vvEPA
'Environmental-Protection -
Agency
Development : •v• ""•''
Washington DC 20460
:EP/!y600/R-97/130
November 1997
An Ecological
Assessment
of the United States
Mid-Atlantic Region
ill
-------
Printed with Soy/Canola ink on paper that contains at least 50% recycled fiber, and although the paper is "coated" (to help maintain the
color brilliance/intensity of the inks), it is recyclable.
-------
An Ecological Assessment
of the United States
Mid-Atlantic Region:
A Landscape Atlas
K. Bruce Jones1, Kurt H. Riitters2;5, James D. Wickham2, Roger D. Tankersley Jr.3,
Robert V. O'Neill4, Deborah J. Chaloud1, Elizabeth R. Smith2, and Anne C. Neale1
1 Environmental Sciences Division, U.S. Environmental Protection Agency,
Las Vegas, Nevada
2 Environmental Research Center, Tennessee Valley Authority,
Historic Forestry Building, Norris, Tennessee
3 Department of Geography, University of Tennessee, Knoxville, Tennessee
4 Environmental Sciences Division, U.S. Department of Energy,
Oak Ridge National Laboratory, Oak Ridge, Tennessee
5 Currently: Biological Resources Division, U.S. Geological Survey,
Knoxville, Tennessee
-------
Dedication
This Atlas is dedicated to our friend and colleague Mason J. Hewitt, whose leadership and inspira\
tion made many of the landscape analyses and displays used in this atlas possible. Masor,
pioneered and laid much of the foundation for Geographic Information System applications in the
EPA. He made it possible for many government agencies to use and apply indicators highlighted
in this atlas. Mason also contributed substantially to the education of many young people througf\
the Boy Scouts of America, teaching young people how to respect and live in harmony with t
natural environment. Mason's impact on the conservation of our environment will be felt for year
to come, but his kindness, leadership, and vision will be sorely missed.
-------
Table Of Contents
Chapter 1. Taking a Broader View - :- -^--^^s^^^ >«..:. -i. ^~~=~:— ..-.. . -j
Purpose and Organization of This Atlas -^^^^^r^agssssB.^^^- -••. • ..--.— ^.-r^—-. 2
Landscape Ecology and the Analysis of Broad-Scale
Environmental Condition --:::— —^——^^^^ 3
How Can Landscape Indicators Help Us Understand
Environmental Conditions? - ----^g^am^ 4
How Were the Landscape Indicators Selected? --^^=^=r=- -^^r^===^=:. Q
How Were the Landscape Indicators Measured? --^^-^^^-- .-.- — —- 10
How Were the Landscape Indicators Summarized? --~-=^==s^a^ss^=====^== 14
How to Read the Maps and Charts in this Atlas ^^^i^ss^^^ass^sm^--^. -) 5
Chapter 2. The National Context - ^r^^aa^^ 17
Data Sources ^^r^r^^aa^^ 1 7
Human Use Patterns •-•^--— "-~=sggg^^^^^ 20
Forest Patterns ^--—^-rBsmg^^ 22
Patterns Affecting Water Quality -:— -^nrsaags^^ 26
National Context Summary :;:::^:3:^-^3a»a^ 28
Chapter 3. The Mid-Atlantic Region -•--~-~^^^^^ 31
Biophysical Setting of the Mid-Atlantic Region xag^,--^.^.-..^.,...-,..... ..... .v-^r 31
Humans in Landscapes in the Mid-Atlantic Region ;::r-^.,-^!»«-». .......... ; ,^==::- 33
Water ^--.T^SSS^^ 49
Riparian Indicators --^^^^gsggs:^ ^ ,^.^« aiM.*^." =-^^r^ssiBs^-3i=^^^:--:. 49
Watershed Indicators --^^^s^^^ 52
Forests -^—r-^rsg^^ . 00
Landscape Change (1975-1 990) -r^ra^ga?^^ .*~~^~^~r^z=z^=^. Q7
Chapter 4. A Comparative Assessment of Mid-Atlantic Watershed Conditions 77
Glossary -•':--^~-"~-i— -^-^^ 89
Appendix. Additional Information About the Indicators in Chapter 3 .:~^-^-- •-- 91
Suggested Further Reading - :::::r--gm-asr^ia«as^^ ..... • -rr—,-~r^:.. 103
Acknowledgements -^-^>»^^ Inside Back Cover
-------
-------
Chapter 1: Taking a Broader View
Environmental quality affects our health, our quality of
life, the sustainability of our economies, and the futures
of our children. Yet pressures from an increasing popu-
lation coupled with the need for economic development
and an improved standard of living often have multiple
effects on our natural resources. So just as a person
with a less-than-healthy lifestyle is more prone to infec-
tion, a weakened ecosystem is more vulnerable to
additional stress. Unfortunately, it is often difficult to see
these changes in environmental quality because they
occur slowly or at scales we do not normally consider.
There is growing public, legal, and scientific awareness
that broader-scale views are important when assessing
regional environmental quality. In the past, media atten-
tion has concentrated on dramatic events, focusing our
environmental awareness on local or isolated phenom-
ena such as cleaning up Superfund sites, stopping
pollution from a drainage pipe, saving individual endan-
gered species, or choosing a site for a county landfill. In
an era of environmental regulations, measures of envi-
ronmental quality were based on legal standards, like
those for drinking water or air quality. As a result they
reflected a limited view of the environment and the
multiple factors that contribute to environmental prob-
lems. In response, scientists studied fine-scale model
systems and often considered humans to be external
factors. Today, our perceptions are changing. We
realize that humans and our actions are an integral part
of the global ecosystem, and that the environment is
complicated and interconnected with human activities
across local and regional scales. We have begun to take
a broader view of the world and of our place in natural
systems.
Technological advancements have made it easier to
obtain new views of overall environmental quality. Com-
puters and satellites allow us to study larger patterns
and processes. Combined with a better understanding
of how the pieces fit together, these technologies help
us to assess where we are now with regard to environ-
mental quality, to envision where we hope to be in the
future, and to identify the steps we need to take. This
atlas takes advantage of these advanced technologies in
assessing environmental condition over the mid-Atlantic
region of the United States.
Just as we now watch broad-scale weather patterns to
get an idea of whether it will rain in the next few days,
we can develop a better assessment of current environ-
mental condition by combining regional and local-scale
information. Broad-scale weather patterns are important
because they affect and constrain what happens locally
on any given day. By taking a broader view of the envi-
ronment, or widening our perspective about how the
environment is put together, it becomes easier to see
where changes occur and to anticipate future problems
before they materialize.
Chesapeake Bay Program
The Chesapeake Bay Program is one of the groups which helped to identify the
environmental issues of concern in the mid-Atlantic region. The Chesapeake
Bay watershed covers a large portion of the area considered in this atlas.
In the past, public and legal attention has been focused on site-specific
environmental problems such as what is coming out of individual drainage pipes.
-------
Purpose and Organization of this Atlas
This atlas is an ecological assessment of the mid-
Atlantic region of the United States (Figure 1.1). The
assessment was done using measurements derived from
satellite imagery and spatial data bases. The information
presented in this atlas is intended to help the reader
visualize and understand the changing conditions across
the region, and how the pattern of conditions can be
used as a context for community-level situations. This
atlas does not provide site-specific analyses of small
areas such as individual woodlots. This atlas was
developed as part of the Environmental Monitoring and
Assessment Program (EMAP), and is part of a larger,
multi-organizational effort to assess environmental
condition in the mid-Atlantic region.
The atlas is divided into four chapters with one
appendix. This chapter introduces the reasons
for doing a broad-scale regional analysis of
environmental condition.
Chapter 2 places the mid-Atlantic region into
the context of the lower 48 states. In Chapter 3,
the landscape conditions in the mid-Atlantic region are
analyzed and interpreted in terms of a set of ecological
indicators, summarized by watersheds within the region.
Chapter 4 summarizes the overall picture painted by
these landscape indicators and compares relative
conditions among watersheds in the region. The Appen-
dix provides methodological information which is not in
Chapter 3, and has a listing of all indicator scores for
every watershed in the mid-Atlantic region.
Figure 1.1.
The mid-Atlantic region,
-------
Landscape Ecology and the Analysis of
Broad-Scale Environmental Condition
To most people, the term "landscape" suggests either a
scenic vista or a backyard improvement project. To
ecologists and other environmental scientists, a land-
scape is a conceptual unit for the study of spatial
patterns in the physical environment and the influence of
these patterns on important environmental resources.
Landscape ecology is different from traditional ecology
in several ways. First, it takes into account the spatial
arrangements of the components or elements that make
up the environment. Second, it recognizes that the
relationships between ecological patterns and processes
change with the scale of observation. Finally, landscape
ecology includes both humans and their activities as an
integral part of the environment.
There are many applications for landscape ecology and
broad-scale information in regional assessments. For
example, we can identify the areas that are most heavily
impacted today by combining information on population
density, roads, land cover, and air quality. In the mid-
Atlantic region, we already have good information (from
the U.S. Census Bureau) about which counties are most
urbanized. But which counties have only a small pro-
portion of adjacent forest cover along the stream length?
Which counties are charac-
terized by a high degree of
forest fragmentation? What
about information for water-
sheds instead of counties?
Broad-scale measurements
can be taken in order to
make relative comparisons of
these indicators over the
entire region. Broad-scale
data can also help in identify-
ing the most vulnerable areas
within the region. Vulnerable
areas are not yet heavily impacted, but because of their
circumstances they are in danger of becoming so. One
example might be a watershed that has a relatively high
percent of forest cover, but is also experiencing rapid
population gains. Such an area might be more vulnerable
to forest fragmentation than a similar area with less
population or less forest area.
The ability to place localities into a regional context is
another benefit of this approach. Some individual cities
and neighborhoods in the mid-Atlantic Region may seem
isolated, perhaps within a large forested area. However,
all are connected by physical features and by ecological
processes. Water flows from one place to another, roads
provide a connecting infrastructure, and land cover
patterns of forest and agriculture form a connected
backdrop for all of our activities. While land management
decisions are made and implemented at a local scale, a
regional perspective can guide our decisions and make
us better stewards of our environment. By placing our
homes, neighborhoods, and government organizations
into a regional landscape picture, we can begin to make
informed decisions that consider not only our goals and
actions, but our neighbors' as well.
-------
E
Figure 1.2 illustrates how a single community is linked to
the landscape at several different scales and across
different mapping units (watersheds and counties in this
example). A small city is highlighted in the middle of the
figure. At this scale we concentrate on individual land
parcels and roads, and our decisions are based on a
local perspective. Broader-scale perspectives emerge
as we follow the lines up either side of the figure. We
see that the community is part of both a watershed (left)
and a county (right), which, in turn, are components of
groups of watersheds and counties. These larger groups
are components of the entire region.
How Can Landscape Indicators Help Us
Understand Environmental Conditions?
An indicator is a value calculated by statistically combin-
ing and summarizing relevant data. Well-known
economic indicators include the seasonally-adjusted
unemployment percentage and number of housing
starts, both of which indicate overall economic condition.
