Final Report: Riparian Reforestation in an Urbanizing Watershed: Effects of Upland Conditions on Instream Ecological Benefits

EPA Grant Number: R825798
Title: Riparian Reforestation in an Urbanizing Watershed: Effects of Upland Conditions on Instream Ecological Benefits
Investigators: Hession, W. C. , Charles, Donald F. , Hart, D. D. , Horwitz, R. J. , Kreeger, D. A. , Newbold, J. Denis , Pizzuto, J. E. , Velinsky, D. J.
Institution: Academy of Natural Sciences , University of Delaware
EPA Project Officer: Manty, Dale
Project Period: June 1, 1998 through May 31, 2001
Project Amount: $837,685
RFA: Ecosystem Restoration (1997) RFA Text |  Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Hazardous Waste/Remediation , Land and Waste Management , Ecosystems


The overall objective of this research project was to determine if the ecological benefits of streamside forests persists in watersheds with urban land cover and, if not, at what level of urbanization the benefit becomes insignificant. The specific objectives of this research project were to determine: (1) how adjacent riparian forests affect the structure and function of low-order stream ecosystems; (2) how urban development within watersheds affects the structure and function of low-order stream ecosystems; and (3) whether the effects of riparian forests change among streams with different levels of urban development within their watersheds. The restoration of riparian forests has become a focus of watershed initiatives to improve degraded stream ecosystems. However, in urban and urbanizing watersheds, the ecological benefits of riparian restoration may be diminished because of widespread hydrologic and biogeochemical disturbance within the watershed. Without a better understanding of how riparian forests benefit streams in urban watersheds, there is a danger of wasting effort on techniques that are inappropriate to local situations.

A general hypothesis of the possible relationships between stream ecosystem integrity, the extent of urban development within watersheds, and riparian vegetation for low-order streams is shown in Figure 1. The curve labeled "Streams Without Riparian Forest" indicates a potential pattern of decreasing ecosystem integrity with increasing urban development. The upper curve, associated with "Streams With Riparian Forest," illustrates the beneficial effects of riparian forests on various structural and functional ecosystem attributes. The two curves merge at a high level of urban development, suggesting that when upland or contributing watershed disturbance reaches a certain level, riparian forest has a limited influence on stream ecosystems. The maximum potential benefits that could be achieved through riparian restoration are given by the vertical distance between the curves at a given level of urban development.

We conducted an interdisciplinary research project designed to evaluate the influence of riparian forest buffers on stream ecosystems in urban versus nonurban watersheds.

Figure 1. A Hypothetical Relationship Between Stream Ecosystem Integrity, Riparian Vegetation, and the Extent of Urban Development Within Watersheds

Summary/Accomplishments (Outputs/Outcomes):

Study Area and Sampling Design

The study area was located in the Piedmont of southeastern Pennsylvania and northern Delaware, with broad, rolling hills in the Piedmont Uplands and less steep, rolling hills in the Piedmont Lowlands. The region supports a fragmented, mixed-hardwood deciduous forest with large areas cleared for agricultural and urban development. This area encompasses the northern and western areas of the Philadelphia metropolitan region. Precipitation averages about 110 cm per year.

Our sampling design was based on direct comparisons of important ecosystem attributes in forested and nonforested sections of 12 streams, with different amounts of urban development within their watersheds. Each of these sites is comprised of nearly contiguous forested and nonforested stream reaches, with no major tributary inputs or other disturbances between reaches. These 12 sites are referred to as paired-reach sites. The paired-reach design represented a major strength of our analysis because watershed characteristics were virtually identical for the two riparian types within each stream. The sampling design also included sampling of forested sections of four streams in highly urban watersheds. These sites are referred to as unpaired forested sites.

A total of 18 sites (streams) were selected (12 paired-reach sites, 4 unpaired forested sites, and 2 restored sites). Physiographic and stream size differences among study sites were minimized by including only first through third order streams on the Piedmont Plateau, north and west of Philadelphia, PA. Sites influenced by dams, bridges, ponds, wastewater discharges, cattle grazing, construction projects, and channel/floodplain modification also were not included in the study. Reaches approximately were 20 channel-widths in length. Forested reaches had mature, relatively undisturbed riparian forest of 30 m or greater in width along both banks. Non-forested reaches consisted of grasses or small herbaceous vegetation of 30 m or greater lateral to the stream on both banks.

