portable UV monitors are not making measurements. As we continue this project, we will characterize seasonal changes in canopy during leaf drop and leaf growth, and variations caused by differences in tree species and age as well as by variations correlated with stream order.

Data. We established the importance of wetland area and forest cover in controlling DOC concentration. We were surprised to find that stream DOC was largely derived from algal production in regions where carbonates were more concentrated, but we cannot resolve the causal factor (stream shading versus chemistry). Specific conductance was a good predictor of DOC concentration at any given site because rain and groundwater contribute to different degrees to streamflow over time and also are different in typical DOC concentration. Specific conductance also was highly correlated with variations in the source of DOC (soil versus algal) for reasons that are still unclear.

Canopy photographs in the first year using a less-than-hemispherical field of view were correlated with UV measurements made at the same time. Diffuse light appeared to be important in regulating UV transmittance. Small order streams appeared to have lower UV transmittance through the canopy, but the results appeared to be site-specific and not easily extrapolated to other areas.

We succeeded in recording stream data for a number of small storms and baseflow conditions with our paired automated stream samplers. We also recorded the remnants of hurricane Ivan passing through our region and the resulting major flooding. High water moved one of our stations slightly and left it muddy, but nothing at these sites was damaged and all samples and data were safe. At another stream monitoring site where we were logging stream stage, we were not so fortunate, and lost a small data logger to the 50-year flood. Comparing storm flow over a range of precipitation, we can confirm the general impression from last year that DOC and CDOM vary inversely with specific conductance, but we now can add that throughfall DOC from the canopy makes a major contribution during rain events.

Our canopy optics investigation confirmed the need to manually adjust hemispherical photographs because the highest part of the canopy is usually much brighter than the average canopy and may be lumped with sky when computing transmittance. Our preliminary analysis suggests that this brightness is not a source of UV and thus should be excluded from the sky view fraction that we correlated with UV transmittance. Also, we have confirmed by measurements of diffuse and direct irradiance in the open and under the canopy that diffuse light is the dominant contributer to UV penetration and must be measured explicitly. Typically, the UV-B wavelengths under a clear sky are 60 to80 percent diffuse, while the UV-A wavelengths are closer to the visible waveband (photochemically active radiation [PAR]) and range from 50 to20 percent diffuse. We have begun to gather and analyze remote sensing and GIS data that we expect will contribute to a model for UV transmittance to the stream surface across our watershed.

Influence of UVR on Macroinvertebrates

The study of the effects of UVR on lotic invertebrates has had two components during the past year. The first component examined in situ attenuation of UVR with a BIC submersible profiling radiometer. The idea was to both characterize the underwater light regimes to which macroinvertebrates are exposed, and simultaneously examine particulate and dissolved absorbance to test optical models being used by others in the project. Profiles of UVR and PAR were taken in all of the standard sampling sites on the tributaries and main stem of the upper and lower reaches of the Lehigh River. Water samples also were collected and brought back to the laboratory for processing by Bruce Hargreaves. The UVR profile data were archived on the server.

The second component of the project focused on the behavioral response of stream macroinvertebrates to changes in exposure to UVR and visible light. This subproject involved a series of in situ experiments looking at the behavioral response of macroinvertebrates to stream reaches with different riparian vegetation and hence solar shading regimes. This work is being done in the context of land-use changes and the possible effects these changes may have on stream macroinvertebrate communities.

Site Description. The experimental reach is located within a section of Little Lehigh Creek, a tributary to the Lehigh River. This section of the stream is composed of two distinct sites. One site is shaded by thick forest canopy cover, created by dense riparian vegetation that blocks much of the incoming UVR and PAR. The second site is exposed to sunlight, with only limited streamside vegetation, allowing both UVR and PAR to reach the water surface. The differences in the two adjacent sites were chosen to signify a change in land use, which leads to a loss of streamside vegetation.

Preliminary Experiments. A set of preliminary experiments was run during the summer of 2004 and continued in the fall. These experiments were established to determine whether the experimental design would provide adequate results and whether the constructed, artificial habitat would withstand stream conditions. The artificial habitat design of this project evolved from mesocosm designs found throughout scientific literature, which focused primarily on predation and nutrient limitation responses by macroinvertebrates. The artificial habitat is composed of a plastic tray filled with rocks of a similar size and number and an anchoring slab of slate. The tray is connected to the slate slab using self-locking, plastic zip ties. Twelve of these habitats were constructed and placed in the stream to control for vast amounts of heterogeneity within stream beds.

The first preliminary experiment placed six artificial habitats in two sites along the stream: shaded (forested) and nonshaded (open). After a 1-week incubation period, the contents of the trays were removed from the stream and analyzed in the laboratory for macroinvertebrate colonization (total abundance) and biodiversity (species richness, Shannon-Weaver index, and evenness).

