Integrating GIS with Benthic Metrics: Calibrating a Biotic .../67531/metadc4425/m2/1/high_res_d/thesis.pdf · Earnest, Steven F. P., Integrating GIS with Benthic Metrics: Calibrating

INTEGRATING GIS WITH BENTHIC METRICS: CALIBRATING A BIOTIC INDEX TO

EFFECTIVELY DISCRIMINATE STREAM IMPACTS IN URBAN AREAS OF THE

BLACKLAND PRAIRIE ECO-REGION

Steven F. P. Earnest, B.S.

Thesis Prepared for the Degree of

MASTER OF SCIENCE IN APPLIED GEOGRAPHY

UNIVERSITY OF NORTH TEXAS

December 2003

APPROVED: James H. Kennedy, Major Professor Minhe Ji, Minor Professor Donald Lyons, Committee Member and Geography

Graduate Advisor Reid Ferring, Chair of Department of Geography

Sandra L. Terrell, Interim Dean of the Robert B. Toulouse School of Graduate Studies.

Earnest, Steven F. P., Integrating GIS with Benthic Metrics: Calibrating a Biotic Index to

Effectively Discriminate Stream Impacts in Urban Areas of the Blackland Prairie Eco-region. Master of Science (Applied Geography), December 2003, 45 pages, 3 tables, 17 illustrations, references, 34 titles.

Rapid Bioassessment Protocols integrate a suite of community, population, and

functional metrics, determined from the collection of benthic macroinvertebrates or fish, into a

single assessment. This study was conducted in Dallas County Texas, an area located in the

blackland prairie eco-region that is semi-arid and densely populated. The objectives of this

research were to identify reference streams and propose a set of metrics that are best able to

discriminate between differences in community structure due to natural variability from those

caused by changes in water quality due to watershed impacts. Using geographic information

systems, a total of nine watersheds, each representing a different mix of land uses, were chosen

for evaluation. A total of 30 metrics commonly used in RBP protocols were calculated. Efficacy

of these metrics to distinguish change was determined using several statistical techniques. Ten

metrics were used to classify study area watersheds according to stream quality. Many trends,

such as taxa presence along habitat quality gradients, were observed. These gradients coincided

with expected responses of stream communities to landscape and habitat variables.

Copyright 2003

by

Steven F. P. Earnest

ii

ACKNOWLEDGEMENTS

I would like to thank T. Bennett, C. Cortemeglia, B. Dunlap, T. Hardison, M. Kavanaugh,

H. Perry, and J. Sandberg for assistance in the field, as well as M. Ji, J. H. Kennedy, and D.

Lyons for academic support during the completion of this project. Also, special thanks go to

Clay Jones for granting access to the Mary L. Cooke National Wildlife Federation Preserve on

Red Oak Creek, Ellis Co.

iii

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS iii

LIST OF TABLES v

LIST OF ILLUSTRATIONS vi

INTRODUCTION 1

Policy Background 1 Monitoring Programs 4 Dallas Monitoring 5 Objectives and Hypothesis 11 Description of Study Area 12

MATERIALS AND METHODS 14

Geographic Information Systems 14 Reference Site Selection 14 Urban Site Selection 15 Physico-chemical Water Quality 16 Habitat Evaluation 17 Macroinvertebrate Field Sampling 18 Macroinvertebrate Lab Processing 18 Invertebrate Metrics 19 Benthic Index Calibration 20 B-IBI Metric Scoring Categories 20 Statistics & Biological Condition 21

RESULTS 22

GIS: Watershed Land Use 22 Physico-chemical Water Quality 23 Habitat Quality 24 Macroinvertebrate Metrics 26 Statistical Analysis 31

DISCUSSION 33

CONCLUSIONS 41

REFERENCES 43

iv

LIST OF TABLES

Page

1. Metrics included in the Benthic Index of Biotic Integrity (B-IBI) 19

2. Physico-chemical water quality measurements of study area streams 23

3. Regional B-IBI metric scoring categories and ranges 26

v

LIST OF ILLUSTRATIONS

Page

Figure 1 - Texas and the Dallas city limits 6

Figure 2 – Dallas area biomonitoring stations 2000 7

Figure 3 – Map of industrial land use concentrations in Dallas watersheds 8

Figure 4 – Map of single-family residential land use in Dallas watersheds 9

Figure 5 – Map of nitrate concentrations in Dallas watersheds 10

Figure 6 – Map of land use in Dallas watersheds 22

Figure 7 – Chart of Dallas watershed land use 23

Figure 8 – Total suspended solids of study area streams 24

Figure 9 – Map of habitat quality index (HQI) scores 25

Figure 10 – Map of Benthic Index of Biotic Integrity (B-IBI) scores 27

Figure 11 – Scatterplots of Shannon’s Index, taxa richness & EPT v. B-IBI 28

Figure 12 – Scatterplots of dominance and Diptera v. B-IBI 29

Figure 13 – Scatterplots of predators & midge taxa v. B-IBI 30

Figure 14 – % of midges as sub-families and tribes 31

Figure 15 – Canonical Correspondance Analysis joint-plot 32

Figure 16 – Scatterplot of B-IBI v. Chironomidae abundance ratio 38

Figure 17 – Scatterplot of B-IBI v. vacant land use 40

vi

INTRODUCTION

The United States Environmental Protection Agency’s Water Quality Act states specific

goals for maintaining the health of US waters. As stated, “the objective of this act is to restore

and maintain the chemical, physical, and biological integrity of the Nation’s waters” [PL 92-500,

Clean Water Act, §101(a)]. In accordance with Phase-I of the National Pollutant Discharge

Elimination System (NPDES) contained within this Act, municipalities with populations over

100,000 must obtain storm water discharge permits. As of 2003, Phase-II requires smaller

population centers to join this effort.

Historically, Clean Water Act amendments have progressed from an industrial discharge

permitting system to municipal monitoring of storm water discharge. Surface water quality often

suffers from high urban use including: industrial and construction site runoff, high-density

residential fertilizer application and vast impervious surfaces that alter landscape hydrology

(Hynes 1970). Urbanization often increases flow and suspended solids in streams, contributes

fecal-coliform bacteria, increases nutrient levels (which often promote algal blooms), and

degrades natural aesthetics (Mitton, 1997; TWDB, 1997). These non-point source pollution

(NPSP) pressures often change a water body’s ability to support diverse populations of aquatic

(and terrestrial) organisms. Since most water pollution generated in urban environments is NPSP,

in the form of land use residue, increased streambed scouring and bank instability, problem

sources are typically difficult to trace.

The state of Texas has established four categories of water usage in order to guide

monitoring standards and protect water use. These categories include: aquatic life use, contact

recreation, public water supply and fish consumption. Specified within each of these categories

are descriptors of several pollutants including: toxic metals, bacteria, toxic organics such as

1

pesticides, and dissolved solids; all of which are present in urban environments. In order to

contend with potentially dangerous consequences of pollutant loading in urban streams, "…the

State of Texas must develop action plans to remediate or protect water bodies... these watershed

action plans include a quantitative plan that identifies responsible parties and specifies actions

needed to restore and protect a water body." (TCEQ 1999).

A number of regional, state and federal agencies, including the North Central Texas

Council of Governments (NCTCOG), Texas Commission on Environmental Quality (TCEQ,

formerly TNRCC), Texas Water Development Board, and the EPA, are currently developing and

evaluating tools for monitoring impacts to watersheds in urban areas. Others have pursued

refinement of existing runoff models and continue to search for solutions to non-point source

pollution. (Camp 1993).

Cities are required to monitor surface runoff and evaluate its impact on water quality.

NPSP, by definition, cannot be traced to any single source, and consequently is difficult to

approach with any single solution. Some comprehensive municipal programs are showing

progress, however. Municipalities have tried to analyze the effects of land use on the quality of

surface rainfall runoff; including the potential effects of roadways, industrial sites, treatment

facilities, and other land uses. Such studies can be enhanced beyond traditional techniques using

spatial analysis tools such as geographic information systems (GIS).

