Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
Relationships Among Land-Use, In-Stream Stressors, and Biological Condition
in Prince George’s County, MD
JoAnna L. Lessard1, Mow Soung Cheng2, Chris Akinbubola2
James B. Stribling1, Erik W. Leppo1
1
Tetra Tech, Inc., 400 Red Brook Blvd., Suite 200, Owings Mills, Maryland, 21117 (Phone: 410356-8993; Fax: 410-356-9005)
2
Department of Environmental Resources, Prince George’s County Government, 9400
Peppercorn Place, Largo, Maryland 20774 (Phone: 301-883-5836; Fax: 301-883-9218)
Abstract
Changes in land use / land cover have been repeatedly shown to be strongly associated with
overall ecological degradation of streams. As natural areas are developed, there is a resulting
change in the hydrologic characteristics. The development activities that alter natural ecosystem
structure and function are considered stressors of that ecosystem. The intermediate linkages and
complex mechanisms that relate these stressors, which are produced by the landscape features
(sources), have not been as well illustrated. Diagnostic analyses performed in this study illustrate
linkages not only between the sources of stressors and the stressors themselves, but also between
the stressors and the biological response variables (benthic macroinvertebrates and fish).
Prince George’s County, for the past five years, has conducted a biological monitoring
program that covered every subwatershed in the County. Using the database developed by this
monitoring program of more than 255 sites, stepwise multiple regression was performed. Results
demonstrated that the strongest source-stressors associations were between medium and high
density residential and commercial and industrial land use as the sources, and reduced overall
physical complexity of the stream habitat (i.e., decreased availability of gravel substrate,
diminished cover, reduced channel sinuosity, and reduced pool variability and substrate), as the
stressors. The biological response variables that had the strongest association with these
stressors were the benthic macroinvertebrate and fish indexes of biotic integrity (B-IBI and FIBI), the Ephemeroptera-Plecoptera-Trichoptera (EPT) Index, Beck’s Biotic Index, % Dominant
Fish Species, and % generalists, omnivores, and invertivores (%GOI). These results have been
used as the basis for developing the County’s Green Infrastructure Plan and the decision making
tool for reviewing new development proposals. The purposes of this paper are to present the
results of these findings and to recommend solutions that can enhance the likelihood of desirable
biological changes. The management recommendations are to reduce, eliminate, or buffer the
most likely sources causing production of the stressors.
Introduction
Determining the relationship between biological condition and human disturbance is
difficult because of the numerous factors and scales acting upon the stream biota. Examples of
factors include: the biological indicator used, geographic setting, proximity and spatial position
1
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
of the human disturbance to the water body, cumulative effects of other stressors, the buffering
effects of hydrology, geomorphology, and riparian vegetation, and the mitigation effects of Best
Management Practices (BMPs). The description of human induced disturbances allows
managers to act upon broader, landscape- (i.e., watershed-) sources of stress in addition to direct
management of stream stressors from more traditional point sources of pollution at individual
sites.
Prince George’s County Department of Environmental Resources has conducted a
countywide biological monitoring program over the past 5 years (1999-2003) to investigate the
ecological condition of its streams and watersheds. Over the five-year study a total of 310
samples were collected from 255 sites which included physical habitat, water quality, benthic
macroinvertebrates and fish. The results of this program indicated that most of the streams
throughout the county were ecologically impaired. One of the recommendations from these
results was to conduct further diagnostic analyses to determine the linkages between biological
condition, the proximal stream stressor(s) leading to the biological responses, and the distal
watershed land-use characteristics that are related to the stressors (i.e., sources of the stress).
The term stressor refers to a human-induced physical or chemical factor that directly causes a
change in some aspect of the stream biota. The stressors can be either positively or negatively
related to the biological condition of the stream, with positive stressors leading to degradation at
low levels (i.e., dissolved oxygen) and negative stressors leading to degradation at high levels
(i.e., deposition). We use the term source to refer to the watershed land-use/land cover
characteristics that potentially produce the stressors to which the stream biota are exposed.