The second preliminary experiment followed the same timeline as the first, except for an additional incubation period. For this period, all 12 habitats were placed in the open site for 1 week to establish a uniform algal substrate on the rocks.

Data. Data for the preliminary experiments were collected on three occasions: June 23, 2004, June 30, 2004, and July 27, 2004. There was a significantly higher number of total macroinvertebrates in the open site compared to the forested site for all three dates. There was a higher number of species in the open site on June 23. There also was a higher Shannon-Weaver Index for the open site on June 23 and for the forested site on July 27. The forested site also had significantly more evenness than the open site on June 30 and July 27.

There seems to be a distinct shift in community structure and taxa composition between the open site and the forested site, which are separated by only an estimated 25 m stretch of stream. Results have been analyzed only to the major group level at this point. The categories include: ephemeroptera (mayflies), chironomidae (midges), plecoptera (stoneflies), hydracarina (water mites), megaloptera (dobsonflies), oligochaeta (aquatic worms), trichoptera (caddisflies), coleoptera (beetles), tipulidae (craneflies), and other diptera (flies). On June 23, mayflies contributed the greatest numbers to both sites, with stoneflies and chironimids second in abundance. The forested site had lower numbers of megalopterans oligochaets and tipulids, while in the open site, only water mites were observed. On June 30, mayflies again dominated both sites, with caddisflies and stoneflies being second in abundance in the forested site, and chironimids and stoneflies being second in abundance in the open site. On July 27, mayflies dominated the forested site with six other groups present in lower abundance, while in the open site, chironimids dominated with mayflies in secondary abundance and several other taxa at low abundances.

Current and future experiments are designed to separate two possible variables that may influence these results: behavioral avoidance and food limitation. It is believed that if macroinvertebrates are not food limited, either through nutrient or PAR limitation, they will selectively avoid habitats that expose them to higher levels of UVR. Land development and land-use change can alter the optical qualities of water through both a loss in streamside vegetation and changes in the amount of DOC entered into a system. This project is focusing on the former of the two processes.

Photobleaching Potential for CDOM of the Lehigh River

CDOM accounts for a large proportion of the absorption of UVR in the water column of both lakes and rivers. Photobleaching alters the optical properties of CDOM and may be an influential instream process regulating UVR transparency. The main purpose of this study is to: (1) show how river conditions and environmental factors such as rainfall and discharge affect the photobleaching potential; and (2) demonstrate potential optical changes in the river resulting from photobleaching. Photobleaching specifically refers to the degradation of the chromophoric structural groups in the DOC that are responsible for the characteristic color as well as the absorption of UVR. The degree of photobleaching is typically assessed by evaluating optical changes in the DOC that reduce the color and UVR absorbance of a particular sample of water. For the purposes of this study, photobleaching is determined by the loss of absorbance of a sample. Photobleaching also is reflected in the alteration in other optical parameters such as spectral slope and molar absorptivity. Both of these parameters have been used as indices of photobleaching.

Water samples were placed in three 1-cm quartz test tubes. The samples were analyzed before and after a 48-hour exposure to a UV lamp system (Q Panel 340). Initial and final measurements included: temperature, dissolved oxygen (DO), pH, DOC, fluorescence (total and a ratio of emission at 450 and 500 nm), and absorbance (200-800 nm).

Data. Photobleaching caused significant changes in all the variables measured in this study (see Table 1). The greatest changes during the 48-hour incubations were observed in total fluorescence and absorbance (at 320 nm), with a 57 percent and 37 percent decrease, respectively. The DOC-specific absorbance (-30%) and the UV-B spectral slope (+25%), indices of DOC quality, also changed substantially. A regression analysis indicates that the change in UV-B is highly correlated to K_b (r² = 0.69, p ≤ 0.001), suggesting that photobleaching does not decrease in absorbance uniformly across the UV-B spectrum. Small but statistically significant changes also were observed in pH (-2%), DO (-8%), and DOC (-10%).

The values of K_b (at 320 nm) varied substantially over the course of the study. The average value of K_b was 0.0108, but significant seasonal variations were observed (see Figure 1 and Table 2). Average values of K_b were lowest in the spring (0.0094) and highest in the fall (0.0122). The lower spring values could be related to snowmelt, whereas the reoccurring high fall values could be related to leaf drop.

Univariate regression analyses were performed between K_b and other parameters to determine which factors are important in regulating photobleaching in the Lehigh River (see Table 3). A significant suite of variables related to DOC concentration and quality were each inversely correlated with K_b (p ≤ 0.001) and explained 17 to51 percent of the variation in K_b. The fluorescence ratio (51%), UV-B spectral slope (44%), and DOC-specific absorbance (30%), possible indicators of DOC source and prior photobleaching, explained the greatest amount of variance. The chemical parameters (conductivity, pH, and alkalinity) were each significantly correlated with K_b (p ≤ 0.001) and explained only 19 to37 percent of the variation. Lastly, the environmental variables (discharge and precipitation) were inversely correlated with K_b, and explained about 20 percent of the variation. These variables do not directly influence K_b but are probably important in regulating the quantity and quality (source) of CDOM as well as the ionic composition of the river, which may directly influence photobleaching rates.