A particular management task included in the NPDES permit application specifies that a

city shall: " Identify each watershed's boundary and categorize each watershed according to

urban or suburban conditions (i.e. rank the watersheds according to potential for development or

significant redevelopment). And, identify which watersheds in each category contribute storm

water runoff to the water supply of the city" (Dallas NPDES permit 1997). This leads directly to

2

an investigation of watershed (i.e. sub-basin) land use characteristics and illustrates a

commitment by local government to study all aspects of stream health.

Hoos (1996) discusses urban areas as major sources of NPSP and recommends that

information be collected on pollutant loads within individual watersheds. This implies that

objectives beyond simply maintaining a reactionary program environment are required for

progress. Unfortunately, the implementation of such policy can be costly and municipalities

often only report what is required to minimally comply with regulatory requirements.

Determining which urban land uses contribute to which types of NPSP, and to what

degree, could be highly beneficial to local monitoring efforts. Considering the magnitude of

landuse complexity found in large urban environments, GIS is an ideal technology for

deciphering and calculating the complex spatial relationships of urban hydrology (Haubner,

1996). Ventura et al (1993) describe the utility of GIS to calculate NPSP modeling input

parameters in Dodge County, Wisconsin, stating that, "By definition, non-point source pollution

is a problem associated with large geographic areas. To understand the causes and impacts of

non-point source pollution requires the integration and display of several types of geographic

information." In recent years, GIS has been used to interpret terrain characteristics and

hydrology, as well as pollutant loadings. Many urban municipalities, particularly those affected

by NPDES (Phases 1 & 2) are increasingly aware of GIS applications for these inquiries and will

likely incorporate them into overall pollution control efforts in the future.

Society uses streams, rivers, and lakes as preferred outdoor recreation opportunities as

well as for potable consumption and fisheries. Surface water bodies are only suitable for these

uses in a naturally balanced state that is minimally affected by human development. Often

referred to as biotic, or biological, integrity (BI) an ecosystem’s ability to recover from

3

perturbation, such as chemical input or hydrological damage, often serves as an experimental

reference for biological monitoring. These references conditions are imperative for successful

implementation of monitoring programs.

Rapid Bioassessment Protocols (RBP) are used throughout the United States by

municipal, state, and federal government agencies to assess the biological condition of streams.

RBP’s integrate a suite of community, population, and functional metrics based on benthic

macroinvertebrate or fish communities into a single assessment of health. These metrics may be

combined as a representative value of stream conditions, otherwise known as a multi-metric

index (Karr & Chu 1999). This combination of metrics is commonly known as an index of biotic

integrity (IBI) and has been developed to take various aquatic biota into account. The use of

benthic macroinvertebrates, as originally developed by Kerans and Karr (1994), is known as the

benthic index of biotic integrity, or B-IBI.

Water quality indices have been created to help determine the relative health of stream

systems. Fish assemblages, flora abundance, and benthic macroinvertebrate communities are

commonly assessed for this purpose. Within these indices, a variety of metrics have been created

to describe biotic communities. Some measures are based on invertebrate taxa-richness, others

mention primary production, and still others consider physical conditions of the water column.

To better understand the effects of urbanization on stream quality, it is necessary to

include measures of stream habitat, functional measures of invertebrate biota, and chemico-

physical parameters indicative of water quality. Of these, chemical inputs to streams are transient

and may only occur periodically. Ecological monitoring is often dominated by benthic

invertebrates due to a relative lack of habitat flexibility within certain taxa groups. benthic

invertebrate communities change according to the cumulative effects of chemical and habitat

4

change. Additionally, habitat structures remain intact unless benthic particle sizes are small

enough to be transported by the water column (urban flash flows commonly scour gravel and

sand beds). Unlike more mobile organisms, such as fish, invertebrates are unable to avoid habitat

destruction associated with urbanization. This has led many researchers to consider stream

invertebrates as good indicators of stream water quality (Barbour 1991, Somers 1998). These

fauna, when properly identified, can be counted and assessed as groups that represent long-term

trends in stream health.

Dallas Monitoring

Dallas’ first attempts at qualifying stream conditions began in 1971 in response to the

Clean Water Act. Twenty-nine years of stream quality data were collected to evaluate ambient

chemical and biological conditions of stream health in the Dallas area. Initially, protocols for

stream biomonitoring were not well established and the Dallas program evolved to reflect new

monitoring approaches over time. As a result, changes were made to accommodate fluctuations

in funding for the monitoring program. Most of the parameters measured were not continuous;

making it impossible to follow biological trends in water quality through time. As per

requirements of the EPA’s NPDES permitting cycle for urban streams, these data included a

series of chemico-physical parameter’s, such as pH, temperature and dissolved oxygen that

influence the health of local biotic communities.

Of the programs established within the City of Dallas, biological assessment

(bioassessment) monitoring has been used consistently, but with varied success. This lack of

success may stem from a limited understanding of stream basin processes, and use of insensitive

methods of stream evaluation.

5

Recently, however, several methods for investigating stream health integrity have been

devised in the city of Dallas’ monitoring program. These include: chemical testing of surface

runoff toxicity, biological assessment of stream communities, and aquatic habitat assessment of

stream corridors. Each of the above mentioned monitoring efforts often include the spatial

location of results using GIS or similar tools. This promotes effective communication of

biomonitoring efforts, increasing the likelihood of monitoring data being considered in urban

planning initiatives, and easing political stressors.

Initially, biomonitoring stations were placed in stream sub-basins based primarily on ease

of access. This method of site selection can misrepresent actual conditions of aquatic habitat

availability along a stream reach. In order to minimize this concern, Dallas’ biological

assessment team has conducted rapid bio-assessment evaluation across multiple stream reaches

within each sub-basin. Each stream is represented with three 33-meter reach sections placed at

various intervals within each sub-basin to reflect the 100-meter protocol. While this would seem

to best represent diversity of habitat, it can also increase resource demands for project

completion. Careful analysis of watershed characteristics and major program components could

simplify this obstruction and efficiently achieve monitoring objectives.

C i t y o f D a l l a s

Figure 1. City limits of Dallas, Dallas County, Texas

6

A total of 38 watersheds are currently monitored by Dallas' Storm Water Quality

Division and reported in the existing EPA-NPDES storm water discharge permit cycle. These

watersheds are referred to later in this paper as the Dallas Watershed Area (DWA). Stream

basins are classified as such because watershed drainage boundaries fall within, or cross, the city

limit. Currently, the city limits encompass approximately 340 square miles of the total 946

square miles contained within the DWA (Figure 2). As of Spring 2000, 116 biomonitoring sites

had been chosen to represent local conditions for stream monitoring purposes (The primary site

list was reduced to 38 sites in 2002 to optimize personnel resources for sample collection).

Figure 2. Dallas Biomonitoring stations as of 2001

Within the DWA’s 946 square miles, approximately 51 square miles are classified as

industrial, 262 are single-family residential, 405 are vacant, 5 are under construction and 23 are

surface water. The remaining 208 square miles are comprised of various other land uses such as

multi-family residential, commercial retail, and infrastructure support.

7

Though the DWA contains more than 600 square miles outside the city limits, sampling

locations have been primarily sited within the city boundaries in order to avoid jurisdictional

conflict and reduce operating costs. The following maps give a partial overview of watershed

land use concentrations in the DWA:

Industrial (Figure 3)- Industrial land uses generally consist of production facilities and

warehouse districts that are normally grouped into development corridors. These areas can have

significant cumulative impacts within watershed areas. Though EPA regulations have greatly

improved the environmental state of these areas, industrial land uses are still considered to have a

potentially high-impact on watershed ecosystems. In addition to highly impervious surface

ratios and unexpected toxic spills, permitted amounts of waste discharge may be exceeded. It is

for these reasons that industrial areas are a primary focus for the NPDES.

Figure 3. Industrial concentrations of the Dallas RWA.

8

Figure 4. Single family landuse concentrations of the Dallas watersheds.

Single-Family Residential (Figure 4) - Single-family uses can be considered to have a mid-range

impact potential due to increased instances of yard fertilizer application, as well as moderate

imperviousness to rainfall. These two combined provide an ideal opportunity for phosphate and

nitrate build-up in runoff to area streams.