In these indicators, seasonal adjustment is made with a
model, and most economists look at several indicators
together instead of just one at a time. Similarly, land-
scape indicators can be measurements of ecosystem
components (such as the amount of forest) or processes
(such as net primary productivity), and modeled adjust-
ments can be used to help interpret the measurements in
order to understand overall ecological conditions.
Figure 1.2.
This atlas may help to understand how
a city (bottom center) fits into a larger
context of either watersheds (left
branch) or counties (right branch).
-------
Figure 1.3 shows an example of measuring spatial
patterns as an indicator of stream conditions. The
distribution of streamside land cover has been mapped
for the same county that was shown in Figure 1.2.
Stream segments that are green have adjacent forest,
and segments that are yellow and red have adjacent
agriculture and urban land cover, respectively. The
pattern of streams in relation to land cover is an indicator
of conditions within the stream. Forests filter pollutants,
Figure 1.3.
Spatial patterns of land cover in
relation to streams fora county in the
mid-Atlantic region. Stream
segments are colored green, yellow,
or red, depending on whether the
segments are adjacent to forest,
agriculture, or urban land cover.
preventing them from reaching the water, whereas
agriculture and urban land often contribute pollutants to
streams. A simple summary indicator might be the
percentage of stream length in the county that is adjacent
to forest land cover. To refine this indicator, a model
might help to account for "natural" conditions, for example
whether or not forest was the natural land cover for the
county.
•"••-, - "-• - -: V; ;-- >:^.i . - ..?'. - ;••- '"_'-
-------
How Were the Landscape Indicators Selected?
As a starting point for selecting indicators, we considered
what people in the region said they cared about. For
example, concern for wildlife populations provides a
reason to examine indicators of habitat fragmentation.
Fragmentation of natural habitats can severely affect
animal populations, as shown by the conceptual model
illustrated in Figure 1.4. Concerns from the mid-Atlantic
were then matched to our ability to take meaningful
measurements, recognizing that some things just can't
be measured very well given the available data or models.
As a result of workshops and advice from people in
the mid-Atlantic region, four general environmental
themes were identified — people, water, forests, and
landscape change. Figures 1.5 and 1.6 are pictorial
representations of key landscape attributes that affect the
sustainability of environmental condition across broad
scales. Figure 1.5 shows some key landscape compo-
nents that sustain a high quality environment, and Figure
1.6 shows some human modifications of the landscape
that can reduce the sustainability of natural resources.
These illustrations represent some of the important
landscape indicators analyzed in this atlas.
X
/I
*
t
•
Large-scale
disturbance
(fire or flood)
1,
I
Population
persists
Recovery
Large-scale
disturbance
J1 f
-------
Forest connectivity
is crucial for the
persistence of
forest species,
especially in areas
with moderate
amounts of
agriculture
Riparian zones
filter sediments
and pollutants,
especially in
agricultural
areas, in
addition to
providing
important
wildlife habitats
Large blocks of
interiorforest
habitat are
importantfor
manyforest
species
Figure 1.5.
A pictorial representation of some
landscape components that sustain
a high-quality environment.
Thenumberof
forest scales
surrounding a
point in the
landscape
determines the
varietyof
forestspecies
found there
Forest edge
habitat is
importantfor
manyspecies
that require more
than one habitat
type to survive
-------
Agriculture on steep
slopes increases soil
loss and sediment
loading to streams
Dams alter the natural
habitats and hydrology
ofstreams
Agriculture areas near
streams increase stream
sediment loads and
chemical inputs
Figure 1.6
A pictorial representation of some human
modifications of the landscape that reduce the
sustalnability of natural resources
The amount and
location of agriculture
in a watershed
influences
landscape pattern
-------
Humans reduce riparian
cover along streams,
which decreases
filtering capacity
Forestharvest
practices influence
forest connectivity
and patch sizes
Air pollution spreads
across the landscape,
affecting regional
air quality
Roads near streams
increase sediment and
pollution loads by
increasing surface
runoff
Population growth
results in loss of
forest and changes in
overall watershed
landscape pattern
-------
The indicators reported in this atlas are not appropriate
for addressing some kinds of questions. For example,
they are not intended to assess conditions for very small
areas. The goal was to develop a consistent and com-
prehensive look at the entire region, and there were
trade-offs between the level of detail and the size of the
area that could be considered. Future work would look
at smaller areas using more detailed data sets. The
regional perspective would be a valuable guide to deter-
mine where this additional expense might be warranted.
The indicators reported here were not evaluated in
absolute terms; only relative comparisons were made. In
order to set absolute standards like the ones which exist
for drinking water and air pollution, the system must
either be very simple or intensively studied to provide
very detailed scientific information. Regional ecosystems
are simply too complicated to set absolute standards
using our current technology and understanding.
Landscapes are very complicated, and the generality of
the conceptual models is an accurate reflection of level
of scientific understanding concerning landscape dynam-
ics. Scientists who study landscape ecology are trying to
improve our ability to interpret landscape indicators
relative to environmental values. The improvements will
help to interpret the information that is contained in this
atlas and may suggest new landscape indicators that we
have not considered. In the meantime, it is worth explor-
ing how much is known about regional environmental
conditions and what conclusions can be made using
state of the art landscape indicators.
How Were the Landscape Indicators Measured?
Many kinds of data were used to prepare the indicators
shown in this atlas. Federal agencies were the primary
source for data, including maps of elevation, watershed
boundaries, road and river locations, population, soils,
land cover, and air pollution. Sources included the U.S.
Geological Survey (USGS), the U.S. Environmental
Protection Agency (USEPA), the U.S. Department of
Agriculture (USDA), the U.S. Census Bureau, and the
Multi-Resolution Land Characteristics Consortium
(MRLC).
Data collected by satellites were used to map land cover
and its change over time. The sensors carried on satel-
lites measure the light reflected from the Earth's surface.
Because different surfaces reflect different amounts of
light at various wavelengths, it is possible to identify land
cover from satellite measurements of reflected light.
Figure 1.7 illustrates the differential reflectance properties
of water, sediments suspended in water, and land sur-
faces for a typical satellite image. Examples of land
cover maps derived from satellite images appear later in
this atlas.
In a typical digital map, data are stored as a series of
numbers for each map. These maps can be thought of
as checkerboards, where each grid square (or pixel,
which is an abbreviation of "picture element") represents
a data value for a particular landscape attribute (for
example soils, topography, or land cover type) at a
specific location.
-------
Figure 1.7.
Illustration of differential light
reflectance properties for
water, sediments suspended
in water, and land surfaces
over a portion of eastern
Virginia and the Chesapeake
Bay. These images can be
manipulated in various ways
to extract information about
the Earth's surface.
Source: North American
Landscape
Characterization
Program
-------
Several techniques are used to take a measurement of a
landscape indicator. One method ("overlaying") simply
examines maps of different themes in order to extract
information about spatial relationships among the themes
(Figure 1.8). For example, by overlaying maps of land
cover and topography, we can look at the occurrence of
agriculture on steep slopes. These relationships are then
stored as a new map which combines the information
from the original set of maps. Another method ("spatial
filtering") can be thought of as using a "sliding window" to
calculate indicator values within small areas that are part
of a larger map (Figure 1.9). Spatial filtering is used here
to create surface maps of indicator values; these surface
maps help us to visualize the spatial pattern of indicators
in more detail than is provided by the watershed-level
summaries described in the following section.
Land cover (with agriculture in red) is combined with topography to
indicate agriculture on steep slopes. The combined map shows
agriculture on slopes greater than 3%.
Figure 1.8,
Example of overlaying digital maps to produce a new map of an indicator.
-------
Original land
cover data:
Stepl
Step 2
StepS
Resulting, spatially
filtered data:
In this example of the spatial filtering process, a 3 pixel by 3 pixel
window (outlined in red - top row of figures) is used to map land
cover diversity. In step 1, there are 2 cover types in the window
which maps to a single blue pixel at the center of the window. In
step 2, the window slides over one pixel. There are 3 cover types
in the new window, mapping to a single green pixel in the center of
the window. In step 3, the window again slides over one pixel. The
third window includes 3 cover types and maps again to a single
green pixel.
Legend for Filtered Coverage
One Cover Type
Two Cover Types
Three Cover Types
Four Cover Types
More Than Four Cover Types
Figure 1.9.
Illustration of spatial filtering which creates a surface map.
-------
How Were the Landscape Indicators Summarized?
This atlas uses watersheds, as defined by USGS hydro-
logic accounting units, to summarize landscape indicator
values (Figure 1.10). Roughly speaking, hydrologic
accounting units follow watershed boundaries. In many
ecological studies, especially those which assess water-
related concerns, watersheds are an appropriate unit for
summarizing data. A watershed is defined as an area of
land that is drained by a single stream, river, lake, or other
body of water. The dividing lines between watersheds
are formed by ridges. Water on one side flows into one
stream, while water on the other side may flow into a
different stream. Thus, watersheds are a natural unit
defined by the landscape. Strictly speaking, the USGS
hydrologic accounting units are not watersheds in the
classical sense of a topographically-defined catchment
area. They are used in this atlas because they are gener-
ally accepted and consistent across the entire nation.
To determine relative condition, the watersheds were
ranked by the values for a given indicator, from highest to
lowest, and then were divided into five groups. Each
group had an equal number of watersheds; at the na-
tional scale (Chapter 2) there were approximately 425
watersheds in each group. At the mid-Atlantic regional
scale (Chapters 3 and 4) there were 25 watersheds in
each group. All watersheds within the same group were
colored with one of five colors, using green to represent
more-desirable conditions and red to represent less-
desirable conditions. Maps based on rankings are useful
for comparing relative conditions, but they do not convey
the actual values of the indicators. That information is
summarized in the companion bar charts which show the
number of watersheds with different indicator values. By
looking at the map and bar chart together, it is possible to
estimate the ranges of indicator values associated with a
given watershed group.
As a practical matter, the authors of this atlas made
judgment calls when assigning 'red' and 'green' colors to
the maps, and 'more desirable' and 'less desirable'
interpretations to the indicator values. For example,
forest edge was colored 'green' and interpreted as 'more
desirable' when its values were high because the mea-
surement was included as an indicator of a type of
habitat.' Similar judgment calls were made for other
indicators. Higher values for the vegetation-increase
indicator were considered to be a negative impact be-
cause much of this change did not represent restoration
of the potential natural vegetation, but rather was more
strongly associated with human activities. One of the
advantages of presenting indicator scores for all water-
sheds (see Appendix) is that any reader can simply
redefine the color scheme and make new judg-
ments based on other criteria.
To calculate indicator
values for a watershed,
the watershed boundary is
overlayed on a GIS
coverage. Information for
that watershed is then cut
out from the larger
dataset.
Figure 1.10.
Illustration of the cookie-cutter
process that was used to
summarize information by
watershed.
-------
How to Read the Maps and Charts in this Atlas
Figure 1.11 illustrates the types of maps and charts that
appear in Chapters 2 and 3, with some description of
their various elements. It may be useful to bookmark
this page for later reference.
EThe map of mid-Atlantic watersheds is •color-abided.'''