To explain more of the variance in K_b, a multivariate regression was developed using four types of models: (1) models using only environmental variables (discharge and precipitation); (2) models using only chemical variables (pH, conductance, alkalinity, and DOC); (3) models using only optical variables (absorption coefficients, spectral slope, DOC-specific absorbance, and fluorescence ratio); or (4) models using both optical and chemical variables (see Table 4). The best model based on only environmental variables explained 18 percent of the variance (p ≤ 0.001). Models using only chemical variables explained 36 percent of the variance (p ≤ 0.001). Models using only optical variables had better results, explaining 54 percent of the variance (p ≤ 0.001). The model that included both the optical and chemical variables, however, performed the best, explaining 75 percent of the variance.

Table 1. A Summary of Chemical and Optical Changes After Photobleaching. Each variable had a sample size of n = 72.

Figure 1. Photobleaching Rate Constant (k +/- s.e.) With Date

Table 3. A Summary of the Correlation Analysis Between Various Parameters and K_b. The r-values are reported as well as the significance, which is denoted as * = p < 0.05, **

Table 4. A Summary of Multivariate Models Run Between K_b and a Variety of Environmental Variables

Future Activities:

We will devote the final year of geomorphic research to documenting changes in channel form and function over the past few decades. This will be accomplished through a study of historic air photos from which channel width can be determined. Prior results provide the basis for interpreting channel width in terms of the prevailing basin hydrology shaping the channels.

The remaining work on the small watershed automated sampling will involve comparisons after leaves are down, and then during the leaf-out phase in the spring. More replication of throughfall measurements will need to be added to our protocol. The remaining work on the basic canopy optics model is to repeat measurements without leaves and use our refined photographic technique to correlate these with UV measurements across time-of-day and clear version overcast sky conditions to determine whether a more complicated model of direct light (and stream orientation) will be needed.

One ongoing experiment and one future experiment are planned at this time for the study of the influence of UVR on macroinvertebrates. The ongoing experiment is a reciprocal transplant experiment, which will place six artificial habitats in each of the two sites (forested and open) for 1 week. Following the incubation period, four habitats from each site are switched. After 24 hours, all of the habitats are collected and analyzed in the laboratory.

The future experiment, planned for spring and summer 2005, is a manipulative experiment compared to the natural experiments previously run. This experiment will place all 12 artificial habitats in the open site. The incoming UVR will be manipulated using acrylic covers. Six habitats will be covered with OP-2 acrylic and six habitats will be covered with OP-4 acrylic. Both OP-2 and OP-4 acrylic permit the transmittance of PAR wavelengths (400-700 nm). When looking at UVR (200-400 nm), however, OP-2 acrylic completely absorbs these wavelengths, whereas OP-4 acrylic allows a general increase of transmittance throughout this portion of the UVR spectrum. Therefore, this experiment will manipulate incoming UVR without decreasing the availability of PAR for primary production.

In combination with colonization sampling, several other variables will be measured on each sampling date including water temperature, conductivity, pH, and periphyton chlorophyll content.

Journal Articles:

No journal articles submitted with this report: View all 37 publications for this project

Supplemental Keywords:

scaling, multiple stressors, biotic resistance, air, geographic area, water, EPA region, ecological risk assessment, ecology and ecosystems, hydrology, state, water and watershed, wet weather flows, climate change, global climate change, UV radiation, aquatic ecology, aquatic ecosystems, aquatic food web, climate models, climate variability, dissolved organic matter, ecological research, global change, hydrologic dynamics, hydrologic models, land and water resources, land management, land use, regional hydrologic vulnerability, solar radiation, urban runoff, vegetation models, vulnerability assessment, water resources, watershed sustainability,, RFA, Scientific Discipline, Air, Geographic Area, Water, Water & Watershed, climate change, State, Environmental Monitoring, Wet Weather Flows, Ecological Risk Assessment, EPA Region, Watersheds, water resources, dissolved organic matter, wetlands, hydrologic dynamics, global change, regional hydrologic vulnerability, aquatic food web, urban runoff, hydrologic models, climate models, UV radiation, hydrology, vulnerability assessment, aquatic ecosystems, watershed sustainablility, Lake Superior, solar radiation, water quality, land and water resources, Region 5, aquatic ecology, stormwater runoff, climate variability, Global Climate Change, land use, vegetation models, ecological research, Michigan (MI), land management

Progress and Final Reports:

Original Abstract

2004

2005

Final Report