Sources can also be either positively or negatively related to the stream stressors and biological
condition.
Purposes and Objectives
The purpose of this project was to use the results of the county’s routine biological
monitoring and assessment program to better understand the linkages between the stressors of
biological condition and watershed disturbances (or sources) that were occurring throughout the
county and likely producing these stressors. We also envisioned this project as an addition to the
growing body of literature linking human disturbance and biological condition in streams and
providing examples of these relationships for the coastal plain region. The objectives of this
project were to:
1) Determine which stressors were most related to in-stream biological condition in the
county
2) Determine which sources were most related to in-stream biological condition in the
county
3) Determine the relationship between sources and stressors in the county
4) Develop multiple regression models for the most responsive biological metrics to
stressors and sources in the county
5) Develop separate models relating biological condition to sources and stressors for the
three major river basins in the County: Potomac River-Anacostia River (ANA), Potomac
River -Non-Anacostia (PNA), and Patuxent River (PTX) (see Figure 1). ANA is the
most urbanize watershed in the County and PTX is the least.
2
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
Analytical Methods
The first step for this analysis was to compile
the database containing the biological metric,
habitat stressor and related land use/land cover
(LULC) data for each stream sample station.
Biological and stressor data (from 1999-2003) were
readily available from the existing monitoring
database developed previously. All LULC values
were based on 1997 Maryland Department of
Planning data. Primary land cover categories were
calculated as a percentage of the whole area under
consideration (i.e. drainage area upstream of each
site). The percent LC imperviousness was also
calculated for the area under consideration. To
determine whether LULC at the local or watershed
scale was a better predictor of stressor and
biological condition, the LULC data were analyzed
at 6 scales (i.e. areas of consideration): the entire
catchment upstream of each site (catchment or C),
the area upstream of the site within a 200 m, 100 m
or 50 m buffer on either side of the stream (B200,
B100 or B50 respectively), and the area within a
radius of 1 km or 2 km around the site (R1K and R2K).
Figure 1. Drainage Area Map
The next step was to remove redundant, ambiguous or unimportant variables from the
dataset. In order to improve the interpretability of the regression models, metrics that were
highly qualitative or non-continuous were removed as well as any non-numeric or narrative data.
In nearly all cases, each variable removed had a coupled data value that provided the same and
more quantitative information as the variable removed (e.g., the numeric value for total insect
taxa was kept but its score was removed). While initially included, percent Ephemeroptera and
percent Tanytarsini were later removed as response variables due to their high skewness and
non-responsive nature to normalizing data transformations (i.e., could not get them to adhere to
normality assumptions with any transformation). This allowed for a more robust analysis of the
data because the biological response variables used fit well to the normality assumptions
required for parametric testing. Fish total biomass was eliminated after the correlation analyses,
as it showed no relationship to any stressors or sources. Redundant stressor variables or those
variables highly correlated to other stressors were also removed. The individual stressor metrics
used here were the Maryland Biological Stream Survey (MBSS) rapid bioassessment metrics
(Stribling et al. 1998), therefore higher values represent a better condition than lower values (but
not necessarily more of that stressor). Total habitat was, by definition, correlated to most of the
individual habitat stressors that combine to create the total habitat score for each site. Total
habitat as well as the individual stressors were included in the analyses, however, to determine if
overall habitat was the most important driver of biological condition, or if there were a fewer
number of important individual stressors. Stressors were removed, therefore, based on their
3
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
correlations with other individual stressors (e.g., left and right bank riparian zone width are
highly correlated to each other and the mean riparian width, so only riparian zone mean width
was retained). The final list of response variables, stressors and sources that were considered for
the multiple regression models are in Table 1.
Backward, stepwise regression (SYSTAT statistical software) was used to build the best
predictive models relating each biological metric with important stressors and/or sources
(backwards regression rules: tolerance<0.01, =0.05 to enter, =0.05 to remove). Sources were
analyzed at each scale (catchment, within the three buffer widths and within the two radial
scales) and the scale the produced the highest correlation with stressors and biological metrics
were selected as the “important” scale(s) to be included in the multiple regression analyses.