Jto show relative conditions among watersheds. The
icblors range from red to green, indicating relatively
B'less desirable" and "more desirable" conditions,
Irrespectively.
7A quintile
^contains 1/5 of
the watersheds.
Quintiles are
^formed after
•Cranking
watersheds for
R the indicator.
1
Quintile
1
2
3
4
5
Data Range (Percent)
<70.600
70.600-76.869
76.870-84.579
84.580-89.889
>89.890
iThe Data Range
:shows the
indicator values
*for watersheds
contained in each
le.
irThe value shown on the X axis is the upper limit of a 1
data range. For example, this bar shows the number of
watersheds with data values between 60-70. 1
A brief explanationoftheessential methods js given.
_Jp>jgtails are in tjhg Appendix.
40
I 20
10 20 30 40 50 60 70 80
Indicator Value
Figure 1.11.
How to read the maps and charts in this atlas.
Woody landcover along streams was calculated as the percent
of streamlength with forest landcover types. By intersecting a
buffer zone around each stream with the landcover, a dataset is
created which records all landcover types within a specified
distance to stream center.
Sources: USGS 1:100,000 River Reach 3 stream data, and
MRLC 30 meter Landsat land-cover data.
90 100
-------
-------
Chapter 2: The National Context
Before looking in detail at the mid-Atlantic region, it is
helpful to place the region within a national perspective.
This chapter paints a picture of the lower 48 United
States, showing differences and patterns among water-
sheds at a continental scale. A national context helps us
interpret the overall condition of the mid-Atlantic region,
relative to the rest of the country. It also helps to deter-
mine if the conditions like those found in the mid-Atlantic
region exist elsewhere.
While it would be desirable to look in great detail over the
entire nation, in practice only a few aspects of environ-
mental condition can be described in a consistent fashion
nationwide. The coarse-scale maps in this chapter show
watershed rankings based on a variety of landscape
indicators (Table 2.1). The rankings portray relative condi-
tions across the nation but do not show the absolute
values of indicators for each watershed. Indicator values
are summarized in the companion bar charts.
Data Sources
Four main data sources were used here. The most
important was a national map of land cover (Figure 2.1)
which shows areas dominated by urban communities,
water, or vegetation such as forest, crops or pasture.
Although the resolution (spatial and land cover) is fairly
coarse (1 square kilometer units, each assigned to 1 of 9
general land cover classes), the familiar national pattern is
apparent — forests in the East, grasslands and crops in
the Midwest, and shrublands, deserts, and mountain
forests in the West. The mid-Atlantic region is typical of
other eastern regions — coastal and riverside urban
areas, agricultural valleys and coastal plain, and forested
mountains and plateaus. Relative to other regions in the
United States, the complexity of land cover in the mid-
Atlantic region can make spatial pattern an important
factor for environmental decisions.
Three other sources of information were used to calculate
landscape indicators nationwide. Figure 2.2 shows the
maps of roads, streams, and watersheds. Clearly, not all
the roads and streams are included. These maps may be
appropriate for a nationwide overview, but much more
detailed maps are needed for regional assessments such
as the mid-Atlantic analysis. The watershed boundaries
identify 2,099 individual watershed units.
Figure 2.1.
National land cover map. The
U.S. Geological Survey produced
this map of "Seasonal Land Cover Regions of the
Coterminous United States." The map was derived
from March-October (1990) 1-km Advanced Very High
Resolution Radiometer (AVHRR) imagery, digital elevation,
ecoregions, and climate data. The original 160 classes of land
cover have been grouped into the 9 broad categories shown here.
Western Forest
Eastern Forest
[H| Croplands
Shrublands
Grasslands
Wetlands
Water
Barren
Urban
-------
For each watershed, the nine indicators included in this
chapter were calculated from land cover and from the
spatial relationships among roads, streams, and land cover.
The maps are color-coded to show relative conditions
among watersheds (as described in Chapter 1).
V;;r«¥,! I;:; S
(a). roads, (b) rivers,
|fc| waferafisd boundaries,' The
<|i" 'TifiifiS are from the ArcUSA distiibu- "'
''~
ngl Line Graph maps of rivers
||73J and roads (1980), and (he ]
t map of8r- ' I
Table 2.1. List of landscape indicators used for the national context.
U-lndex (proportion of watershed area with anthropogenic land cover)
Agriculture Index (proportion of watershed area with agriculture land cover)
Number of natural land cover types per unit area
Proportion of watershed that has forest land cover
Average forest patch size as a percentage of watershed area
Index of forest connectivity
Proportion of total stream length with forest land cover
Proportion of total stream length with anthropogenic land cover
Number of roads crossing streams per unit stream length
-------
t^^£^L^ss^^^MjJ3^.Mi^J^.^f^i^Mat^.^L.^'^ ^Lzt*._t
-------
Human Use Patterns
One of the simplest and most informative indicators of
environmental impact is the extent to which humans
have changed the natural vegetation to crops or urban
land cover. These indicators are easy to interpret be-
cause profound land cover changes influence almost
every aspect of the environment from wildlife habitat to
soil erosion.
The national maps of human use intensity (Figure 2.3)
show watershed rankings for both total human use
(agriculture plus urban, Figure 2.3a) and agriculture alone
(Figure 2.3b). Urban areas are relatively minor in terms of
total area, and farming areas are more extensive, so the
two maps are very similar. Most of the human land use
has occurred in the central United States and along the
eastern seaboard. Higher elevations and the dry south-
west appear to have been less impacted by conversion to
agricultural or urban land cover. Like most of the eastern
coast, the mid-Atlantic region has a complicated pattern
of land use that deserves more detailed attention.
The chart gives some details about the distribution of
human use intensity among watersheds. About 10%
(200) of the nation's watersheds have been almost com-
National Rank
Quintile Data Range
1 • < 3
2 B 3-12
3 Q 12-34
4 H 34-72
5 • >72
1Q2O30405O6070809O100
Indicator VWu.
Flguro 2.3.
Proportion of watershed area with: (a) agriculture or urban land cover, (b)
agriculture land cover.
-------
.____„„„_
pletely converted to agricultural land. These are located
mostly in the fertile central United States.
About 40% (800) of the watersheds have only small
amounts of agriculture. These watersheds are primarily
located in arid and mountainous areas. Some human
uses of the land are undetectable at this scale. For
example grazing, an important agricultural activity in the
western United States, does not change the grassland
cover type designation at this scale.
©
National Rank
Quintile Data Range
1 • < 3
i 3-11
3 11-31
1 31-69
• >69
10 20 30 40 50 60 70 80 90 100
Indicator Value
-------
The complicated spatial patterns in the mid-Atlantic
region are evident in the map of land cover diversity
(Figure 2.4). The map shows the watershed ranking for
the number of different natural land cover types (anything
except urban and agriculture) per unit area. These
rankings are based on the original 160-class version of
land cover and not the 9-class version shown in Figure
2.1.
The greatest diversity of natural land cover is found in the
western United States, where large changes in elevation
produce different vegetation types at the top and bottom
of the same watershed. But there are also diverse water-
sheds in coastal areas, including parts of the mid-Atlantic
region.
Forest Patterns
Forest patterns are particularly relevant in the eastern
United States because forests are the dominant natural
vegetation cover. In contrast, natural land cover in the
western United States also includes grasslands and
shrublands, so forest patterns alone do not describe
departures from potential natural vegetation types. We
used three different indices of forest pattern in the water-
sheds: amount of forest, average forest patch size, and
forest connectivity. The resulting national rankings of
watersheds for these forest indices are shown in Figure
2.5.
eoo
400
200
0
National Rank
Quintile Data Range
< 0.19
0.19 -0.28
0.29 - 0.42
0.43 - 0.76
1 •
3 Q
4 m
5 • > 0.77
0.1 02 03 0.4 OS 0.8 0,7 0.8 OS 1 l*re
kx&ahx Value
Figure 2.4.
Number of natural land-cover types per 100 square kilometers of
watos/iod area.
-------
^•7 Eco/og/ca/ Assessment of the Un/fecf §ggfs TWtfrAtfanticfiegion Chapter 2 1J23 i
L^^ieJA^^.^^.^1^F^^J^_,J_^^Jife^i.|^^^^
The first map (Figure 2.5a) shows the watershed rankings
of forest area, expressed as the percentage of total
watershed area. Patterns of forest loss are evident along
the east coast, and once again the mid-Atlantic region
has a complicated pattern that will be interesting to
explore in more detail. The chart indicates that about
20% (400) of the nation's watersheds are almost com-
pletely forested, and that about 30% have little forest
cover. About 100 watersheds have no forests at all when
measured at this scale. Forest cover is the most common
vegetation type in nearly all of the watersheds east of the
Ohio River. Many western watersheds are only forested
at higher elevations.
The two other maps are different ways of looking at
whether the forests that do occur in a watershed are
continuous, or fragmented into smaller patches. Figure
2.5b shows watershed rankings of average forest patch
area or size, expressed as a
percentage of total watershed
area. Figure 2.5c shows
watershed rankings of forest
connectivity, defined as the
probability that a randomly-
selected forested spot on the
map is adjacent to another
forested spot.
All three maps have a similar
pattern. Forest cover is usually
continuous where most of the
watershed is forested. In other
cases, such as some water-
sheds in the southwest, forest
cover is a minor component overall, and yet is still con-
tinuous where it does occur.
Compared to potential natural cover conditions, forest loss
and fragmentation of the remainder is significant in the
northeast United States, along the east coast, and in the
Mississippi River valley. The patterns in the mid-Atlantic
region are typical of those found in other places in the
eastern half of the country.
Although the three maps have a similar pattern, the charts
illustrate different views obtained by using different indica-
tors. The distribution of watersheds is more or less
uniform for the indicator based on percentage of forested
area. The charts for the other two indicators suggest that
in most watersheds, the average forest patch is a small
percentage of total area, but that forest cover tends to be
connected in whatever amount actually exists.
-------
legion: Cfta/oterZ
' ""
National Rank
Quintile Data Range
< 2.0
2.0 - 22.0
22.0 - 60.0
60-° - 89-°
>89.0
102030.405080708090100
Indicator VWue
-------
~ ^Jk^?**5*^'
An Ecological Assessment of the Un/fedlstates Mid-AtianticRegion: Chapter 2
National Rank
Quintile Data Range
1 • < 0.1
0.1 - 0.5
o.s - 1.9
1.9-11.6
10 20 30 40 SO 60 ' 70 80 90 100
Indicator Value
600
National Rank
Quintile Data Range
1 • < 0.31
2 H 0.31 - 0.58
3 El 0.59 - 0.78
4 B 0.79 - 0.92
5 • > 0.92
10 20 30 40 50 60 70
Indicator \felue
80
-------
Patterns Affecting Water Quality
Water quality and aquatic life are intimately related to
land cover near streams. The plant life near streams is
referred to as riparian vegetation. It forms an important
buffer zone protecting water quality. Natural vegetation
absorbs agricultural nutrients, slows the rate of water
movement, and is a settling zone for soil particles sus-
pended in runoff. Riparian conditions are often evaluated
within a few meters of a stream, but the larger landscape
context is also important.