The ecological condition of each study reach was assessed based on an interdisciplinary sampling of channel morphology, bank erosion, nutrient biogeochemistry, benthic algae, benthic macroinvertebrates, fish, and food web linkages.

Watershed metrics describing the percent of forested, agricultural, urban/suburban, and impervious cover within study watersheds were calculated for 16 sites using data developed by the Delaware Valley Regional Planning Commission. These data are derived from 1-meter resolution aerial photography acquired during the spring of 1995. Two study sites were outside of the region covered by this data, and similar metrics were calculated using Landsat TM satellite imagery (MRLC NLCD). Watersheds ranged from 1-16 km2 in size and about 1-66 percent impervious surfaces. All but 3 of the 18 study watersheds have 30 percent or greater impervious surface. Therefore, the sites sampled represent watersheds with a relatively high level of urban development.

Riparian Vegetation

Riparian vegetation metrics were calculated at each study reach, including tree canopy density percent, shrub canopy density percent, herb percent cover, light intensity, tree species richness, species diversity, mean tree diameter, and the prevalence of exotic species. Each study reach was measured and used to describe the species composition and robustness of riparian trees, shrubs, and herbaceous vegetation at each site. Key findings included:

(1) Forested reaches in nonurban watersheds were reasonably similar in most vegetation metrics;

(2) Nonforested reaches differed widely in vegetative character; and

(3) Tree canopy density and herb percent cover were both reduced at urban locations compared with nonurban locations.

Channel Morphology

The channel morphology analysis combined data from this study with 14 paired-reach sites from a study recently conducted in the same geographic area using similar methods. Twenty-six paired stream reaches with drainage basins of 0.39-50 km2 in southeastern Pennsylvania, northern Maryland, and Delaware were examined. We measured channel morphologic characteristics, including bankfull width, cross-sectional area, bankfull depth, sinuosity, slope, and median bed particle size (d50) of all study reaches during base-flow periods between 1997 and 1999. Key findings included:

(1) Channels with forested riparian zones are wider than channels with nonforested riparian zones;

(2) Channels in urbanized watersheds are wider than channels in nonurbanized watersheds; and

(3) The effect of riparian vegetation is independent of the level of urbanization. In other words, riparian vegetation exerts a strong influence on channel morphology regardless of the level of urbanization in the watershed.

Physical Habitat

In-stream physical habitat at each study site was assessed by mapping and characterizing basic habitat types and reach-scale characteristics, including the depth of pools, riffles, runs, substrate size, substrate embeddedness, bed compaction, the presence of leaf litter and large woody debris, undercut banks, and rootwad features. Key findings included:

(1) Forested reaches have significantly greater volumes of woody debris per unit length, undercut rootwads along streambanks per unit length, and volumes of leaf litter per unit streambed area;

(2) Urban development negatively was correlated with the percent leaf litter per unit streambed area; and

(3) The variations in habitat indices were not impacted by the level of urbanization because of riparian vegetation.

Streambank Erosion/Bedload

Studies were conducted to determine rates of bank erosion at cutbanks, measure rates of deposition on point bars in laterally migrating forested/nonforested pairs, estimate the height, stiffness, and density of grassy vegetation on accreting point bar surfaces, determine rates of bed material transport in a single forested/nonforested pair, and assess how rates of sediment storage and production vary between forested and nonforested reaches. Key findings included:

(1) Bed material transport rates were higher in the forested reach of one paired-reach site where studies occurred. We estimated that during a 10-year period, bed material transport in the forested reach exceeded transport in the nonforested (grassy) reach by a factor of 3;

(2) Point bars in the nonforested reach significantly were larger than those in the forested reach. This suggests that point bars in the nonforested reach serve as locations of sand and gravel storage. Results suggest that deposition rates are higher in nonforested reaches because dense grassy vegetation traps sediment and increases the resistance of deposited sediment to erosion;

(3) Bank erosion rates were lower in forested reaches because tree roots resist bank erosion. Differences in the texture of the soils in the two reaches were not significant, suggesting that soil texture was not an important control on bank erosion rates in our study reaches; and

(4) Data on bank erosion rates, combined with estimates of bed material transport rates, suggest that rates of floodplain storage and reworking significantly are higher in nonforested reaches than in forested reaches. Bed material composed of sand and gravel tends to be stored in point bars in grassy reaches. Silt and clay sized sediment carried in suspension mostly is transported through both reaches.