Because of the large number of variables, and high degree of interactions among many of the
variables, we limited the variables to be included in the stepwise regression to those that showed
moderately significant pairwise correlations to biological condition (arbitrarily set as Pearson
correlation coefficients >0.2-0.3). This helped prevent spurious relationships from driving the
final models. Each biological response variable from Table 1, was modeled for the entire county
and for each of the three major basins (ANA, PNA, and PTX).
Table 1.
List of parameters (biological dependent variables, significant stressors and
significant sources) used in the stepwise multiple regression analyses.
Dependent Variables
Beck's Biotic Integrity (BeckBI)
% Invertebrate Clinger Taxa
(Clingers)
Important Scales of Sources
Catchment B200 R1K R2K
Vegetated Cover
Sources
Imperviousness
Medium-density
residential
High-density
residential
Commercial
Conductivity
Industrial
Catchment B200
pH
% silt/clay
%gravel
%Sand
% Cobble
%Bedrock
Institutional
Open urban land
Cropland
Pasture
Deciduous forest
Evergreen forest
Catchment B200
R2K
Catchment R2K
R2K
Catchment R1K
R2K
Pool Variation
Mixed forest
Catchment B50
Pool Subustrate
Brush
B50
Stressors
Total Habitat Score
Channel Alteration
Total Number of EPT taxa (EPT) Sinuosity
Benthic IBI (B-IBI)
Invertebrate Scraper Taxa
(Scrapers)
Total Invertebrate Taxa (TotTax)
Fish IBI (F-IBI)
Native Fish Taxa (NativeTax)
Benthic Fish Taxa (BenthTax)
Intolerant Fish Taxa (Intol)
% Tolerant Fish Taxa (Toler%)
% Dominance of most dominant
fish taxa (Dom%01)
% Fish that are generalists,
omnivores or invertevores
(GOI%)
# of Fish per meter square
(Density)
Catchment R2K
Catchment R2K
Catchment R2K
R1K
R2K
R2K
R2K
Vegetative Protection
Mean Riparian Width
4
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
Results and Discussion
Summaries of the biological and physical data showed that the ANA basin was consistently
the most degraded and the PNA was similar to or slightly better than the PTX (Lessard et al.,
2005). Based on the correlation analyses, a list of parameters was created for the stepwise
regression analyses (Table 1). The results of the best models from the multiple regressions are
summarized in Table 2. The most common stressors retained in the models were total habitat
score, % gravel, channel alteration and sinuosity for the invertebrate metrics, and pool variation
and total habitat score for the fish metrics (Table 2). The most common LULC source variables
retained in the models that were significantly, negatively-related to invertebrate biological
condition and physical habitat condition were percent imperviousness, medium-density
residential, commercial development, and industrial areas. Forested areas and agriculture were
also retained in many of the models and were significantly, positively-related to invertebrate
biological condition. Agriculture is typically considered a negative stressor for stream systems
leading to nutrient and sediment inputs that degrade streams. This dataset, however, lacked
nutrient data and so the influence of agriculture from a nutrient enrichment perspective could not
be addressed. Similar to forested areas, agriculture is a largely vegetated land-use for this area.
It was only a significant factor in two models while forested land-use was significant in 15 of the
multiple regression models (Lessard et al., 2005), indicating that forest land-cover is much more
important for mitigating urban disturbances. For the fish metrics, the most common land use
sources that were retained in the models, and were negatively-related to biological condition
(commercial development, imperviousness, pastureland, institutional development and highdensity residential) (Table 2). Low-density residential land-use was revealed as a significant
positive factor for fish metrics in each basin. This is likely due to the fact that where low-density
residential areas occur, the stressor sources more common in higher density residential
development are lower. For example, low density residential areas may cause fewer stressors
because they contain more open spaces (parks, fields, large yards, treed areas, golf courses etc.).