One way to measure environmental conditions is to look
at whether streams flow through predominantly forested
or developed landscapes within a watershed. If there are
no large urban areas or agricultural zones anywhere near
streams, then it is less likely that water quality is being
affected by these land uses. If forest cover dominates in
the vicinity of streams, then there is greater opportunity for
forests to buffer the conditions within streams.
Watershed rankings of the proportions of stream length
dominated by different land cover types are shown in
Figure 2.6. These proportions are based on forest cover
(Figure 2.6a) or urban and agriculture cover (Figure 2.6b)
within about one-half kilometer of streams in each
watershed. In the eastern United States, the rankings for
forested riparian zones show a contrast between the
highly developed northeast and the more rural southeast.
The mid-Atlantic is a transition zone between these
regions. Rankings based on the proportion of agriculture
National Rank
Quintile Data Range
1 • < 1.5
1.5 - 22.5
22.5 - 56.0
56.0 - 88.7
>88.7
°' 10 ' 20 ' 30 ' 40 ' so eo 70 80 so 100
IndicalorVnluo
Figure 2.6.
Proportion of total streamlength that is: (a) forested, or
(b) agriculture and urban.
-------
_
"'
or urban land cover in riparian zones show similar pat-
terns in the eastern United States. Many watersheds in
the Upper Mississippi River basin have relatively high
proportions of urban or agriculture land cover near
streams and rivers. The differences are more compli-
cated in the western United States because non-forest
vegetation may also be shrublands or grasslands.
Nationwide, the charts indicate that about 40% of the
watersheds have riparian landscapes that are at least
70% forested, but an equal number of watersheds have
very little forest cover in riparian landscapes. About 10%
(200) of the watersheds have riparian landscapes that
are nearly all agriculture or urban, and about 30% are
almost completely undeveloped.
Spatial variation in land cover near streams and rivers
across the nation suggest some potentially large differ-
ences in sediment and contaminant loadings to streams
and rivers between regions. For example, the Upper
Mississippi River basin has relatively more watersheds
with agricultural and urban riparian zones, and this may
contribute to relatively higher levels of sediment loadings
in the streams and rivers. Large forested areas of the
Appalachian Mountains have high proportions of forested
riparian zones, and relatively little agriculture. Sediment
and contaminant loadings to streams in these areas are
likely lower than in the Upper Mississippi River basin.
©
National Rank
Quintile Data Range
1 •:< 3.4
3.4 - 15.7
15.7 - 37.1
37.1 -74.3
>74.3
10 20 30 40 SO 60 70 80 ' 90 ' 100
Indicator \felue
-------
National Context Summary
Water quality is also related to larger patterns of land use
over entire watersheds. For example, roads near streams
affect water quality not only as direct pollution sources, but
also because they represent paths for rapid runoff. The
frequency of roads crossing rivers was expressed here as
the number of road crossings per unit river length in each
watershed. This expression helps to adjust for differences
in the total length of rivers between watersheds.
The map of watershed rankings for this indicator (Figure
2.7) is complicated, and it does not closely resemble the
national patterns found earlier when looking at land
cover. The mid-Atlantic region, like most of the northeast
and upper midwest, has extensive road networks. The
mountainous areas of the mid-Atlantic have more cross-
ings than would be expected based on land cover alone;
this is so because most roads in the mountains follow
river valleys and can cross the same river many times.
Several important features of the mid-Atlantic region can
be identified by placing it into a national context. The
mid-Atlantic region certainly has complicated spatial
patterns of land cover, and the finer-scale analyses
shown later in this atlas seem warranted. In fact, the mid-
Atlantic region should be an excellent case study area
because of the variety of conditions that it contains.
Some patterns in the mid-Atlantic region are typical of
other areas along the eastern seacoast. This means that
what is learned in the mid-Atlantic may be applicable in
other regions. Because the mid-Atlantic is also a transi-
tion zone between regions of more or less impact to the
north and south, further studies here may also be relevant
to environmental monitoring in these other areas.
123 45 6 7 8 9 10 Mm
National Rank
Quintile Data Range
1 • <1.4
2 B 1.4 - 2.2
3 Q 2.2 - 3.0
4 H 3.0 - 4.2
5 • >4.2
Figure 2.7,
Number of road-stream crossings per 100 kilometers of streams.
-------
The mid-Atlantic is probably not the most highly impacted
region in the eastern United States, but it is different from
the less impacted areas that are found at higher elevations
in the south and west. The complexity of patterns in the
region creates an opportunity to consider a full range of
environmental strategies from restoration of the more
developed areas to protection aimed at particular re-
sources such as high-elevation forests or wetlands. This
brief look at the mid-Atlantic region in a national context
has confirmed that many broad-scale aspects of environ-
mental quality can be explored here.
-------
-------
Chapter 3: The Mid-Atlantic Region
This chapter illustrates some landscape indicators that
can be used to assess watershed conditions in the mid-
Atlantic region (Table 3.1). The comparative assessment
that is reported in the next chapter is based on these
indicators. Each environmental measure is discussed
separately, with maps to illustrate the relative rankings or
groupings of watersheds and charts to show the distribu-
tions of indicator values. The Appendix lists the actual
indicator values obtained for every watershed. The
methods and data used to construct the maps and charts
are briefly described in this chapter, and more detailed
methods are provided in the Appendix.
We begin by looking at the biophysical setting of the
mid-Atlantic region and present maps of the data used
to calculate indicator values. Included are regional
pictures of topography, rivers, watershed boundaries, and
land cover. An important criterion when choosing digital
data was consistency across the region. Consistency is
essential because the goal is a regional comparative
assessment, and many detailed maps of
relatively small areas were not
used for this reason.
Figure 3.1
Shaded relief map of the mid-Atlantic region. Source: U.S. Geological Survey,
Digital Elevation Model, 3 arc-second.
The indicators are grouped according to four themes —
people, water, forests, and landscape change. These
groups are subjective, and any given indicator could be
relevant to more than one theme. For example, an
indicator of forest cover along streams appears as part of
the "water" theme, but it also describes certain aspects of
forest condition as well as human impacts. The conclud-
ing section about landscape change is based on analysis
of satellite imagery over a 15-year period from the 1970s
to the 1990s.
Biophysical Setting of the
Mid-Atlantic Region
The mountains, valleys, and coastal plains form the
backdrop for all of the physical and biological processes
that shape the region. When you look at a map of the
region — whether it is a physical map, a vegetation map,
or even a socio-political map — the most striking fea-
tures of the landscape are created by topographic varia-
tion (Figure 3.1). The variety of the different
physiographic regions — Blue Ridge Mountains, Ridge
and Valley, Coastal Plain — creates one of the most
diverse physical and ecological regions in the nation.
In the western section of the mid-Atlantic, the Appala-
chian Mountains rise thousands of feet to dominate the
landscape for hundreds of miles in any direction. The
great valley of the Appalachians, stretching from Penn-
sylvania to Alabama, provides fertile agricultural lands
and gently sloping areas for human development. To the
east of the Appalachians, the coastal plain stretches to
the Chesapeake Bay, one of the most important natural
resources of the mid-Atlantic region. The estuarine and
wetland habitats surrounding the bay are associated with
lowland areas and slowly draining soils which have been
washed from the western mountains.
Topography and soils have a direct and dramatic effect
on the biological character of the mid-Atlantic region.
The diversity of plants and animals is tied to variations in
sunlight and moisture, the basic building blocks for
ecological communities. The amount and timing of
sunlight varies from one hillside to the next, depending
on the direction, or aspect, of the slope. In the northern
hemisphere, south-facing slopes receive more sunlight
than northern faces and this, in turn, causes differences
in available energy and soil moisture. The depth and
nutrient content of the soils themselves also influence
available resources. All of these variations are reflected
in the arrangement of plant and animal communities that
respond to ecological conditions. Even in the relatively
-------
flat coastal plain, a difference of just a few feet of eleva-
tion or a slightly different aspect can dramatically change
the species composition of an ecosystem. Topography is
one of the most important considerations for any land-
scape ecological assessment.
Streams and rivers direct the flow of water across the
landscape and are a dominant feature in the mid-Atlantic
region (Figure 3.2). In addition to carrying water, streams
transport sediment and nutrients that replenish
downstream areas. Because streams also transport
pollutants, it is important to look both upstream and
downstream when assessing water quality issues. The
connected nature of the stream network requires us to
examine not only the streams of immediate concern, but
also how those streams fit into the regional picture of
streamflow and water quality.
Table 3.1
POPDENS
POPCHG
UlNDEX
RDDENS
NO3DEP
SO4DEP
OZAVG
RIPFOR
RIPAG
STRD
DAMS
CROPSL
AGSL
STNL
STPL
PSOIL
FOR%
FORFRAG
EDGE?
EDGE65
List of landscape indicators used to assess watershed conditions in the mid-Atlantic region.
(The abbreviations are used in tables of indicator values in Chapter 4 and in the Appendix.)
Population density (number of people per EDGE600
square kilometer)
Population change (percentage change from /A/77
1970 to 1990)
Human use index (proportion of wateshed area /A/765
with agriculture or urban land cover)
Road density (average number of kilometers of INT600
roads per square kilometer of watershed area)
Average annual wet deposition of nitrate INTALL
(1987 and 1993)
Average annual wet deposition ofsulfate FORDIF
(1987 and 1993)
Average annual value of the W126 ozone
index (1988 and 1989) NDVIDEC
Proportion of total streamlength with adjacent
forest land cover NDVIINC
Proportion of total streamlength with adjacent
agriculture land cover NDVITOT
Proportion of total streamlength that has roads
within 30 meters 1STDEC
Number of impoundments per 1000 kilometers
of stream length
Proportion of watershed with crop land cover
on slopes that are greater than three percent 1STINC
Proportion of watershed with agriculture land
cover on slopes that are greater than
three percent
Potential nitrogen loadings to streams 1STTOT
Potential phosphorus loadings to streams
Proportion of watershed with potential soil loss
greater than one ton per acre per year
Percent of watershed area that has forest ND VI3%
land cover
Forest fragmentation index
Proportion of watershed area with suitable
forest edge habitat (7 hectare scale)
Proportion of watershed area with suitable
forest edge habitat (65 hectare scale)
Proportion of watershed area with suitable
forest edge habitat (600 hectare scale)
Proportion of watershed area with suitable
interior forest habitat (7 hectare scale)
Proportion of watershed area with suitable
interior forest habitat (65 hectare scale)
Proportion of watershed area with suitable
interior forest habitat (600 hectare scale)
Proportion of watershed area with suitable
interior forest habitat at three scales
Departure of the largest forest patch size from
the maximum possible for a given amount of
anthropogenic land cover
Decrease in normalized difference vegetation
index from 1975 to 1990
Increase in normalized difference vegetation
index from 1975 to 1990
Total change in normalized difference
vegetation index from 1975 to 1990
Difference between observed and expected
decreases in normalized difference vegetation
index from 1975 to 1990 in first-order
stream regions
Difference between observed and expected
increases in normalized difference vegetation
index from 1975 to 1990 in first-order
stream regions
Difference between observed and expected
total change in normalized difference
vegetation index from 1975 to 1990 in
first-order stream regions
Proportion of watershed with normalized
difference vegetation index decreases from
1975 to 1990 on slopes greater than
three percent
-------
Figure 3.2.