In addition to land use, the distribution of watershed nutrient averages also reveals

patterns within the current EPA report zone. For example, some relationships could exist

between these averages and upstream land use patterns. For the purposes of demonstration the

only physical parameter taken into account here is total nitrate concentration.

Nitrates (Figure 5): Nitrates have been directly linked to algal blooms resulting from

excess fertilizer application. These inputs, most commonly associated with agricultural land use,

are also found in urban areas of intense grounds-keeping and recreational landscaping. Excess

nutrients are known to contribute to increases of cyanobacteria (blue-green algae) (Hauer 1996).

9

Furthermore, lowered oxygen availability, due to elevated biochemical oxygen demand, is a

common effect of excess nutrient loads. Though nitrate abundance is not the only pollutant

parameter of concern to the city, it is known to be among the most influential on stream nutrient

processes (Connell 1997).

Figure 5. Nitrate concentrations of the Dallas Watersheds (2000).

Attempting to spatially link pollutants to specific land use types is why a GIS was used to

extract landuse patterns of the Dallas area. Current calculations of water quality scores are

derived from seven physical, chemical, and bacteriological measures, including: biochemical

oxygen demand, dissolved oxygen, fecal coliform, total phosphates, total nitrates, turbidity, and

pH (Dallas Bioassessment 1999). These primary indicators of water quality are the separate

scores derived from the averages of contaminants at monitoring sites in each watershed.

10

Bioassessment sites were initially established to represent land use types that occur

within each watershed area. Depending on the size of the basin, between two and twenty

collection points are chosen. At each of these sites, it is normally necessary to survey at least 100

meters of the stream course in order to collect a sufficient sample data set. Typically,

bioassessment data are collected in the winter season to avoid poor conditions that may arise

from higher temperatures in the summer; winter season monitoring helps to ensure optimal

oxygen availability and invertebrate diversity. For this study, summer collection methods, which

help define base-flow conditions, were used the help resolve these issues of seasonality (which

often require base parameters be established at least bi-annually).

Study Objectives

This research combines the utility of GIS analysis with biological techniques in stream

quality assessment procedures. By following previously established methods, an assessment of

protocols for rapid biomonitoring procedures (RBP’s), using benthic macroinvertebrates, is also

included. These procedures yield information on our ability to distinguish relationships between

stream biological condition and human induced perturbation. A reference condition was

established to provide a basis for calibration of biological metrics used to determine impairment

of streams in Dallas. These site conditions are thought to be the least disturbed by urban and

agricultural influences. Unfortunately, reference conditions for stream health evaluation in North

Texas are scarce. Other studies in the area have acknowledged difficulty with finding suitable

reference sites for evaluating urban influence (Davis 1998). Identifying least impacted reference

streams and selecting appropriate biotic metrics for analysis was a primary focus of this study.

11

Specific objectives of this study were:

◙ To describe “best-available” reference conditions for regional streams that represent natural

background conditions.

◙ To examine existing benthic metrics for watershed scale projects and determine which of

those metrics are suitable for semi-arid applications.

◙ To combine applicable biological metrics with GIS methods that enhance analysis

procedures for stream quality assessment.

◙ To evaluate aggregates of metrics (the B-IBI) for its ability to discriminate stream impacts in

urban areas of the blackland prairie eco-region.

Attending to the above goals produced the following hypotheses:

Ha1: The benthic index of biotic integrity is capable of discerning human impacts from

naturally variable conditions in urban areas of North Texas.

Ha0: The benthic index of biotic integrity is not capable of discerning human impacts

from naturally variable conditions in urban areas of North Texas.

Study Area

The city of Dallas covers portions of Collin, Dallas, Denton, and Rockwall Counties.

Dallas County’s natural eco-region, blackland prairie, is dissected by the broad, shallow and

gently rolling Trinity River basin. This area contains narrow riparian woodlands bordered by

wheat and soybean fields; cattle pasture is common. Much of the area is well drained but often

contains high proportions of finely textured soils. Though nearly all of the area is farmland,

urban development is rapidly increasing. The city’s estimated population is 1.2 million residents;

Dallas/Fort Worth’s metropolitan area population presently stands around 5.5 million. Over

100,000 new residents have moved to the region each year over the past seven years (NCTCOG

12

2003). Average annual rainfall is 94 cm (37 inches), while the mean annual temperature is 19

degrees Celsius (67 degrees F) (NOAA 2003).

13

MATERIALS AND METHODS

Geographic Information Systems

Study area watershed layers were created from United States Geological Survey (USGS)

digital elevation model (DEM) data using ESRI Arc/Info (Environmental Systems Research

Institute, Inc.) hydrologic analysis tools. Numerous data sets (streams, roads, etc.) were acquired

from the city of Dallas Storm Water Quality Division’s GIS and Environmental sections. Spatial

land use information was taken from existing 1995 land use data obtained from the North Central

Texas Council of Governments (NCTCOG). Steps involved in the use of GIS were:

Each sample site was recorded as ground coordinate locations using GPS.

Resultant points were then overlayed onto the elevation model for use as pour-points in the

watershed delineation process.

Watershed boundaries were then used to clip land use information for each basin.

Total land use area was calculated from polygon features representing level-3 land use

categories. Summaries of land use occurrence were then calculated to summarize land use

percentages within each watershed.

B-IBI results were then calculated for each watershed and averaged to illustrate benthic

community health within each basin.

Reference Stream Selection

The process of evaluating stream quality within the study area started with choosing

stream sites that illustrate varying degrees and types of perturbation. We began by finding the

“best-available” reference conditions. This was to identify streams with minimal agricultural and

urban influences. Though these reference sites cannot reflect pristine conditions, they are

14

characterized as having the least impacted catchments. Candidate reference streams were chosen

by criteria that assume:

- Close geographic proximity to urban areas (i.e. possessing similar habitat potential as

test streams)

- Minimal human development within surrounding watersheds.

- Watersheds are spatially isolated from urban study site areas, so there is no drainage

interaction

- Accessibility

Using GIS as a source of watershed and landuse information, numerous candidate

streams were investigated in and around the Dallas county area. Four sites were designated as

potential references due to their abundance of macroinvertebrate habitat availability. These were

located outside the urban fringes south of Dallas; each contained within the north central Texas

blackland prairie eco-region. Reference watersheds included portions of Ten Mile Creek

(southern Dallas Co) and Red Oak Creek (northern Ellis Co). With this inclusion of multiple

stream sites a multi-habitat composite reference condition was compiled. These streams were

sampled at two locations each, intentionally overlapping drainage basins to ensure diversity of

habitats sampled within the same stream system. After initial habitat evaluation, however, one of

these sites was dropped because it was impacted by agriculture and lacked riparian buffer

vegetation. This site, on Ten Mile Creek, was dropped from the final analysis.

Urban Watershed Selection

Study sites were identified based on distributions of common land uses. Predominant

uses in the area include residential housing (both single and multi-family), commercial retail,

15

industrial, office space, parks, and vacant space. Urban streams were selected in accordance with

the following criteria:

- Close geographic proximity to reference streams

- Those whose watersheds represented a unique combination of human activities

- Those whose watersheds are spatially isolated from one another, so there is no

drainage interaction

- Accessibility

A total of six streams were selected for analysis. In Dallas proper these included Ash

Creek, Dixon Branch, Floyd Branch, and Turtle Creek. Cottonwood Creek and Fish Creek, each

originating just west of Dallas in Grand Prairie, were also selected.

Physico-chemical Water Quality

At each sampling site the following physico-chemical parameters were measured:

dissolved oxygen, suspended sediment concentration, conductivity, hardness, alkalinity,

temperature, acidity, and nitrate (NO3) concentrations. Of these, oxygen, conductivity, acidity,

and temperature were measured on location. Water was transported to the lab where the

remaining tests were conducted.