Table 2. Best models from the stepwise multiple regression analyses
Basin Response Constant b1
ANA EPTTaxa
-1.01
0.03
b5
0.06
PNA
PTX
B-IBI
IntolTaxa
Beck BI
EPTTaxa
B-IBI
Dom01%
Density_F
F_IBI
Beck BI
EPTTaxa
F_IBI
0.62
0.82
-2.72
-1.75
0.49
6.29
-2.56
3.33
5.68
2.57
1.82
0.02
-0.02
0.14
0.04
0.08
2.93
0.06
0.01
0.08
0.06
0.09
X1
TotHabSc
X5
R2K_41
b2
0.12
b6
-0.04
TotHabSc
R2K_imp
C_21
TotHabSc
Channel_Alt
Pool_var
TotHabSc
TotHabSc
%gravel
%gravel
Pool_sub
-0.01
0.12
0.07
-0.07
0.02
0.53
-0.34
-0.10
-0.16
-0.09
0.06
X2
Sinuosity
X6
R2K_43
C_12
R2K_11
B50_43
C_imp
R2K_41
%Silt/clay
Cover
Pool_var
R2K_imp
R2K_imp
B200_11
b3
X3
-0.07 R2K_13
b4
X4
Adj-R2
-0.06 R2K_15 0.51
0.01 R1K_41
0.21
0.07
0.10
-1.40
0.05
0.05
-0.33
0.05
0.02
0.43
0.53
R2K_41 0.55 R2K_42 0.61
R2K_41
0.55
R2K_42
0.55
C_imp
6.33 C_14
0.74
%Sand
-0.40 R2K_42 0.64
C_imp
-0.23 C_14
0.64
R2K_18 0.09 R2K_43 0.55
R2K_43
0.45
B200_21
0.31
5
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
The scale that the source variables had the highest predictive power varied among the
biological metrics. Overall, however, the LULC variables were more strongly related to
invertebrate biological condition within the 2 km radius around the sample site than either within
buffers or total catchment. This implies that, for invertebrates, the land-use/land-cover
characteristics within the proximal environment are most important to the biological condition of
these streams. It also suggests that this area of impact is not small and extends out at least 2 km
around the site. For fish, the LULC within a 200 m buffer or within the entire catchment were
generally more important than radial area. Pool variation and pool substrates were consistently
important to the fish biological condition for streams in the county. These stream habitats are
edge areas, therefore, the linkages between buffer strips and sources of in-stream stress may be
more important for fish than insects.
The Anacostia Basin (ANA), which covers most of the area of the District of Columbia
(DC), is the most degraded area of the county. EPT taxa was the most responsive benthic metric
related to these habitat stressors and LU sources within the ANA basin (Table 2). This model
contained 6 parameters (Table 2). The most responsive and parsimonious model relating these
stressors and sources to invertebrate biological condition was for the B-IBI, accounting for 43%
of the variation in biological condition with only 3 parameters. B-IBI was positively related to
total habitat score and deciduous forest within a 1 km radius, and was negatively related to
medium-density housing developments at the catchment scale (Table 2, Figure 2). Figure 2
shows the best relationships for invertebrates between effects and each individual parameter.
For fish, intolerant taxa was the most responsive metric to LULC and stream stressors, with
% imperviousness and low-density residential in a 2km radius accounting for 53% of the
variation (Table 2). The most strongly-related stream stressor in the multiple regression models
for intolerant fish was pool variation. These models indicate a potential dichotomy in the
sensitivity to scale of different biotic measures. The more specific measure of EPT taxa richness
was related to similar source parameters as biotic integrity (B-IBI), but at a scale of 2km radial
distance. B-IBI was more sensitive to forest locally (at the 1 km radial scale) and urban
development more distally (catchment scale). This illustrates that catchment scale LU, especially
for urban LU, is as important as riparian habitat for stream conditions.
The main effects relationships between each multiple regression parameter from the best
models across all basins for invertebrates and fish, respectively, are shown in Figures 2 and 3.
These figures represent relationships between land-use gradients and biological condition. The
slopes of the graphs in Figure 2 show how biotic condition falls more steeply as housing in the
catchment increases, compared to increases due to forest. This implies that forest as a mitigator
of disturbance is not likely to cancel out the influence of urbanization. These models also
indicate a strong link between improved biotic integrity and physical habitat (Figure 2).