Streams and water bodies in the mid-Atlantic region.
Source: U.S. Environmental Protection Agency, River Reach File Version 3
(RF3), derived from U.S. Geological Survey Digital Line Graph - streams,
1:100,000-scale.
-------
The landscape indicators in this atlas are summarized by
watersheds, using the national map of watershed bound-
aries that was shown in Chapter 2. A subset of that map
covering just the mid-Atlantic region (Figure 3.3) was
used to summarize the landscape indicators in this
chapter. The figure illustrates one of the problems in
using naturally-defined units such as watersheds to
assess conditions over politically-defined units like
states — parts of some watersheds lie outside of the
assessment region. As a result, the indicators calculated
for these watersheds are probably not as reliable as the
indicators calculated for watersheds that had complete
data coverage.
Figure 3.3.
Watershed boundaries within the mid-Atlantic region. The numbers are
USGS hydrologtc unit codes (HUCs). See Table 3.2 for watershed names.
Source: U.S. Geological Survey, Hydrologic Unit Code Boundaries (HUC250),
1:250,000-scale.
-------
Table 3.2 Watershed names
2040101 Upper Delaware 2070004
2040103 Lackawaxen 2070005
2040104 Middle Delaware- 2070006
Mongaup-Brodhead 2070007
2040105 Middle Delaware- 2070008
Musconetcong 2070009
2040106 Lehigh 2070010
2040201 Crosswicks-Neshaminy
2040202 Lower Delaware 2070011
2040203 Schuylkill 2080102
2040205 Brandywine-Christina 2080103
2040207 Broadkill-Smyrna
2050101 Upper Susquehanna 2080104
2050103 Owego-Wappasening 2080105
2050104 Tioga 2080106
2050105 Chemung 2080107
2050106 Upper Susquehanna- 2080108
Tunkhannock 2080109
2050107 Upper Susquehanna- 2080110
Lackawanna 2080201
2050201 Upper West Branch 2080202
Susquehanna 2080203
2050202 Sinnemahoning 2080204
2050203 Middle West Branch 2080205
Susquehanna 2080206
2050204 Bald Eagle 2080207
2050205 Pine 2080208
2050206 Lower West Branch 3010101
Susquehanna 3010102
2050301 Lower Susquehanna- 3010103
Penns 3010104
2050302 Upper Juniata 3010105
2050303 Raystown 3010106
2050304 Lower Juniata 3010201
2050305 Lower Susquehanna- 3010202
Swatara 3010203
2050306 Lower Susquehanna 3010204
2060002 Upper Chesapeake Bay 3010205
2060003 Gunpowder-Patapsco 3040101
2060004 Severn 4120101
2060005 Choptank 4130002
2060006 Patuxent 5010001
2060007 Blackwater-Wicomico 5010002
2060008 Nanticoke 5010003
2060009 Pocomoke
2060010 Chincoteague 5010004
2070001 South Branch Potomac 5010005
2070002 North Branch Potomac 5010006
2070003 Cacapon-Town
Conococheague-Opequon 5010007
South Fork Shenandoah 5010008
North Fork Shenandoah 5010009
Shenandoah 5020001
Middle Potomac-Catoctin 5020002
Monocacy 5020003
Middle Potomac- 5020004
Anacostia-Occoquan 5020005
Lower Potomac 5020006
Great Wicomico-Piankatank 5030101
Rapidan-Upper 5030102
Rappahannock 5030103
Lower Rappahannock 5030104
Mattaponi 5030105
Pamunkey 5030106
York 5030201
Lynnhaven-Poquoson
Western Lower Delmarva 5030202
Eastern Lower Delmarva 5030203
Upper James 5050001
Maury 5050002
Middle James-Buffalo 5050003
Rivanna 5050004
Middle James-Willis 5050005
Lower James 5050006
Appomattox 5050007
Hampton Roads 5050008
Upper Roanoke 5050009
Middle Roanoke 5070101
Upper Dan 5070102
Lower Dan 5070201
Banister 5070202
Roanoke Rapids 5070204
Nottoway 5090101
Blackwater 5090102
Chowan 6010101
Meherrin 6010102
Albemarle 6010205
Upper Yadkin 6010206
Chautauqua-Conneaut
Upper Genesee
Upper Allegheny
Conewango
Middle Allegheny-
Tionesta
French
Clarion
Middle Allegheny-
Redbank
Conemaugh
Kiskiminetas
Lower Allegheny
Tygart Valley
West Fork
Upper Monongahela
Cheat
Lower Monongahela
Youghiogheny
Upper Ohio
Shenango
Mahoning
Beaver
Connoquenessing
Upper Ohio-Wheeling
Little Muskingum-
Middle Island
Upper Ohio-Shade
Little Kanawha
Upper New
Middle New
Greenbrier
Lower New
Gauley
Upper Kanawha
Elk
Lower Kanawha
Coal
Upper Guyandotte
Lower Guyandotte
Tug
Upper Levisa
Big Sandy
Raccoon-Symmes
Twelvepole
North Fork Holston
South Fork Holston
Upper Clinch
Powell
-------
~-^^"^—
Land cover is the product of human land uses on the
backdrop of the biophysical setting. A map of land cover
is essentially a picture of the dominant vegetative, water,
or urban cover in an area. The map of land cover in the
mid-Atlantic region (Figure 3.4) is based primarily on
images taken by the Landsat satellite (Thematic Mapper)
earlier in this decade. The land cover map was prepared
by the Multi-Resolution Land Characterization (MRLC)
project, a Federal effort to create similar maps for the
•Water
•Residential
• Urban
M Pasture
BROW Crops
CProbable Row Grops^
•conifer Forest
•Mixed Forest
•Deciduous Forest
•Woody Wetlands
•Emergent Wetlands
@ Quarry
•coal Mine
Beeach
•Transitional
entire country. The resolution of the land cover data is
30 meters, so each pixel (picture element) represents an
area about the size of the infield of a major league
baseball park. Although individual pixels are far too small
to be rendered accurately here, the visual impression of
broad-scale regional patterns is readily apparent.
Flguro 3.4.
Land cover in the mid-Atlantic region, ca. 1990. Source: Multi-Resolution Land
Characteristics (MRLC) Consortium, derived from Landsat Thematic Mapper
(TM) data, 30 meter resolution (shown here at 90 meter resolution).
-------
The two most dominant land-cover types in the mid-
Atlantic region are forest, which covers about 70% of the
area and agriculture, which covers about 25% of the
area. Most of the watersheds are primarily forested, and
some approach complete forest cover (Figure 3.5). Only
a few watersheds have less than one-third forest cover.
Where forests have been removed, agriculture and urban
land-cover become more dominant, yet they are rarely as
extensive as forest in terms of total cover. The median
amount of urban land cover per watershed is about 2%,
and only five watersheds have more than 25% urban
land cover. Agriculture is an extremely important land
use in the region, yet only six watersheds have more
than 50% of that land cover overall, and the median
amount is only about 25%.
Water
Urban
Agriculture
Forest
Other
Figure 3.5
Proportion of forest, agriculture, urban, water, and other land cover types for
watersheds in the mid-Atlantic region.
-------
Humans in Landscapes in the
Mid-Atlantic Region
Humans structure the landscape for their purposes, and
landscapes constrain human activities. For example,
humans may decide the shapes and sizes of individual
agricultural fields, but regional patterns of topography,
soils, and geology determine if there can be fields at all.
Because human-ndominated landscapes are used for
different purposes which impose different patterns, land
use history is always important for understanding local
landscapes. The interplay between humans and land-
scapes has created a tapestry of multi-scale patterns in
the mid-Atlantic region, and combinations of these two
factors influence the sustainability of ecological pro-
cesses that maintain a high quality environment.
Figures 3.6 through 3.16 illustrate some of
these patterns.
Census data and road maps were used to
translate county-level population values to
watershed-level estimates. The indicator
value is the number of people per square
kilometer.
Quintile Data Range
(PopVSq.Km.)
Population Density and Change
According to the United States Census Bureau, the
population of the mid-Atlantic region in 1990 was about
26,000,000 people, which represents about 10% of the
total population of the United States. The watershed
rankings for population density (people per unit area) are
illustrated in Figure 3.6. As would be expected, the
watersheds with the highest density of people are
2
3
4
5
< 37
37-67
67 - 111
111 - 280
> 280
40 80 120 160 200 240 280 320 360 400 Mora
Indicator Val JB
Figure 3,6
Population density In the mid-Atlantic region. Source: U.S. Census Bureau,
1990 census.
-------
-------
>' r
iS™. ""
located around the larger metropolitan areas — Norfolk,
Washington-Baltimore, Philadelphia, and Pittsburgh —
and the watersheds with the least density are located in
rural and mountainous areas. The chart indicates a
median population density of about 70 people per square
kilometer. The watershed with the highest population
density has over 3,500 people per square kilometer,
while the watershed with the lowest population density
has about 11 people per square kilometer.
Between 1970 and 1990, the population density in-
creased in some watersheds and decreased in others
(Figure 3.7). The median change for all watersheds was
an increase of about 10 people per square kilometer, and
The total population was estimated for each
watershed for 1970 and 1990. The percentage
change was determined by subtracting the 1970
value from the 1990 value, and then dividing the
result by the 1970 value.
Qulntile Data Range
(Pet Change)
1 | <-1.0
2 | -1.0 - 3.0
3 |HJ 3.0 - 18.0
4 • 18.0 - 38.0
5 • >38.0
the extreme values were a gain of 197 and a loss of 15
people per square kilometer. A quarter of the watersheds
in the region had either no change or a reduction in
population. The most noticeable losses in population
densities occurred in watersheds near the Pittsburgh
metropolitan area. Some of the larger gains were in
watersheds just outside of the Baltimore-Washington
area, and in a few watersheds on the DelMarVa peninsula
east of the Chesapeake Bay which includes parts of
Delaware, Maryland, and Virginia.
0 20 40 60 80 100 120 1400 160 180 More
Indicator Vaba
Figure 3.7
Population change (1970 to 1990) in the mid-Atlantic region. Source: U.S.
Census Bureau, 1970 and 1990 census.
-------
Human Use Index
The proportion of an area that is urbanized or used for
agriculture is a measure of human use known as the
U-index. We often assume that humans tend to simplify
' their environment, because agricultural fields and urban
areas, for example, are less complex than the natural
land cover that they replace. At landscape scales,
however, the map of human land use displays compli-
cated patterns (Figure 3.8). The scale at the transition
from simple to complicated patterns might be a measure
of the scale to which humans have structured a land-
scape, or conversely, the scale at which geophysical
processes constrain human activity. By looking at re-
gional patterns of the U-index, it is possible to identify
those areas which have experienced the greatest
land cover conversion from forest cover that
historically dominated the region.