Nutrient Biogeochemistry

Water samples were collected during baseflow at the upstream and downstream end of each study reach via multiple grab samples taken throughout the year (spring, summer, and fall). Key findings included:

(1) Concentrations of different forms of nitrogen and phosphorus substantially varied with most reaches dominated by nitrate (22 to 5,800 µg N/L) and soluble reactive phosphorus (SRP) (<0.5 to 136 µg P/L);

(2) All reaches were potentially limited by the level of available P (as SRP), given the high N to P ratios (>200); and

(3) There was no clear trend between nutrient concentrations and the amount of impervious cover in watersheds. Local riparian conditions (forested versus nonforested reaches) exerted a strong influence on the bioavailable forms of nitrogen (ammonium) and phosphorus (SRP). This most likely was a function of the increased algal activity and biomass in nonforested reaches.

Benthic Algae. Algal samples were collected from four substrate types in each reach: sand/silt, gravel, rock, and wood/sticks. Chlorophyll a and ash-free dry mass were determined for each sample, and 300 diatom valves per sample were identified to the lowest taxonomic level and counted using U.S. Geological Survey (USGS) National Water Quality Assessment Program (NAWQA) protocols. Key findings included:

(1) The amount of algal biomass in nonforested reaches significantly was greater than in forested reaches. Riparian cover (forested/nonforested) did not affect the species composition of diatom assemblages;

(2) The percent of impervious cover in watersheds strongly was correlated with differences in diatom assemblages, but not with concentrations of algal biomass; and

(3) The number of diatom taxa in samples correlates most strongly with watershed size and probably reflects greater habitat diversity. It also correlates positively with percent forest, and negatively with percent urban and percent impervious cover.

Benthic Macroinvertebrates

Three Surber samples for benthic macroinvertebrates were collected at each reach during February 1998 and 1999, and subsamples of approximately 200 organisms were identified to the lowest practical taxonomic level. Statistical analyses included comparisons of biological metrics, biomass, and general taxonomic similarity. Key findings included:

(1) All macroinvertebrate metrics significantly were correlated with the amount of watershed impervious cover. The direction of response was consistent with reduced biological integrity;

(2) Most macroinvertebrate metrics showed no significant difference in the response to impervious cover for forested and nonforested reaches; and

(3) Two measures of community dissimilarity between forested and nonforested reaches each indicated significant relationships. These relationships no longer were significant when sites with low levels of impervious cover were excluded from the model, indicating that most of the differences between forested and nonforested communities occur at low levels of watershed imperviousness.


Fish were collected at each reach during early fall 1998 and 1999 by depletion (removal) sampling of a blocked section (the sampling reach) within each study reach. Two to four sampling passes were made within each reach, using electroshocking gear appropriate to the stream size. Four types of biological measures were used to evaluate the effects of riparian vegetation and urban development on fishes: numerical density of species, biomass density of species, relative abundance of species, and metrics. Key findings included:

(1) Forested and nonforested reaches showed similar community characteristics, with greater abundance of a few species in nonforested reaches. The lack of a clear negative effect in the nonforested reaches may reflect: (a) relatively short length of nonforested reaches, so that these receive some effects from forested upstream areas; (b) presence of intact vegetation, even in the nonforested reaches, in particular, none of the sites were regularly grazed by livestock; and (c) greater difficulty in detecting effects on more sensitive species, which are found in fewer pairs of reaches; and

(2) There was a strong urban gradient in fish communities. Major effects were seen at low levels of urbanization (loss of most sensitive species) and at most urban sites, where omnivorous, tolerant species become common. A suite of native and widespread introduced species is present across much of the urban gradient. There were few riparian-urban interactions which may reflect: (a) relatively weak riparian effects for many measures; and (b) difficulty of this design for detecting some interactions, e.g., relating to intermittency, and extirpation of fish in very small streams.