Dissolved oxygen was measured using a YSI model 50B electronic meter. Conductivity

was measured using a YSI model 3300 conductivity meter. A portable Orion 250A pH meter

(and triode sensor) was used to record pH. Hardness, alkalinity, suspended solids, and nitrate

measurements were performed within 24 hours of water collection.

Total suspended solids (TSS) was measured from one-liter water samples vacuumed

through pre-fired (500o C for 30 minutes), pre-weighed, glass fiber filters (ASTM 2000). These

samples were taken from the surface of undisturbed water along each reach.

16

A LaMotte nitrate concentration (NO3) kit was employed, in conjunction with a Milton-

Roy Co. Spectronic 1201 photospectrometer at 543nm across a one-centimeter cuvette, to

measure total nitrates at each study site.

Habitat Evaluation

For the purpose of minimizing weather-induced variability no stream location was

sampled until at least seven days after any rain event of greater than 25 mm (0.10 inches). After

this time, if rain was not anticipated, field sampling commenced across the study area.

Habitat was evaluated at each sampling station using criteria developed by EPA and

TCEQ (TNRCC 1999). Five transects were delineated at each stream site. Across each

streambed, wetted area and bank-full width was measured using a 50-meter field measuring tape.

Flow was measured using a Marsh-McBirney Flowmate 2000 portable flowmeter at pre-

determined intervals, one-tenth the stream width, across each wetted stream section. These

intervals were offset by one-half of an interval to yield ten total measurements across the stream.

The only exception to this rule occurred in smaller streams that did not exceed two meters width

at a given transect; these were divided into fewer sections. Total discharge was then calculated

from these data.

Additional stream reach measurements included bank slope, erosion potential, specific

habitat abundance, algae or macrophyte occurrence, dominant substrate type, riparian vegetation,

tree canopy, and available instream cover (snags, cobble, gravel, etc.). Each of these were

assessed across a two-meter width on each side of each transect. An example of the field data

sheet used to record these variables in the field is provided in the appendix (Appendix Figure 1).

Habitat abundance data, such as percent of in-stream cover, average bank erosion and bank

stability, was compiled along TCEQ habitat quality index (HQI) guidelines. Photograph’s of

17

transects and habitat features were also taken at each sampling station. These metrics were then

used to score each habitat metric category for comparison to, and synthesis with, invertebrate

community distributions.

Macroinvertebrate Field Sampling

Three replicate invertebrate kick-samples were collected in each stream. Kick samples,

using D-framed nets with a 1200-micron mesh, were collected in proportion to the abundance of

each habitat type found on site. These totaled a maximum of ten one-minute kicks measuring 0.3

x 1.0 meters. Sampling always proceeded upstream to reduce substrate disturbance between

samples. Each replicate’s kick samples were combined into one of three 40-liter (10-gallon)

water buckets and separated from inorganic, and large organic debris, prior to bottling in one-

liter containers. Composite samples were labeled and preserved using 70% ethanol.

Macroinvertebrate Laboratory Processing

In the laboratory each kick-sample was placed into an 810cm2 (126 in2) porcelain pan

divided into ten sections (11.5 x 7 cm; 4.5 x 2.8 inches). A random number was generated, by

rolling two dice, to determine from which section the invertebrate sub-sample would be taken.

Numbers were generated in this fashion by rejecting a value of 12 and considering a value of 11

as the first square (i.e. values 2-11 were used to generate 10 unique numbers). If the resultant

sub-sample clearly contained greater then 100 organisms (+- 20%) it was further sub-divided

into quarter samples under the microscope. Each sample was picked until this requisite number

of organisms was achieved. All organisms were then stored in 70 percent ethanol for later

identification. The remainder of the original samples were returned to the sampling bottles and

set aside for storage.

18

Chironomidae larvae from each sample replicate were slide mounted using CMC-9

mounting media (Epler 2001). When possible, invertebrates collected during this study were

identified to genus using Merritt & Cummins (1996), Wiggins (1995), or Wiederholm (1989).

Invertebrate Metrics

Table 1 lists the invertebrate metrics that were evaluated for their ability to detect

changes in water quality.

Table 1: Candidate invertebrate metrics and expected direction of metric response to perturbation (compiled from Barbour et al 1996 and Karr & Chu 1999).

Category Metric Definition Expected response to perturbation

Richness measures # of taxa Overall variety of invertebrate assemblage Decrease

# of EPT taxa Number of taxa in the order Ephemeroptera, Plecoptera, and Trichoptera Decrease

# Coleoptera taxa Number of beetle taxa Decrease # Chironomidae taxa Number of chironomids Decrease

# Orthocladiinae taxa Number of chironomids as orthocladiinae Decrease

# Tanytarsini taxa Number of chironomids as tanytarsini Decrease

# Crustacea + Mollusca taxa Number of calcium-dependent taxa Decrease

Composition measures Shannon-Weiner Index Richness and evenness in a measure of Diversity and composition Decrease

% Dominant taxon Dominance of most abundant taxon Increase % Pyrallidae Percent aquatic Moths Decrease % Odonata Percent dragonfly nymphs Increase % Ephemeroptera Percent mayfly nymphs Decrease % Trichoptera Percent caddisfly nymphs Decrease % Plecoptera Percent stonefly nymphs Decrease % Coleoptera Percent beetles Decrease % Elmidae Percent of beetles as elmids Decrease % Diptera Percent of dipterans Increase % Tanyarsini to chironomids Percent chironomids as Tanytarsini Decrease

% Orthocladinae to chironomids Percent chironomids as Orthocladinae Increase

# Non-insect taxa Number of non-insect taxa Decrease

# Intolerant taxa Number of taxa not generally tolerant of perturbation Decrease

% tolerant Percent of taxa generally tolerant of perturbation Increase

Ratio_I/T Ratio of intolerant to Tolerant Taxa Decrease

Ratio_T/I Ratio of tolerant to Intolerant Taxa Increase

Hilsenhoff Biotic Index Measure of taxa sensitivity Decrease

19

Trophic measures % gatherers Percent of gatherer feeding group Variable % filterers Percent of filterer feeding group Decrease % predators Percent of predator feeding group Variable % scrapers Percent of scraper feeding group Decrease % shredders Percent of shredder feeding group Decrease

Benthic Index Calculation / Metric Selection

A simple variance routine was used to select benthic metrics for further analysis. Metric

sensitivity was based on the combined replicates from all reference streams (n = 9). The

coefficient of variation (C.V.) was used to evaluate the variability of candidate metrics. Metrics

with a C.V. of 30%, or less, were conditionally accepted for use in the B-IBI calculation. The

only exceptions were in cases of metric redundancy (i.e. high correlations, r2 > 0.7, between

measures of similar community metrics), or low counts (i.e. average organism counts less than

one). When the coefficient of variation (CV) of any one metric was greater than 30%, it was

rejected due to associated variability. Additionally, the Hilsenhoff Biotic Index was rejected due

to its tendency to incorporate taxa dominance as a measure of richness. Metrics selected were

used to calibrate the B-IBI for regional conditions.

B-IBI Metric Scoring Categories

When calculating each stream site’s index score, metric categories were created such that

higher values were given to those sites most similar to reference conditions. Each replicate

sample was placed into one of the three conservative scoring categories: 5 points for sites that

exceeded the 25th percentile in metrics that were expected to decrease with perturbation or were

below the 75th percentile for those expected to increase with disturbance; 3 points for sites that

fell between 25% and the minimum value in decreasing metrics, 75% and the maximum value in

increasing metrics; and 1 point for sites that did not exceed the minimum value of decreasing

metrics or exceeded the maximum value of increasing metrics.

20

Statistics & Biological Condition

Using the ten accepted metrics, each site’s B-IBI was scored, for each replicate, as the

sum of the ten metric scores. This created a possible B-IBI score range of 10-50. Total B-IBI

scores were then averaged across each site. After quadri-secting the total index range (10 – 50

pts), one of four categories of condition (excellent, good, poor, or very poor) were assigned to

the sample site watersheds for GIS display.