Biological condition metrics in the Potomac Non-Anacostia Basin (PNA) basin were
generally the most responsive to the land-use sources and habitat stressors that we measured
(Table 1). Similar to the ANA basin, the primary stressor related to biological condition (benthic
macroinvertebrates) was total habitat (mainly from channel alteration and reduced sinuosity)
(Table 1). Total habitat score and pool variation were also the most significant stressors to fish
biological condition (Table 1). Individual correlations between land-use types and biological
6
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
condition demonstrate that the primary driver of invertebrate biological degradation in this basin
is % impervious land-use at the catchment and 1km radius scales (Lessard et al., 2005), although
these parameters were not retained in the stepwise analysis. The fish multiple regression models
retained commercial development at the catchment scale as the primary source of biological
degradation (fish) (Table 1).
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
B-IBI
B-IBI
B-IBI
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
ANA
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
R2 =0.13
10 20 30 40 50 60
% Deciduous Forest (R1K)
PNA
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
R2=0.17
0
5
10
Evergreen Forest (R2K)
4.5
4
3.5
R2 =0.19
3
2.5
2
1.5
1
0.5
0
20
40
60
80
100 0
% Medium-Density Residential (C)
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
R2=0.30
20
40
% Impervious LU (R1K)
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
R2 =0.33
50
100
Total Habitat Score
150
15
60
R2=0.36
0
5
10
15
20
Channel Alteration
25
Figure 2. Main effects models for the significant parameters from the multiple regression
analyses relating invertebrate biotic integrity with land-use and stream habitat.
7
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
PNA
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
R2=0.34
0
R2=0.38
0
2
4
6
8
Low-Density Residential (R2K)
R2=0.07
0
10 20 30 40 50 60
Imperviousness (R2K)
Fish IBI
2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
Fish IBI
Intolerant Taxa
ANA
Intolerant Taxa
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
The main effects models in Figures 2 and 3 show the relationships between biotic integrity
and channel alteration, evergreen forest and % imperviousness (in 1 km radius) for invertebrates,
and total habitat and commercial development for fish, respectively. For the PNA basin,
invertebrate biological condition was more related to local land-use influences than in the ANA
basin, which should make targeting management efforts near streams even more effective for
biological recovery in this basin. Fish showed a different trend with biological integrity being
significantly, positively related to total habitat and significantly, negatively related to
commercial development at the catchment scale (Figure 3). The lack of predictive power for the
fish metric models indicates that they may be less useful diagnostic analyses in this basin.
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
10
50
100 150
200
Total Habitat Score
250
R2=0.45
0
10
20
30
40
50
Commercial Development (C)
Figure 3. Main effects models for the significant parameters from the multiple regression
analyses relating the most significant fish metrics with land-use and stream habitat.
8
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
The biological metrics in the Patuxent Basin (PTX) basin were generally less responsive
to these sources and stressors, compared to the other two basins. The most important predictors
of biological condition in the PTX were % gravel substrate and % impervious land-cover for
invertebrates, and pool substrate and total habitat for fish. Percent gravel substrate, %
imperviousness and mixed forest combined to explain 45% of the variation in EPT taxa richness,
which was the best, most parsimonious invertebrate model for this basin. The best fish model
was for F-IBI, which had 31% of its variation explained by pool substrate, low density
residential and pastureland (Table 1). The individual main effects models for the Patuxent are
provided in Lessard et al. (2005). In this basin, lack of gravel substrate was an important habitat
stressor in all of the multiple regression models for invertebrates (Lessard et al. 2005). Improving
substrate conditions in the PTX, therefore, may be the best place to start for mitigating the
negative influences of urbanization. The F-IBI main effects models were very similar to those
for the PNA, with biotic integrity increasing with better pool habitat (e.g., pool substrate) and
also increasing with more low-density residential areas. Unlike the PNA, however, the scale of
greatest importance in the Patuxent basin was in the 200 m buffer strip. Again, these models
point to the importance of pool habitat for fish biological condition and the influence of more
local disturbances on pool habitats.