Low
Figure 3.8
Surface map of the human use index (U-index) in the mid-Atlantic region. The
map shows the percentage of urban and agriculture land cover within 65-hectare
windows.
-------
The regional pattern of human use is reflected in the
watershed rankings over the region (Figure 3.9). The
accompanying bar chart shows that the highest U-index
value for a watershed is about 70%, which means that
70% of that watershed has agriculture or urban land-
cover. The lowest value is about 3%, and the median
value is about 30%. The proportion of area with urban or
agriculture land cover exceeds 50% in about 15 water-
sheds, and about the same number of watersheds have
U-indices less than 10%.
Roads
Roads and other transportation corridors are designed to
connect the human-dominated elements of a landscape.
The network of roads in the mid-Atlantic region permits
access, commerce, and communication throughout the
region. Roads also impact the connectivity of ecosys-
tems, and ecosystem connectivity influences the dis-
persal of plants and animals. Sometimes roads restrict
dispersal, as in the case of animals that are unable to
cross roads, and sometimes they enhance it, in the case
of plant species that spread along disturbed roadsides.
The human use index is the percentage of
total watershed area that has either urban or
agriculture land cover.
Quintile
<16J
2 | 16.7 - 26.0
3 fjj 26.0 - 32.5
4 H 32.5 - 43.4
5 • >43.4
Figure 3.9
The human use Index in the mid-Atlantic region.
-------
The influence of any road extends for some distance,
depending on factors such as road size, traffic volume,
and type of use. There are few places in the mid-
Atlantic region that are entirely free of their influence.
According to the road maps used in this atlas, there are
about 700,000 kilometers of roads in the mid-Atlantic
region. This dataset includes all types of roads catego-
rized by the United States Geological Survey as Class
1 (Interstates, United States highways) through
Class 4 (minor roads and city streets). Placed
end to end, these roads would circle the
Earth more than 17 times. It is no wonder
that roads are one of the most important
human features in the mid-Atlantic landscape
today.
In fact, there are so many roads that a detailed
regional map cannot be shown on a single page.
Instead, the regional distribution of roads is
indicated (Figure 3.10) by using a coarse-scale
indicator of relative road density. On this map,
bright colors indicate places that have higher road
density, and dark colors indicate places with lower
road density. It is immediately apparent that
roads are not distributed uniformly throughout the region;
their locations are directly comparable to the maps of
population, elevation, and land cover shown earlier.
There are more roads in urban areas than in rural areas,
and there are more roads in the eastern
half of the region than in the
western half.
Low
Figure 3.10
Surface map of road density in the mid-Atlantic region. The map shows an
estimate of total length of roads within each square-kilometer window. Source:
U.S. Geological Survey, 1:100,000-scale Digital Line Graphs— Transportation.
-------
There are concentrations of roads along the corridors
that link urban areas, and areas with the fewest roads
show up as dark patches, especially along the higher-
elevation ridges. The watershed rankings of road density
over the region (Figure 3.11) further illustrates this pat-
tern. Watersheds surrounding urban areas have the
highest concentration of roads, while mountainous
watersheds in the west have the lowest.
Air Pollution
Air pollution is truly a regional phenomenon, because air
does not stop at political boundaries. It is one of the
more important human-caused stresses in the mid-
Atlantic region. Air pollution presents a changing spatial
pattern over the landscape as pollution sources and
circulation patterns change over time. The atmosphere
interacts with the terrestrial watersheds below in many
ways.
Road density was calculated as length of
road (kilometers) per total watershed area
(square kilometer) using a clipping procedure.
Quintile
123456789101*16
Indicator \telut
Figure 3.11
Road density In the mid-Atlantic region.
-------
Figures 3.12 through 3.15 show regional patterns of
estimated annual sulfate and nitrate deposition and
cumulative annual ozone concentrations, averaged over
several years. Nitrate deposition appears to increase
from south-to-north within the region; the highest levels
are found in central Pennsylvania, extending east to the
Pocono Mountains and south through western Maryland
and northern West Virginia along the ridges of the Appa-
lachian Plateau. The lowest levels occur in central and
southern Virginia, and along the extreme western portion
of the region. This regional pattern of nitrate deposition
may reflect prevailing winds from the west that carry air
pollutants from other regions.
Terrestrial features can influence pollution deposition
indirectly. For example, air pollution deposition models
predict a topographic trend as well as a south-to-north
trend in the region. Estimated nitrate wet deposition
(Figure 3.12a) and sulfate wet deposition (Figure 3.12b)
are greater at higher elevations because topographic
features influence the deposition of rain and fog which
carry these pollutants dissolved in water droplets. The
surface maps of sulfate and nitrate deposition appear
almost identical because the measurement scales have
been normalized to make it easier to see relative trends
across the region. This also helps to highlight the topo-
graphic effect on air pollution deposition. However, the
two maps are different enough to yield slightly different
watershed rankings for nitrate (Figure 3.13) and sulfate
(Figure 3.14) deposition.
The map of watershed rankings for tropospheric (surface)
ozone (Figure 3.15) demonstrates that not all pollution
indicators follow the same regional pattern. The ozone
index is not closely associated with topography, but
rather with the distribution of urban and agricultural
areas. This map was prepared from extremely coarse-
scale information, and the surface map for this index (not
shown here) has only several dozen pixels for the entire
region. Even at this coarse scale, there are obvious
differences across the mid-Atlantic.
©
Figure 3.12,.
Surface maps of estimated average annual wet deposition of (a) nitrate, and (b) I
sulfate for 1987 and 1993 in the mid-Atlantic region. Source: J. Lynch,
Pennsylvania State University.
-------
Average annual wet deposition of nitrate was estimated from data
for 1987 and 1993 by resampling the original maps at 90-meter
resolution and then finding average pixel values for each watershed.
Quintile Data Range
(Kg./Ha.x100)
1
2
3
4
5
<1213
1213 - 1456
| 1456 - 1677
| 1677 - 1965
I >1965
£00
200 400 800 800'1O» 1200 MOO WOO 1EOO 20002200
Indicator Valuo
Figure 3.13
Average annual wet deposition of nitrate in the mid-Atlantic region (average of 1987 and 1993).
Average annual wet deposition of sulfate was estimated from data
for 1987 and 1993 by resampling the original maps at 90-meter
resolution and then finding average pixel values for each watershed.
Quintile Data Range
(Kg./Ha.x100)
1 | <1978
2 | 1978 - 2242
3 ffl 2242 - 2562
4 | 2562 - 2814
5 • >2814
400 S001200W0020002«02a0032003H)0«00
ImScatorValuo
Flguro 3.14
Average annual wet deposition of sulfate in the mid-Atlantic region (average of 1987 and 1993).
-------
Figure 3.15
Average annual value of the W126 ozone index
(1988 and 1989) in the mid-Atlantic region.
Source: U.S. Environmental Protection Agency.
Average annual ozone exposure was
estimated from data for 1988 and 1989 by
resampling the original maps at 90-meter
resolution and then finding average pixel
values for each watershed
Quintile Data Range
(W126 Index)
1 | <39
2 g| 39-41
3 |g| 41-44
4 f| 44-48
5 • >48
30 35 40 45 50 55 60 65 70 75
Indicator Value
Air pollution disperses
quickly over large
regions.
-------
Landscape Units
Landscape analysis methods provide the opportunity to
look at regional patterns of land use at a range of scales.
Maps of land cover are created one pixel at a time, which
ignores some of the information about local-scale pat-
terns of land cover. By recognizing these local patterns,
new landscape map themes can be created which
suggest the types and intensities of human activities that
are occurring in a given place. For example, if you are
standing in a spot that is forested, and if most of the
spots around you are also forested, then it is likely that
you are in a part of the landscape that has a general land
use or activity theme of "forest." If, however, your for-
ested spot is embedded in a pattern of forest and
agriculture, then it is more likely that your part of
the landscape has the land use theme of
"rural agriculture" instead of "forest." But
land cover alone is not always an accurate
guide to actual land use. For example, if
the pattern surrounding your forested spot is
mainly urban, then you might be standing in
an area with a "city park" theme, but (without
more information) it could just as easily be an
area that is planned for "future development."
A landscape may be described by the relative
proportions of forest, agriculture, and developed
(urban) land cover it contains. The map of landscape
units for the mid-Atlantic region (Figure 3.16) has 19
classes, labeled with combinations of the letters F, A, and
D, referring to forest, agriculture, and developed land
cover. The labels are interpreted as follows. An upper-
case letter indicates an area with more than 60% of that
land cover, and a lower-case letter indicates an area with
less than 40% of that land cover. The ordering of letters
corresponds to the relative amounts of land cover in an
area. If a land cover is less than 10% of an area, the
corresponding letter is left out.
• A
• D
• F
QA/d
BA/d
BA/f
BD/a
BF/a
HF/d
BA/d/f
BA/f/d
B D/a/f
B D/f/a
B F/a/d
B F/d/a
Bad
EJaf
Bdf
Badf
B No Data
Figure 3.16.
Landscape units In the mid-Atlantic region (see text for explanation).
-------
Compare this map with the map of land cover presented
earlier (Figure 3.4). Some individuals with a practiced eye
can perform the "mental blending" of land cover propor-
tions that the computer did to create the map of land-
scape units. The phenomenon is similar to some
people's ability to see the pattern hidden in 3-D stereo-
grams. For many people, however, it helps to use a
computer to extract the "hidden" information. By simpli-
fying the patterns into these landscape units, we can
identify zones of human use that are difficult to see using
the land cover map alone.
Water
Everyone knows the importance of water. But many
people do not realize how much its quality depends on
the surrounding landscape. Water quality, like landscape
condition, is an integrated response to environmental
stress and land management practices at watershed
scales.
This section presents landscape indicators that are
related to water quality in the streams of the mid-Atlantic
region. "Riparian" indicators describe landscape condi-
tions near streams and "watershed" indicators describe
conditions over entire watersheds. The riparian indica-
tors include measures of human activities near streams.
The size and amount of riparian buffers along
streambanks is an important determinant of soil loss and
sediment movement, which in turn affect water quality.
The group of watershed indicators presented here prima-
rily measure the potential for soil and nutrient losses from
surrounding landscapes which would ultimately be
deposited in streams. Put simply, watersheds covered
by forests are likely to be in better condition than water-
sheds with high percentages of intensive land uses.
Because intact riparian areas buffer streams from the
potentially adverse effects of watershed-scale events like
erosion, both types of indicators need to be evaluated
when considering overall landscape influences on stream
condition and water quality. The interplay of processes
operating over the entire watershed with processes in the
riparian zone will ultimately determine the condition of
streams in the mid-Atlantic.
Riparian Indicators
The vegetation along a stream influences the condition of
both the stream bank and the water in the stream. This
strip of vegetation, known as the riparian zone, is com-
monly described by the types of vegetation it contains
-------
and by the presence of roads and other human activities.
In an ideal situation, many pollutants and fertilizers will be
Intercepted or absorbed by the riparian vegetation, and
this helps to keep the streams healthy. Bank erosion is
also mitigated by intact riparian vegetation. The condi-
tions of the riparian ecosystem over a whole watershed
can be studied in order to learn where, for example, a
restoration project would most improve water quality.