In addition to B-IBI calculation, multi-variate statistical procedures were also applied to

habitat and benthic data. These included Pearson’s coefficient of correlation, detrended

correspondence analysis (DCA), principle component analysis (PCA) and canonical correlation

analysis (CCA). Stream replicate counts of benthic taxa were analyzed with DCA to evaluate

similarity between and within sample sites.

Invertebrate sample counts were also analyzed with PCA to determine gradients of change of

benthic communities. A total of 57 parameters from three sets of environmental variables –

watershed land cover, stream habitat, and physico-chemical water quality - were used to build a

model with CCA to characterize taxa-habitat relationships.

21

RESULTS

GIS: Reference and Urban Watershed Land Use

Each urban basin contains varying concentrations of land use types while each reference

location contains marginally developed, highly buffered streams that represent the “best-

available” conditions (Figure 6).

Figure 6. Landuse map of study area watersheds selected for metric evaluation of B-IBI

Three land uses were found to be most dominant across the study area. These included

single-family residential (Ash Ck, 67%; Floyd Br, 34%; Turtle Ck, 61%), industrial (Dixon Br,

39%), and vacant space (Red Oak Ck, 80%; Ten Mile Ck, 58%; Fish Ck, 54%; Cottonwood Ck,

34%) (Figure 7).

22

Landuse Percentages of Study Watersheds

0.0%

20.0%

40.0%

60.0%

80.0%

100.0%

RED

1

RED

2

TEN

ASH

FLO

YD

CTN

WD

DIX

ON

TUR

TLE

FISH

Perc

ent o

f Tot

al A

rea

Industrial Multi-Family OfficeParks & Recreation Retail Single FamilyVacant

Figure 7. Primary land use composition of study area watersheds

Physico-Chemical Water Quality

Few notable differences were observed in basic water chemistry parameters among sites.

Stream temperatures ranged from 23.50C in Red Oak Creek to 330C in Cottonwood Creek.

Dissolved oxygen levels were 100% saturated at all sampling sites except Fish Creek (%60

saturation).

Table 2. Physico-chemical water quality of study area streams

Stream Temp (Co) pH DO

(mg/L) DO%

ALK (mg/L

CaCO3) HARD (mg CaCO3/L)

COND (micromhos)

NO3 (mg/L)

Red Oak1 23.5 8.4 6.5 101 200 408 420 3.188 Red Oak2 31.1 8.4 7.6 99 203 416 440 4.802 Ten Mile 23.3 8.16 6.45 90 190 200 347 0.456 Ash Ck 30.3 7 11.5 145 170 456 350 0.398 Floyd Br 31 8.2 10.5 150 115 236 530 7.684 Cottonwood Ck 33 8.14 6.8 95 122.5 208 510 0.287 Dixon Br 28 8.5 8.7 116 110 230 500 0.492 Turtle Ck 26.4 8.58 6.8 105 67.5 196 690 0.263 Fish Ck 27.3 7 4.65 59.5 192.5 272 540 0.355

23

Turtle and Dixon Creek, and Floyd Branch, exhibited the lowest alkalinity (at 68, 110,

115 mg/L CaCO3 respectively) while Red Oak Creek’s 200 mg/L CaCO3 was highest. Hardness

measurements ranged from Turtle Creek (lowest with 196 mg CaCO3 /L) to Ash Creek (highest

with 456 mg CaCO3 /L) which was similar to Red Oak Creek’s 416 mg CaCO3 /L. The only

anomalous conductivity was recorded in Turtle Creek (690 micromhos; other sites ranged from

350 – 550 micromhos). High total nitrates (NO3) were found on both Red Oak Creek and Floyd

Branch (4.8 & 7.7 mg/L respectively). Total suspended solids (TSS) were highest in Cottonwood

Creek at 29 mg/L (85% as inorganic solids) (Figure 8). Also with regard to TSS, organic solids

were highest in Ash Creek (36%) and Floyd Branch (29%) and lowest in Ten Mile Creek (6%).

Total Suspended Solids (TSS) as Seston or Inorganic Carbon

0

5

10

15

20

25

30

35

Red1 Red2 Ten Ash Floyd CtnWd Dixon Turtle Fish

TSS

mg/

L

Seston Inorganic

5.12.20.90.21.20.8 1.4 0.9 1.83.5

4.7

23.8

6.3 6

2.9 0.72.1 1.6

Figure 8. Observed amounts of total suspended solids represented as mg/L of organic seston and inorganic carbon.

Habitat Quality

Physical habitat varied across the study area. Stream flow volume, total number of

channel bends, percent of in-stream cover, total number of in-stream cover types, average bank

erosion potential, riparian vegetative buffer width, and tree canopy cover varied most. Red Oak

24

Creek’s base flow was recorded at 0.2 m3/s, while Ash Creek only discharged 0.007 m3/s. Stream

channel bends were most numerous in Red Oak Creek (with 4 moderately-defined bends) while

each urban stream only maintained 1-2 poorly-defined bends. Ash Creek exhibited the highest

percent of in-stream cover at 91%, while Floyd Branch only contained 36% cover. Total number

of in-stream cover types were least at Ash Creek (2), and highest in Red Oak and Ten Mile

Creek (each with 5). Dixon and Floyd Branch also maintained 5 cover types each. Average

stream bank erosion potential was clearly highest along Cottonwood Creek (75%), and lowest

along Floyd Branch (4%). Riparian vegetative buffer width was highest along each reference

stream (each greater than 100m) and lowest along Turtle and Cottonwood Creeks (3m & 5m,

respectively). Percent tree canopy cover ranged from Ten Mile Creek, Floyd Branch, and Turtle

Creek (68%, 72% & 75%, respectively) to Ash Creek’s 12% canopy. TCEQ’s habitat quality

index (HQI) reflects these habitat qualities as a combination of each physical parameter (Figure

10).

Figure 10. Results of habitat quality index calculations as applied to study area watersheds

25

Macroinvertebrate Metrics

Of the 30 invertebrate metrics evaluated 10 were summarized to create B-IBI scoring

categories (Table 3).

Table 3: Descriptive statistics for "best available" reference sites and scoring criteria for B-IBI metrics Summary Statistics Score Metric Min 25% Median 75% Max 5 3 1 # of taxa 19 27 27 30 33 >27 19 - 26 <19 # of EPT taxa 7 8 9 10 12 >8 7 <7 # Chironomidae taxa 4 7 8 9 11 >7 4 - 6 <6 Shannon-Weiner Index 3.708 4.136 4.303 4.38 4.515 >4.136 3.71 - 4.135 <3.71 % Dominant taxon 10.08 10.92 11.97 13.19 21.43 <13.19 13.19 - 21.43 >21.43 % Diptera 21.43 27.73 36.13 40.66 46.15 <40.66 40.66 - 46.15 >46.15 # Intolerant taxa 9 11 11 12 16 >11 9 - 10 <9 % tolerant 40.91 47.83 50 55.17 57.14 <55.17 55.17 - 57.14 >57.14 Ratio Int/Tol 53 81.82 104.25 109.43 137.21 >81 53-81 <53 % predators 16.33 17.09 18.49 20.88 31.87 >17.09 16.3 - 17.09 <16.3

All together, 1,468 midges and 1,342 other invertebrates were identified to genus. After

the B-IBI was compiled, values were assigned to their corresponding watersheds. Reference

streams reflected the highest B-IBI values (30-50) for the study area (Figure 10). Urban streams,

scored between 10-29.

26

Figure 10. Map of B-IBI quality score across the study area; Darker green shades denote higher quality. Yellow and orange symbols represent those watersheds reflecting poor or very poor stream quality.

These data were then evaluated with statistical techniques (Pearson’s correlation, PCA,

CCA) relating specific metrics to recorded habitat variables. B-IBI metrics that were expected to

increase or decrease with disturbance are reflected by scatter plots of metrics. Shannon’s

diversity index, taxa richness, and number of EPT taxa (r2 = 0.8392, 0.8626 & 0.7525,

respectively) clearly reflect an increase in B-IBI as a function of overall diversity (Figure 11).