Food Web

The structure of the aquatic food web was evaluated at study sites by measuring the carbon and nitrogen stable isotopic compositions of various producer and consumer groups. The transfer and sources of C to various consumer groups were followed by comparing the ratios of stable isotopes of C and N in the tissues of consumers (i.e., macroinvertebrates and fish) to those in their foods (suspended particulate material, flocculated sediment, benthic algae, and leaf litter). Key findings included:

(1) There were significant differences (p< 0.001) between the delta13Cperiphyton in forested versus nonforested reaches, with forested reaches generally more depleted in 13C (i.e., difference (delta13C) = forested minus nonforested averaged of circa -2 percent). This difference was slightly more pronounced in the more urban reaches;

(2) The N isotopic composition of fish increased with increasing urban area. The mechanisms of these changes need to be determined to better understand the impact of urbanization on stream ecosystems and the trophic structure; and

(3) Results from the different study sites indicated a complex interaction between food sources and consumers.

Food Quality

We measured the proximate biochemical composition (protein, lipid, and carbohydrate) of various forms of organic material that can serve as nutritional resources for primary consumers. These samples included suspended particulate material, flocculated sediment, benthic algae, and leaf litter. We also examined the biochemical composition of various fish and macroinvertebrate consumers. Key findings included:

(1) Biochemical makeup did not differ appreciably between paired forested and nonforested reaches. However, preliminary analysis indicates that important differences in the nutritional quality of key biota may exist between urban and nonurban streams; and

(2) Benthic microalgae collected from urban landscapes contained lower levels of protein, lipid, and carbohydrate (i.e., were less nutritious to herbivores) than algae collected from nonurban streams.


We conducted an interdisciplinary research project to evaluate the benefits of riparian reforestation in watersheds along a rural to urban gradient. The results of this research will help restoration planners decide if riparian reforestation would be adequate to improve aquatic ecosystem health, or if additional watershed-level restoration activities might be needed on a watershed-by-watershed basis.

We hypothesized that riparian reforestation improves aquatic ecosystem health, but as a watershed becomes more urbanized, this benefit will diminish until, at some level of urbanization, the benefits of riparian reforestation would be insignificant (see Figure 1). Across all the interdisciplinary aspects of the research and the numerous ecological attributes measured, there was no consistent or general trend that supports or rejects the hypothesis presented in Figure 1. However, among individual attributes, there were some clear research results that supported or rejected the null hypothesis (see individual results summarized above). Overall, our research indicates that the combined influence of riparian forests and watershed-level urban development is complex and variable among different ecosystem attributes.

This research resulted in many important research findings. However, some riparian and/or urbanization effects may not have been detected in this study because of the high amount of urban development in the study watersheds. The percent total impervious cover in study watersheds ranged from about 1-66 percent, with only 3 of the 18 sites below 25 percent. Therefore, the majority of sites were well beyond the 10-25 percent impervious cover threshold where significant stream impairment has been documented.

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

Other project views: All 33 publications 2 publications in selected types All 2 journal articles
Type Citation Project Document Sources
Journal Article Hession WC, Pizzuto JE, Johnson TE, Horwitz RJ. Influence of bank vegetation on channel morphology in rural and urban watersheds. Geology 2003;31(2):147-150. R825798 (Final)
  • Abstract: GSA Abstract
  • Supplemental Keywords:

    streams, ecology, restoration, urbanization, aquatic ecosystem restoration, chemical mixtures, ecological effects, human health, ecological indicators, ecological risk assessment, ecosystems, ecosystem protection, ecosystem assessment, indicators, forestry, hydrology, restoration, watersheds, exploratory research, environmental biology, riparian ecosystem, conservation, ecological benefits, ecological exposure, ecological recovery, environmental assets, environmentally stable landscape, forest, forested watershed, public policy, reforestation, regulations, riparian, riparian reforestation, risk assessment, upland conditions, water quality., RFA, Scientific Discipline, Water, Ecosystem Protection/Environmental Exposure & Risk, Water & Watershed, Ecosystem/Assessment/Indicators, Ecosystem Protection, Restoration, Forestry, Ecological Effects - Environmental Exposure & Risk, Ecological Risk Assessment, Ecology and Ecosystems, Aquatic Ecosystem Restoration, Watersheds, Ecological Indicators, ecological exposure, risk assessment, riparian reforestation, forest, streams, watershed, reforestation, Riparian ecosystem, environmental assets, upland conditions, conservation, ecological recovery, riparian, aquatic ecosystems, economic, urbanizing coastal watershed, water quality, forested watershed, ecological benefits, environmentally stable landscape, public policy

    Relevant Websites: Exit

    Progress and Final Reports:

    Original Abstract
    1999 Progress Report
    2000 Progress Report
    2001 Progress Report