Conclusions
Prince George’s County is an area rich in aquatic resources, but is also heavily urbanized.
Returning these streams to pre-settlement conditions is likely not an attainable goal, but
nonetheless, improving the biological condition from the present state should be achievable. The
stream biological monitoring and assessment program provided important information as to the
biological condition of streams in the county, and the source/stressor models developed in this
study provide insight on the effects of the land-use history of this region. These results provide
targeted information for managers in each of the major drainage basins and can be used to
prioritize management efforts for rehabilitation opportunities throughout the county. The most
consistent and significant pattern that these analyses reveal is that local habitat stressors are the
most important drivers of biological degradation. While these stressors are related to the LULC
sources, it may be possible to de-couple their influence with specific, local management efforts.
Land-cover conversions that lead to loss of natural stream habitat complexity will degrade the
biological community, a result confirmed by many earlier studies (Allan 2004). To buffer the
effects of future degradation, or to contribute to potential biological recovery, it would be best to
manage for habitat heterogeneity. Natural landscape features, like forest, promote natural
channels and provide important functions for streams including shade, organic matter, woody
debris, root habitat, and urban run-off interception and filtration. Management efforts to
maximize these ecosystem “services” should be capitalized on as cost-effective ways to promote
stream integrity and aesthetics.
9
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006
Literature Cited
Allan, J.D. 2004. Landscape and riverscapes: The influence of land use on river ecosystems.
Annual Reviews of Ecology, Evolution and Systematics 35:257-284
Downloaded from ascelibrary.org by University Of Maryland on 02/25/15. Copyright ASCE. For personal use only; all rights reserved.
Bressler, D.W., J.B. Stribling, M.J. Paul, and M.B. Hicks. 2005 (in review). Stressor tolerance
values for benthic macroinvertebrates in Mississippi. Submitted to Hydrobiologia.
Bryce, S. A., D. P. Larsen, R. M. Hughes, and P. R. Kaufmann. 1999. Assessing relative risks to
aquatic ecosystems: a Mid-Appalachian case study. Journal of the American Water Resources
Association 35:23-36.
Frissell, C.A., W. J. Liss, C. E. Warren, M. D. Hurley. 1986. A hierarchical framework for
stream habitat classification: viewing streams in a watershed context. Environmental
Management 10(2): 199-214.
Lessard, J.L., J. Stribling, and E. Leppo. 2005. Relationships between land-use, in-stream
stressors and biological condition in Prince George’s County, MD. Prepared by: Tetra Tech, Inc.
Owings Mills, MD. Prepared for: PG DER, Largo, MD.
Naiman, R. J., P. A. Bisson, R. G. Lee, and M. G. Turner. 1997. Approaches to Management at
the Watershed Scale. Chapter 16 (pp. 239-253), In, Kathryn A. Kohm and Jerry F. Franklin
(editors), Creating Forestry for the 21st Century. The Science of Ecosystem Management. Island
Press, Washington, DC.
Omernik, J.M. 1995. Ecoregions: A Spatial Framework For Environmental Management. Pp.
49-62, In, T. Simon & W. Davis (editors), Biological Assessment and Criteria: Tools for Water
Resource Planning and Decision Making.
Prince George’s County Department of Enviornmental Resources (PG DER) 2003. Biological
Assessment of the Streams and Watersheds of Prince George’s County, Maryland, Spring Index
Period . Prince George’s County, Maryland. Department of Environmental Resources, Programs
and Planning Division, Technical Support Section. Landover, Maryland.
Pringle, C.M., R.J.N., G. Bretschko, J.R. Karr, M.W. Oswood, J.R. Webster, R.L. Welcomme,
and M.J. Winterbourn. 1988. Patch dynamics in lotic systems: the stream as a mosaic. Journal of
the North American Benthological Society 7(4): 503-524.
10
Copyright ASCE 2006
World Environmental and Water Resources Congress 2006
World Environmental and Water Resource Congress 2006