27

a)

B-IBI Score v. Shannon's Index

r2 = 0.8392

-10

0

10

20

30

40

50

60

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Shannon's Index

B-IB

I Sco

re

b)

B-IBI v. Taxa Richness

r2 = 0.8626

0

10

20

30

40

50

60

0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0

# of Taxa

B-IB

I

c)

B-IBI Score v. Number of EPT Taxa

r2 = 0.7525

0

10

20

30

40

50

60

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0

# EPT taxa

B-IB

I Sco

re

Figure 11. Scatterplots of relationships of B-IBI scores versus Shannon’s Index (a), Taxa richness (b), and EPT taxa (c) for all replicate samples from Dallas area watersheds (n = 27).

Some metrics revealed negative relationships to B-IBI scores. This is consistent with

expected responses due to index calibration ranges of increasing or decreasing metrics. Of

28

particular note are %Diptera and %Dominance (r2 = -0.6305 & -0.5912, respectively) (Figure

12). Diptera included Tabanus sp, Tipula sp, Simulium sp, Ceratopogonidae, and various

Chironomidae.

a)

B-IBI v. Percent Community as Diptera

r2 = -0.6305

0

10

20

30

40

50

60

0.0 20.0 40.0 60.0 80.0 100.0

% Diptera

B-IB

I

b)

B-IBI v. Taxa Dominance

r2 = -0.5912

-10

0

10

20

30

40

50

60

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0

% Dominance

B-IB

I

Figure 12. Scatterplots of relationships of B-IBI scores versus the percent of stream community as Diptera (a), and taxa dominance (b) for all replicate samples from Dallas area watersheds (n = 27).

Some metrics, such as percent of community as predators (r2 = 0.1040) and total number

of midge taxa (r2 = 0.0572) did not directly account for B-IBI variation (Figure 13). Percent of

community as tolerant taxa (r2 = 0.3634) also reflected this characteristic.

29

a)

B-IBI v. Percent of Community as Predators

R2 = 0.104

0

10

20

30

40

50

60

0.0 10.0 20.0 30.0 40.0 50.0 60.0

% Predators

B-IB

I

b)

B-IBI v. Number of Midge Taxa

R2 = 0.0572

0

10

20

30

40

50

60

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0

# Chironomidae Taxa

B-IB

I

Figure 13. Scatterplots of relationships of B-IBI scores versus percent of community as predators (a), and total number of observed midge taxa (b) for all replicate samples from Dallas area watersheds (n = 27).

Variation in taxa composition of midge communities was observed throughout the study

area. When divided into four primary sub-family/tribe types – Tanypodinae, Orthocladiinae,

Chironomini, and Tanytarsini – considerable changes in dominance occurred among the urban

streams. Where reference streams showed greater diversity and evenness of the above-

mentioned taxa, Chironomini and Orthocladiinae were dominant in one or more of the urban

sub-basins (Figure 14). This occurrence also coincided with dominant functional feeding group

categories. In the case of Ash Creek, collector gatherers increased markedly due to the

dominance of Dicrotendipes sp. Floyd Branch showed similar dominance of the midge

30

community as a result of the prevalence of Orthocladiinae in the mostly bedrock, nutrient rich

stream reach.

Percent of Chironomidae Community as Sub-Family / Tribe

29.9 28.2 23.3

69.2

14.9

75.365.2 69.8

30.9

17.3 38.333.5

11.7

79.6

11.1

6.1

29.7

0.6

23.514.0 32.7

7.2

8.29.2

23.9

15.0

25.9 28.8 23.511.0 4.4 6.6

14.11.90.5 2.0

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

RED1RED2

TENMILE ASH

FLOYD

CTNWDDIXON

TURTLEFISH

% o

f Chi

rono

mid

ae

%Chironominae %Orthocladiinae %Tanypodinae %Tanytarsini

Figure 14. Percentage distributions of midge sub-families/tribes in study area streams (n=27). Urban sites clearly exhibit dominance of Chironomini or Orthocladiinae.

Statistical Analysis

Principle component and canonical correlation analysis illustrated gradients of taxa

occurrence occurring in the study area. The CCA triplot indicates the relative weight of habitat

influences on invertebrate taxa (Figure 15). This is shown by plotting taxa located at the

centroid of that taxa’s occurrence within the dispersion of environmental variables; represented

as arrows that reflect habitat variable gradients within that ordination space. Individual streams

can be described via the plotted site-id markers in relation to the metric points and habitat

arrows.

Major gradients were derived from habitat quality metrics such as stream bends, percent

in-stream cover, flow volume, percent vegetative bank cover, and stream buffer width to percent

of bank covered by grass, average bank slope, average erosion potential, and total percent

organic seston. Industrial landuse was aligned with the second ordination axis while vacant space

31

is clearly a strong component of the first axis. Individual stream sites fell within expected ranges

of the habitat variable gradients. PCA results reflected similar gradients of taxa occurrence

among sampling sites.

Figure 15. Canonical Correspondence Analysis of 57 environmental variables versus taxa distributions and sample site community composition. Arrows denote direction of change of each environmental variable along its corresponding ordination axis. Points are placed at the centroid of their occurrence within the ordination space.

32

DISCUSSION

Urban watersheds are a mosaic of land uses drained by constructed channels and sewers

designed to efficiently remove rainfall runoff from city areas. As a result of these modifications,

stream channels receive various inputs of NPSP; making watersheds ideally suited for GIS

analysis. TCEQ acknowledges difficulty in acquiring reference conditions for North Texas due

to scarcity of perennial, minimally impacted streams in the area (Davis 1998). Additionally,

employing agricultural use in lieu of more rural areas can lead to complications with reference

availability. Defining reference conditions in this fashion made it possible to test urban stream

health using corresponding reference scale measures. Inherent problems that may arise from this

approach are avoided by replicating these “best-available” habitat conditions for initial B-IBI

calibration.

Watersheds chosen for this study exhibit a wide range of land uses. This study does not

consider specific chemical toxicity of land uses to a stream. However, benthic data provided

great insight to the health of local aquatic systems effected by urban use.

Reference B-IBI conditions compiled from three of the “best available” stream sites in

the region were sufficient to discriminate stream impacts in urban watersheds. These reference

scales were created independent of urban streams by measuring the natural variability of each of

the 30 B-IBI metrics. This allowed for conservative estimates of regional reference conditions in

favor of urban stream B-IBI scores. Each of the urban streams assessed exhibited greatly reduced

B-IBI scores when compared to reference conditions.

Red Oak Creek maintains the highest quality HQI and B-IBI of these reference streams.

Though some agriculture does exist in the watershed, large areas of riparian vegetation coupled

with un-altered stream channels, provides these streams protection from large-scale runoff events

33

that would otherwise scour in-stream habitat. This portion of northern Ellis County is largely

vacant (i.e. some agricultural) space (80.4%) complimented by rural residential housing (15.8%).

Physico-chemical water quality was somewhat unique in this stream. Both hardness and nitrates

(412 mg CaCO3/L, 3.9 mg/L, respectively) were slightly elevated above urban stream sites. This

is presumably due to large areas of exposed limestone bedrock along the sampled stream reaches

and limited nutrient enrichment from surrounding pasture.

Ten Mile Creek, just north of Red Oak Creek in southern Dallas County, is more

developed. With 59% vacant area and 29% residential space, it still maintains considerably

higher HQI and B-IBI scores than any of the urban streams. This suggests that further research

may be required to evaluate specific land use thresholds that cause detectable changes in benthic

quality. Water quality here was consistent with urban streams with one exception. Conductivity

(347 micromhos) in Ten Mile Creek was 150 – 340 units less than any urban site.

Ash Creek is 64% single-family & 5% multi-family residential use. 10% of this

watershed is comprised of vacant areas, while only 6% is institutional and 5% industrial. No

physico-chemical water quality parameters were notably different from reference streams. This

suggests that habitat parameters, such as channel morphology and lack of in-stream cover types,

contributed most to its low B-IBI score.