Similarly, a characterization of riparian conditions over the
entire mid-Atlantic region can help to identify which
watersheds are most likely to see improved water quality
as a result of riparian improvements.
Forest and Agricultural Land Cover
Along Streams
Forested riparian zones
are a natural part of the
healthiest stream ecosys-
tems in the eastern United
States. They provide an
effective barrier to runoff of
water, pollutants, and excess
fertilizer, and support a variety
of valuable plant and wildlife
species. Conversely, when
forests are removed right up to
the stream, the riparian zone
not only loses its natural
buffering capacity but now
becomes a potential
source of pollution
and excess fertilizer.
Agricultural prac-
tices usually
employ fertilizers,
pesticides, and
other chemicals
that are essen-
tial to crop growth
and yield. These
chemicals can more
readily be moved into
streams which flow
through agricul-
tural fields, in
comparison
FIguro 3.17
Proportion of total streamlength with adjacent forest land cover in the mid-
Atlantic region.
to streams which flow through forests. The maps on
these pages illustrate differences among watersheds in
the length of stream that has either forest or agriculture
cover in the riparian zone.
Figure 3.17 shows the relative percentage of stream
length in each watershed that has forested riparian
zones. The urban areas of eastern Pennsylvania,
Maryland, and northern Virginia have the least percent-
age of forest in riparian zones. Western Pennsylvania,
southeastern Virginia, and portions of West Virginia have
the greatest percentage of forested
riparian cover. The chart
indicates that all watersheds in
the mid-Atlantic region have at
least 50 percent of their total
The proportion of total
streamlength with forest
cover was determined for
each watershed by
overlaying land-cover and
stream maps. The index
value is the total length of
stream with forest land-
cover, divided by the total
length of all streams in
the watershed.
Quintile
2
3
4
5
Data Range
(Percent)
<70.6
70.6 - 76.8
76.8 - 84.6
84.6 - 89.9
>89.9
10 20 30 SO 60 ' 70 80 90 100
Indicator Value
-------
stream length in forest cover, and that over half of the
watersheds have at least 80% riparian forest cover.
Whereas the distribution of riparian forests is an indicator
of natural buffering capacity, the distribution of agricul-
tural land cover in riparian zones is an indicator of poten-
tial problems. Figure 3.18 shows the percentage of
stream length in each watershed that has agricultural
land cover in the riparian zone. Because forest and
agriculture are the two most common land cover types in
the mid-Atlantic region, this figure is almost the inverse
of the riparian forest map. But not quite, because
agriculture is not the only non-forest land use in the
region. Although only a handful of watersheds
have more than 30% of their
stream length with agricul-
tural land cover, every
watershed has at least
some agriculture in the
riparian zone. The water-
sheds with the highest poten-
tial for negative impacts are in
eastern Pennsylvania, Mary-
land, and northern Virginia.
Roads Along Streams
Roads affect stream water in
many ways and roads in close
proximity to streams have
the most potential for
adverse effects on
stream water quality.
Since roads have an
impervious surface,
and ditches are
built to channel
water from
roads into
streams, the rate
of water runoff is
higher where there are
more roads. This
contributes to
increased scouring
of streambanks
and channel
Figure 3.18
Proportion of total streamtength with adjacent agriculture land cover in the
mid-Atlantic region.
alteration. Although large spills of pollution are rare and
often quickly contained, small spills of petroleum prod-
ucts, antifreeze,.and other vehicle-related chemicals
happen every day on every mile of road in the region.
These small spills eventually go somewhere, usually into
streams. Road construction near streams is a temporary
stress on water quality, but after construction, the road-
sides remain. Routine maintenance, including salting
during the winter, can increase pollution and sediment
loadings to streams and contribute to poorer water
quality. Cumulatively, these changes can reduce water
quality and fish habitat suitability in
streams. For these and other
, reasons it is important to
consider how the proximity of
roads to streams might influ-
ence regional water quality.
The proportion of total
streamlength with
agriculture land-cover
was determined for each
watershed by overlaying
land cover and stream
maps. The index value is
the total length of stream
with agriculture land-
cover, divided by the total
length of all streams in
the watershed.
Data Range
(Percent)
70 80 90 '100
Indicator Value
-------
Watershed Indicators
The regional pattern of roads along streams (Figure 3.19)
may be surprising because many of the watersheds with
a high incidence of roads along streams are located in
remote areas which do not have a lot of roads in com-
parison to the rest of the mid-Atlantic region. The expla-
nation lies in the topography of the region. Road
construction is more difficult in steeper topography and,
as a result, the roads are often located in the relatively flat
areas along streambeds. Furthermore, the highly dis-
sected topography that is characteristic of the Appala-
chian Plateau often forces the roads to cross streams
several times in a short distance. So while there may be
relatively fewer roads in these areas, they are nearly all
located adjacent to streams and hence have relatively
high values for the indicator. Certainly traffic volume and
the type of traffic will also influence the actual impacts of
roads near streams, but such information is generally not
available in a format suitable for regional-scale
analyses.
The proportion of streamlength within 30
meters of a road was calculated by
overlaying maps of streams and roads.
The index value is the proportion of total
streamlength in a watershed within 30
meters of a road
While streamside conditions are important, it is also
important to have indicators of potential impacts on
water quality from sources throughout the watershed. It
was mentioned earlier that the watershed indicators
presented here are primarily concerned with soil erosion
and runoff processes. These indicators are relatively
easy to determine from regional databases. In any case,
erosion processes are extremely important. The results
of increased erosion may include reduced agricultural
productivity, reduced storage capacity of lakes and
reservoirs, increased water treatment costs, introduction
of pesticides and fertilizers to water sources, loss of
habitat for fish and other species,
and reduced recreation
potential.
Quinlile
1
2
3
4
5
Data Range
(Percent)
<2.8
2.8 - 4.6
4.6 - 6.2
6.2 - 8.3
>8.3
2 4 6 8 10 12 14 16 18 20
ImScatof Value
Figure 3.19
Proportion of total stream length that has roads
within 30 meters in the mid-Atlantic region.
-------
Impoundments
There are three major reasons why dams are built. They
provide a stable water supply for human uses, they
control flooding, and they channel water through genera-
tors which produce electricity. While all of these are good
reasons, it is unfortunate that an essential feature of dam
design is to disrupt all of the natural processes associ-
ated with stream flow. Dredges operating to keep river
channels open for navigation are evidence that sediment
is deposited behind dams, and fish ladders around dams
demonstrate the direct effects on wildlife populations.
Dams have other, less well-known effects. For example,
many dams are built to raise the water 'head' or pressure
behind the dam, and water is tapped from the deepest
parts to generate electricity. Deep water usually contains
less dissolved oxygen than surface water, and this can
impact life in streams below the dam unless special
measures (for example, adding oxygen) are taken.
Considering all major impoundments, dams are relatively
abundant in the mid-Atlantic region (Figure 3.20). They
are distributed throughout the region, with a surprising
number occurring on flatter topography in the coastal
plain. The pattern may be surprising because most
people associate dams with mountainous terrain, but
only the largest, electricity-producing structures are built
in those areas, and the largest dams are the ones most
people see. Figure 3.21 shows the watershed rankings
for the number of dams per 1,000 kilometers of stream
length. The values range from 0 (16 watersheds have
Figure 3.20
Locations of large water impoundments in the
mid-Atlantic region. Source: U.S. Geological
Survey.
-------
no large dams) to about 85 dams per 1,000 kilometers of
stream. The watersheds with the highest density of
dams are in the northeastern portion of the region,
particularly along the Delaware River. There are also
some high densities in watersheds in southeastern
Virginia.
Agriculture on Steep Slopes
Unless special measures are taken, agriculture tends to
increase soil erosion, which ultimately deposits sediment
in streams and lakes. Potential soil erosion from crop-
land is related to the steepness of slopes being cultivated
and the farming methods used. Percent slope is a
The number of impoundments in each
watershed was divided by the total stream
length for the watershed to estimate the
density of impoundments.
measure of steepness that is calculated as the ratio of
vertical rise in elevation per horizontal distance traveled.
The U.S. Department of Agriculture has classified slopes
into six categories. Based on this classification, slopes
greater than 3% have a greater risk for soil erosion. For
comparison, a 3% slope is about half as steep as the
steepest hill on which roads are built.
Figure 3.22 illustrates the watershed rankings for the
percentage of watershed area that has crop land cover
on slopes greater than 3%. Figure 3.23 shows the same
Quintile Data Range
(Dams/1,000 Km.)
1 0 < 2.4
2 | 2.4 - 4.9
3 H 4.9 - 10.1
4 B 10.1 - 15.9
5 • >15.9
10 20 30 40 50 60 70 80 90 100
Indicator Value
Figure 3.21
Number of impoundments per 1,000
kilometers of stream in the mid-Atlantic
region.
-------
picture, but this time considering all agriculture land cover
types (crops plus pasture). Crops are typically cultivated,
a practice that removes ground cover, exposes soil, and
makes surface erosion more likely on crop lands in
comparison to pasture lands. But pastures on steep
slopes are also potentially at risk to erosion, especially in
comparison to forest cover on the same slope.
Every watershed in the mid-Atlantic region has some
agriculture on steep slopes. The proportions are lower in
remote mountainous areas since there is less agriculture
there, and in some predominantly agricultural regions
such as the Delmarva peninsula because there are fewer
steep slopes. The combination of steep slopes and
agriculture occurs most often on the foothill margins of
the great valleys in the mid-Atlantic region, where agri-
culture is a dominant local land use.
Nitrogen and Phosphorus Export to Streams
Despite the many benefits, there is a potential negative
impact of fertilizers, animal wastes, and other nonpoint
source pollutants coming from agricultural fields and
pastures. The problem was first identified decades ago
The proportion of watershed area that has
crop land cover on slopes greater than three
percent was calculated by overlaying maps
of percent slope and land cover, and dividing
the area of crop land cover on steep slopes
by the total area of the watershed.
Quintile
2
3
4
5
Data Range
(Percent)
<1.5
1.5 - 4.2
4.2 - 6.4
6.4 - 9.7
>9.7
36 9 12 15 18 21 24 27 30
Indicator Value
Figure 3.22
Proportion of watershed with crop land
cover on slopes that are greater than
three percent in the mid-Atlantic region.
-------
as part of the awareness of lake eutrophication. Lake
eutrophication is a process by which excess nutrients in
lake water make it easier for undesirable plants to thrive,
which in turn consume other resources and adversely
affect lake water quality for other purposes. Similarly, the
Chesapeake Bay Program has recognized the impact of
nitrogen and phosphorus loadings on the productivity of
the Chesapeake Bay and other coastal bays. As a
result, the Chesapeake Bay Program has developed a
number of watershed models to assess sources of
nitrogen and phosphorus loadings to the Bay; these
models have resulted in watershed-wide plans to reduce
these two pollutants. The potential effects of the export
of nitrogen and phosphorus from farmlands to streams
have been intensively studied for several decades. It is
now possible to survey the scientific literature to deter-
mine how much nitrogen and phosphorus export can be
expected for different types of land uses in different
areas.