Cottonwood Creek, whose sample site was located in a park area, had lower residential

use (26.6% single-family; 6.9% multi-family) but contained more industrial areas and vacant

space (12.6% and 34%, respectively). Total suspended solids were the much higher in this

stream (28.9 mg/L) than any other stream surveyed. This could be due to pedestrian traffic on

banks and in open grass fields of the surrounding park area, or frequent scouring of sediment

34

deposits. Habitat characteristics that contribute to reducing this parameter, primarily riparian

buffer vegetation and bank stability, were notably absent at his location.

Dixon Branch, whose sample site was also located in a park, had higher rates of industrial

use (39.4%), and similar residential use compared to other urban areas (28.6% total). However,

vacant space was reduced to 17.6% in its surrounding watershed. Though this stream maintains

better physical habitat quality than the other urban streams, B-IBI metrics reflected little change

from the worst scoring streams (gaining a poor quality rating as compared to very-poor).

Fish Creek’s sample site was located in yet another urban park. Vacant space in the

surrounding watershed is around 54%, while all types of residential use only totaled 29.9%. The

sample reach of the stream was heavily channellized while its riparian buffer vegetation was

heavy. The primary chemical water quality parameter of note in Fish Creek was reduced

dissolved oxygen at the sample site (59% saturation). This may be attributed to slow moving

water and lack of turbulence. This location contained few physical in-stream features that could

enhance its ability to absorb oxygen more readily. Habitat quality here was lower than other

urban streams, earning one of the lowest HQI scores (15 of 30 possible points). As a result, B-

IBI scores were also poor, though not as low as more urbanized areas.

Floyd Branch, which drains to White Rock Lake in central Dallas, was sampled in a

small park area near industrial uses. At 13.8%, however, these industries are not primary

components of the stream’s watershed but 40% of the upstream land cover is residential. Habitat

quality in Floyd Branch was lower than Dixon Branch primarily due to an abundance of bedrock

as substrate. This area also contained notably higher nitrate concentrations (7.7 mg/L), as well as

Chironomidae: Orthocladiinae occurrence. The surrounding watershed also maintains 11.8% of

its area for parks and recreation.

35

Turtle Creek, located in a very affluent area of Dallas, exhibited very-poor HQI. Habitat

here has clearly been altered by channellization and runoff scouring events. Each side of this

creek maintains very little vegetative buffer, and is bordered by either very expensive, small-lot

housing, or roadways. In total, Turtle Creek’s watershed contains 72.5% residential, 8.5% park

and very little vacant land use (2.3%). Habitat in the stream reach, though not the worst in the

study set, was of less quality than both Floyd and Dixon Branch. B-IBI results showed very-poor

invertebrate community metrics occurring in this stream.

Human inputs to stream systems often go undetected due to runoff transport and dilution

events. Some basic chemical parameters can provide insight to underlying conditions.

Each of the urban streams exhibited increased conductivity levels in comparison to

reference streams. The normal calcium carbonate buffering (i.e. alkalinity and hardness) capacity

contributed more to the overall conductivity of these streams. This suggests that urban non-point

source pollutants are affecting the electrical impedance of the water column.

Though preliminary results show low nitrate concentrations in most of the sub-basins,

two watersheds, Floyd branch and Red Oak creek, contained elevated nitrate levels (7.6mg/L and

4.8mg/L respectively). This may be indicative of increased fertilizer input from residential or

agricultural applications, or simply a result of terrestrial detritus (Connell 1997). Normal levels

of 2.0 mg/L, or less, are commonly used to initiate further research (Dallas 2000).

A working principal of ecology is that diversity of habitat, if it can be described, can be

indicative of the potential diversity of biota. Pardo and Armitage (1997:111) define "meso-

habitats" as "visually distinct units of habitat within the stream, recognizable from the bank with

apparent physical uniformity". Though these habitats vary widely, basic habitat qualification

generates data that can be useful for predicting stream health. Identifying proportions of meso-

36

habitats, such as snags, riffles and gravel beds, within each stream reach contributed to an

understanding of habitat variation among urban and reference settings.

The use of biological metrics to identify perturbation of stream systems is still a relatively

new practice in water quality monitoring (Karr 1991). The resultant indices, however, can over-

simplify results of quality indices if not used cautiously. As a result, some details are lost and

policy makers are often oblivious to specific impacts other than flooding. This can only be

overcome with the retention of specific metrics as they are analyzed. Identification of stream

insects was commonly done at the order/family level in Dallas reports, but stream assessments

can be enhanced with the added resolution of lower taxonomic levels.

Early studies, in the Dallas area, often only recorded the presence of organisms identified

to the order or family level. These observations often lacked the taxonomic sufficiency required

to determine water quality variations throughout the city. This may be due to invertebrate stress

tolerances that vary widely within order and family level groups. This is evident in recent taxa

tolerance lists used by TCEQ. Comparing results of biological indices, using genera level

identifications, to family level results can further define this problem of taxonomic sufficiency.

Though species levels of taxa identification are most desirable for this type of evaluation,

resources for this are often limited, or not available.

Benthic invertebrates are often used in biomonitoring programs due to the ease at which

many long-lived, or sensitive, taxa can be re-established in healthy streams after damaging

events. Degraded streams, on the other hand, can be assessed by a lack of these diverse aquatic

communities (Hynes 1970).

Taxa dominance was clearly responsible for lowering B-IBI scores in some sites.

Chironomidae taxa showed a tendency to colonize impoverished aquatic communities while the

37

highest diversity and evenness scores were found in reference sites. Though functional feeding

groups were relatively even across reference streams, an increase in scraper taxa on Floyd

Branch, reflected as Orthocladiinae, was coincident with nutrient loading and stable substrate

characteristics. Gathering taxa, represented primarily by Dicrotendipes sp and Polypedilum sp,

exhibited a strong dominance over other invertebrates in all other urban sites.

In addition to standard metrics, one experimental measure was assessed. This was a ratio

created from the number of chironomidae taxa present divided by the total percent of the

community represented by chironomidae. With an r2 value of 0.729, this ratio reflects an

expected increasing trend with the B-IBI (Figure 16), presumably due to an inherent reflection of

diversity and evenness of midge taxa. Though it was not included in the primary analysis, further

research could reveal striking potential for inclusion of such a measure into a regional B-IBI.

B-IBI v. Ratio of Chironomidae Taxa to Percent Community as Chironomidae

R2 = 0.729

0

10

20

30

40

50

60

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00

Ratio Chiro.Taxa : % Chiro

B-IB

I

Figure16: Relationship between observed benthic index and the ratio of the number of midge taxa to percent of stream community as midges

Structural and functional biotic metrics effectively describe stream invertebrate

community health when combined appropriately into the B-IBI. This aggregation of metrics

requires the use of exploratory statistical techniques to discern gradients of change in

38

invertebrate community structure as a function of surrounding watershed inputs. Use of multi-

variate techniques such as PCA and CCA provide tools for examining these complexities.

Within CCA axis gradients, intolerant taxa are reduced when driven towards poor overall

habitat quality. Conversely, numerous tolerant, and often dominant, taxa are present within

streams maintaining lower habitat quality. With the inclusion of land use cover into the CCA

model, correlations indicate a gradient ranging from office space, multi-family residential area,

and hotel space to vacant land and mobile home lots. Common characteristics of these uses

include decreased permeability of the land surface, and increased runoff, from higher density

land uses.

Not surprisingly, habitat quality index results were similar to B-IBI results. Of the nine

streams surveyed, urban HQI conditions were lowest in densely utilized areas and highest in

those that were more sparsely populated. These two index techniques were calculated

independently and only combined in the CCA axis ordination. Other examinations of habitat and

invertebrate occurrence indicate a need to further evaluate land use analysis procedures. Many

more sample replicates would be required for this effort, however.

Those programs that attempt to assess land use inputs to streams tend to group urban use

into broad categories. Similar studies have compared agricultural uses to generalized urban uses

to discover that differences do exist between these broad categories (Allen et al 1997). However,

such studies often disregard variability of urban conditions that are difficult to determine without

sufficiently resolute spatial data. Compilation of detailed land use information in the Dallas area

reveals that urban classifications influence overall stream quality. As this land use data is

difficult to acquire, the current land use data was deemed sufficient for analysis in lieu of the

more accurate, and expensive, information. Some land uses corresponded well with overall B-

39

IBI results. For instance, total area of vacant land use was strongly correlated with B-IBI results

across the study area (r2 = 0.8979) (Figure 17).

Watershed B-IBI Scores v. Acres of Vacant Land use

R2 = 0.8979

0

10

20

30

40

50

60

0 5000 10000 15000 20000 25000 30000 35000

Vacant Land use (ac)

B-IB

I

Figure 17. Scatterplot of B-IBI scores versus increasing vacant land use in study watersheds.

40

CONCLUSIONS

Basic monitoring and assessment methods used by the city of Dallas are generally

suitable to fulfill regulatory requirements. In fact, past results of local monitoring efforts were

influential in the design of this project. However, a more detailed examination of invertebrate

data, coupled with spatial analysis of watershed features, could improve resolution of monitoring

procedures. By reducing the number of sampling stations, and increasing efforts for analysis,

urban monitoring programs can be more informed of human influence on stream communities.

Detection of perturbation to invertebrate communities in urban stream systems can help target

acute monitoring efforts efficiently. Furthermore, there have been few studies of specific urban

influences on biotic stream quality measures.

Biotic index scores indicate that Dallas area stream health ranges from very poor in the

city to excellent in reference areas. This was an expected response to the very rapidly growing

urban landscape and its associated watershed runoff. Such use of biological index techniques

help illustrate variations of urban stream quality and are complimented by GIS and statistical

programs such as MVSP. Combining physical habitat variables and invertebrate metrics with

detailed land use parameters enhances watershed health examination. Ultimately, creating an

ability to define human induced habitat perturbation.

Urban stream quality was discernable from stated “best available” reference conditions.

These conditions were sufficient to define B-IBI metrics for use in the blackland prairie eco-

region of North Texas. However, further research is required to determine specific

characterization of land use inputs.

Watershed hydrologic characteristics, as inferred from land cover attributes, were notably

responsible for degradation of stream quality. Without extensive hydrologic data analysis, such

41

as storm run-off volume extremes and pervious cover analysis, this is an under-informed

conclusion however. This need to further evaluate land uses of watershed terrain infers potential

for increased regulation of specific land uses. In the future, this could lead to further regulation

of land use development (Mitton 1997).

Inclusion of GIS analyses to biomonitoring efforts adds significantly to the ability to

report, foresee, and ultimately help control the pollution of area water resources. Though it

cannot substitute for on-site field evaluation, it is a considerable aid in finding and analyzing

areas within a city that contribute to non-point source pollutants in streams. Ultimately, GIS can

help focus public education efforts and environmental inspection methods in problem areas.

Through the increased education of school children and public officials, water quality problems

that are identified as violations of city, state, and federal law can be targeted effectively.

42

REFERENCES

Allan, J. David et al (1997) The influence of catchment land use on stream integrity across multiple spatial scales. Freshwater Biology, 37, 149-161.

Barbour, M.T. et al (1996) A framework for biological criteria for Florida streams using benthic macroinvertebrates. Journal of the North American Benthological Society, 15, 185-211.

Bearwald, T.J. (1991) Social Sciences and Natural Resources. Renewable Resources Journal,Autumn: 7-11.

City of Dallas. (1993) Survey of Area Streams. Dallas Storm Water Quality Division BioAssay Report.

City of Dallas. (1996) Survey of Area Streams. Dallas Storm Water Quality Division BioAssay Report.

Connell, Desley W. (1997) Basic Concepts of Environmental Chemistry. CRC Press. Florida.

Davis, Jack R. (1998) A Benthic Index of Biotic Intergrity for Texas Lotic-Erosional Habitats. Texas Natural Resources Conservatin Commission. Austin, Tx.

Epler, John H. (2001). Identification Manual for the Larval Chironomidae (Diptera) of North and South Carolina.

Gregory, Kirk. (1997) Application of a Geographic Information System for Urban Watershed Assessment, Management, and Restoration. Papers and Proceeding of the Applied Geography Conferences, 20, 76-82.

Haubner, Steven M. et al. (1996) Using a GIS for Estimating Input Parameters in Urban Stormwater Quality Modeling. Water Resources Bulletin, 32(6),1241-1351.

Hauer, F.Richard & Gary A. Lamberti. (1996) Methods in Stream Ecology. Academic Press.

Hynes, H.B.N. (1970) The Ecology of Running Waters. Blackburn Press.

Jager, J. (1998) Current Thinking on Using Scientific Findings in Environmental Decision Making. Environment 40(2), 14-18, 36-38.

43

Johnson, Lucinda B. & Gage, S.H. (1997) Landscape approaches to the analysis of aquatic ecosystems. Freshwater Biology, 37, 113-132.

Johnson, Lucinda B. et al (1997) Landscape influences on water chemistry in Midwestern stream ecosystems. Freshwater Biology, 37, 193-208.

Johnson, Sherri B. & Alan P. Covich (1997) Scales of observation of riparian forests and distributions of suspended detritus in a prairie river. Freshwater Biology, 37, 163-175.

Karr, James R & Ellen W. Chu. (1999) Restoring Life in Running Waters: Better Biological Monitoring. Island Press, Washington D.C.

Karr, James R (1991) Biological Integrity: A Long-Neglected Aspect of Water Resource Management. Ecological Applications. Ecological Society of America. 1(1), 66-84.

Kaufman, Martin M. (1998) Spatial Response Patterns to Stormwater Pollution in an Urbanized Watershed. Papers and Proceeding of the Applied Geography Conferences,21,179-185.

Kerans, B.L., J.R. Karr & S.A Ahlstedt. (1992) Aquatic invertebrate assemblages: spatial and temporal differences among sampling protocols. Journal of the North American Benthological Society, 11(4), 377-390. Kratz, Timothy K. et al (1997) The influence of landscape position on lakes in northern Wisconsin. Freshwater Biology, 37, 209-217.

Lewis, Michael E. (1998) The Role of Urban Stream Corridors in Stormwater Management Planning. Papers and Proceeding of the Applied Geography Conferences,21,230-237.

Merrit R.W. & Cummins, K.W (eds)(1996) An Introduction to the Aquatic Insects of North America, 3rd edn. Kendall/Hunt publishing Co.

Richards, Carl et al (1997) Catchment and reach-scale properties as indicators of macroinvertebrate species traits. Freshwater Biology, 37, 219-230.

Richter, Brian D. et al (1997) How much water does a river need? Freshwater Biology, 37, 231-249.

Saligoe-Simmel, Jill. (1998) Investigation of Seasonal Streamflow Variability in Oregon for Regional Water Quality Monitoring. Papers and Proceeding of the Applied Geography Conferences. 21, 78-87.

Somers, K.M., R.A. Reid & S.M.David. (1998) Rapid biological assessments: how many animals are enough? Journal of the North American Benthological Society, 17(3), 348-358.

44

Texas Natural Resources Conservation Commission. (1999) Receiving Water Assessment Procedures Manual. Water Quality Division, Surface Water Quality Monitoring Program.

Texas Water Development Board. (1997) Water for Texas Today and Tomorrow: a Consensus-Based Update to the State Water Plan. http://www.twdb.state.tx.us/wrp/state-plan/97-water-plan.

Townsend, Colin R. et al. (1997) The relationship between land use and physicochemistry, food resources and macroinvertebrate communities in tributaries of the Taieri River, New Zealand: a hierarchically scaled approach. Freshwater Biology, 37, 177-191.

Ventura, Stephen J. et al. (1993) Modeling Urban Nonpoint Source Pollution With a Geographic Information System. Water Resources Bulletin 29(2), 189-198.

Water Environment Federation (1997) The Clean Water Act: 25th Aniversary Edition. Alexandria, Va.

Wateshed Protection: A Project Focus. (1995) U.S. Environmental Protection Agency Office of Water. http://www.epa.gov/owowwtr1/watershed/focus.

Wiggins, Glenn B. (2000). Larvae of the North American Caddisfly Genera (Trichoptera) 2nd Ed. University of Toronto Press. Toronto, Canada.

45