The scientific literature provides a simple predictive
model of nitrogen and phosphorus loadings to streams
which is based only on land cover (see Appendix). Of
course, this model does not reflect actual fertilizer appli-
The proportion of watershed area that is
agriculture land cover on slopes greater
than three percent was calculated by
overlaying maps of percent slope and land
cover, and dividing the area of crop plus
pasture land covers on steep slopes by the
total area of the watershed.
Qulntile
3 6 9 12 15 « 21 24 27 30
Figure 3.23
Proportion of watershed with agriculture land cover on slopes that are greater
than three percent In the mid-Atlantic region.
-------
cation rates which determine local export amounts.
However, over a large area such as the mid-Atlantic
region, this type of model is valuable as a screening tool
to rank watersheds based on potential impacts assuming
that average fertilizer rates are used throughout the
region. In a nutshell, if there are no agricultural lands in
a watershed, then fertilizer application is near zero. Such
a watershed has less risk of impacts than a watershed
for which 30% of the area is used for agriculture. One
major drawback of this simple model is that it ignores
fertilizer applications in urban areas, where areas such
as lawns, gardens, and golf courses can receive heavy
fertilizer doses several times a year.
Figure 3.24 shows watershed rankings for estimated
nitrogen (in nitrate form) exports from agricultural lands in
the mid-Atlantic region. These ranks also apply for
phosphorus (in phosphate form), but the actual amounts
differ for the two elements as shown in the charts. The
map shows that the watersheds on the DelMarVa penin-
sula and northern end of the Chesapeake Bay have the
highest potential for fertilizer export based on land cover
patterns. This pattern suggests why the Chesapeake
Bay is such a concern, and why land use practices along
the Susquehanna River are of concern to the Bay's
water quality stewards.
The annual amount of nitrogen and phos-
phorous exported to streams in each
watershed was estimated by multiplying the
amount of vulnerable land by loss coeffi-
cients determined from the literature.
Quintile
15 30 45 60 75 90 105 120 135 150
(ntScator \fetoe
Figure 3.24
Potential nitrogen (top) and phosphorous (bottom) loadings to streams in the
mid-Atlantic region. The map and legend reflect the nitrogen values.
-------
Soil Loss
A significant portion of the Federal budget is devoted to
the reduction of soil loss in the United States. Soil
erosion is important because eroded soil can be trans-
ported to a stream where it becomes sediment and
because topsoil erosion reduces productivity of agricul-
tural lands. Topsoil is expensive to replace and natural
soil-forming processes would require thousands of years
to replenish soil already lost from the nation's farmlands.
One of the tools developed by agricultural scientists to
estimate soil loss from farm lands is the Universal Soil
Loss Equation, or USLE. The USLE is intended to show
farmers how agricultural practices reduce, or contribute
to, soil erosion. The USLE is not generally applied to
non-agricultural land uses, nor is it used to estimate
wind erosion. Soil texture, slope, land cover,
precipitation, and cropping practices are the
primary data that go into the USLE (see the Ap-
pendix). The soil loss estimated from the model is
given in tons of soil lost per acre per year.
The annual amount of soil eroded from
agricultural pixels was estimated by using
the Universal Soil Loss Equation for
agricultural lands within each watershed.
The index value is the proportion of
watershed area where the estimated value
was greater than one ton per acre per
year.
Figure 3.25 shows the watershed rankings for the per-
centage of watershed area that could experience greater
than one ton per acre per year of soil loss. As shown in
the chart, the proportion of area in a watershed where
soil erosion exceeds the threshold value ranges from
less then 1 % to over 35%. The watersheds surrounding
the Chesapeake Bay and the Delaware River estuary
show the greatest potential for soil loss.
Quintile
2
3
4
5
Data Range
(Percent)
<11.2
11.2 - 19.8
19.8 - 24.0
24.0 - 35.3
>35.3
10 20 30 40 50 60 70 80 90 100
Indicator Value
Figure 3.25
Proportion of watershed with potential soil loss greater than one ton per acre per
year In (ho mid-Atlantic region.
-------
Forest Land Cover
At one time, nearly all parts of every watershed iathe
mid-Atlantic region were forested. Today, the remaining
forest helps indicate the probable condition of streams
within each watershed. The forest is not the only indica-
tor, however, because the specific types and patterns of
non-forest land cover are also important.
The proportion of forest cover in each watershed was
calculated from the land cover map. The results are
summarized in Figure 3.26 which shows the watershed
rankings for this indicator, and in the chart which indi-
cates the values obtained. Watersheds with high propor-
tions of forest are located in the northwest and southwest
portions of the mid-Atlantic region. Watersheds with
relatively low proportions of forest cover are
clustered around the upper Chesapeake Bay
and near major urban centers in the region.
The proportion of watershed area with forestr
was estimated from the land cover map.
The index is calculated as total forest area
divided by total watershed area.
Quintile
Overall, about 70% of the region has forest cover. The
chart illustrates that most of the watersheds in the region
are primarily forested, and a majority of watersheds have
at least 60% forest cover. Forests occupy less than 50%
of the area in the 15 to 20 watersheds that are the most
highly-developed and have the most agriculture cover.
Figure 3.26
Proportion of watershed that is forested in the mid-Atlantic region.
-------
Forests
Forests are important elements of natural and human-
dominated landscapes. Forests benefit both humans
and wildlife species, providing wood fiber, outdoor recre-
ation, wildlife habitat, and regulation of some hydrologic
functions. They dominate most of the mid-Atlantic
region today. Historic patterns of land use and develop-
ment have created the present distribution of forests from
what once was essentially ail forest. These patterns
have also caused changes in the plant and animal
species which live in forested environments.
Four related views of forest cover patterns are presented
in this section. Earlier, in the context of stream condi-
tions, the watersheds were ranked based on the propor-
tion of area with forest cover (Figure 3.26). In this
section, the pattern of the existing forest cover is de-
scribed as it affects various environmental values, par-
ticularly wildlife habitat
Forest Fragmentation
As in other regions of the United States, forest fragmen-
tation is an important issue in the mid-Atlantic region.
Although the phrase has several meanings, it is used
here to describe a formerly continuous forest that has
been broken up into smaller pieces. In the eastern
United States, forest loss is generally associated with
conversion to agriculture and urban cover types. These
human land uses remove some forest and leave the
remaining stands in smaller, isolated blocks. The pattern
of forest loss is as important as the amount lost. For
example, a checkerboard pattern exhibits more fragmen-
tation than a clumped pattern of the same amount of
forest. As described and illustrated in Chapter 1, the
degree of connectivity can affect the sustainability of
forest species within and among watersheds. Areas with
large blocks of continuous forests support a variety of
interior forest species, whereas areas with small, frag-
mented forests support fewer interior forest species and
more edge-dwelling species. However, high levels of
connectivity may also promote the spread of certain tree
diseases across the landscape. Our assessment of
forest connectivity is related to habitat for interior forest
species; therefore, high levels of forest connectivity are
considered the most desirable conditions, and low levels
of connectivity are considered the least desirable condi-
tions.
A variety of indicators have been used in the past to
assess fragmentation. For any one of the indicators, the
apparent degree of fragmentation is highly dependent
upon the definition of forest, the scale at which forests
are mapped, and the scale at which fragmentation is
measured. For example, if a given area is completely
covered by forests of any type, then it would not appear
to be fragmented by that definition of "forest." If, how-
ever, the same area was mapped at a finer scale which
recognized, say, age class differences within the forest,
then the "forest" of each age class would appear to be
fragmented. Similarly, apparent fragmentation increases
as smaller and smaller breaks in the forest canopy are
recognized. At some scale, each tree could be consid-
ered as a separate island. These differences simply
suggest that meaningful interpretations of fragmentation
measurements require knowledge of how, and at what
scales, the measurements were taken (see Appendix).
Figure 3.27 shows watershed rankings for the fragmenta-
tion indicator. Forest fragmentation is highest in water-
sheds around the Chesapeake Bay and in western
Pennsylvania. But as indicated in the chart, forested
spots in even the most fragmented watersheds are still
likely to be adjacent to another forested spot. In about
half of the watersheds in the mid-Atlantic region, there is
at least a 90% chance that any given forested pixel is
adjacent to another forested pixel.
A comparison with the earlier map of forest area percent-
age (Figure 3.26) shows, as expected, that fragmentation
is generally higher in watersheds with lesser proportions
of forest cover. Places that don't follow the trend are
places for which there are forest pattern differences. For
example, some watersheds in eastern Pennsylvania
have higher forest fragmentation than is expected based
on the amount of forest there. The reverse is true in
some watersheds in mountain regions.
-------
Forest Edge Habitat
Edge habitat occurs at the boundaries between different
types of land cover. Species that require edge habitat
use the resources in two or more vegetation types.
Some birds, for example, nest in forests and forage in
nearby fields. Forest edge habitat is fairly common
throughout most of the mid-Atlantic region because there
is at least some forest nearly everywhere, and few areas
are completely forested. Partial forest cover is what
creates forest edge habitat, no matter what the scale.
In our assessment, we have made the judgement that
more edge is desirable, as it promotes species diversity
across the landscape. However, forest-edge habitat can
also be viewed as undesirable. Certain nest parasitic
bird species (for example, cowbirds) have their greatest
impact on other native species in areas where edge
habitat is common. The analysis of edge habitat pro-
vided in this atlas should provide useful information on
edge habitats, whether edge is considered desirable or
undesirable.
An index of forest fragmentation was
calculated for each watershed as the
probability that a given forested pixel in the
watershed was not adjacent to another
forested pixel.
Quintile
2
3
4
5
Data Range
(Percent)
< 7.8
7.8 - 11.2
11.2- 13.8
13.8 - 21.4
>21.4
10 20 30 40 50 60 70 80 90 100
Indicator Value
Figure 3.27.
Forest fragmentation index in the mid-Atlantic region.
-------
Because fragmented forests have more edge habitat, the
fragmentation map shown earlier could be a guide to
forest edge. That map, however, is only one realization of
the fragmentation indicator at a particular scale. It is
unlikely that many species perceive forest edges exactly
like that. Some species may require watershed-size
areas made up of mostly edge habitat, but others need
just a bit of edge within a forest. Without a particular
species in mind, there is no single answer to the question-
of how much forest edge habitat there is. Multiple-scale
approaches are necessary to assess habitat for many
species. Looking at different scales helps us to under-
stand if, and how, habitat measurements at one scale
might be extrapolated to other scales.
Maps of forest edge habitat at three scales (Figure 3.28)
were prepared by using calculation windows of about 7,
65, and 600 hectares. The maps illustrate how the
apparent habitat picture changes with window size.
Species that require more extensive areas of edge find
less suitable habitat in the region, and such habitat is
concentrated in the more heavily-fragmented water-
sheds. The areas of suitable habitat derived from a 600
hectare calculation window appear to be predictable from
the those generated from smaller-sized windows. How-
ever, the reverse is not true; that is, it would be very
difficult to predict the spatial distribution of habitat gener-
ated from the smallest window size based on the habitat
map produced from the largest window size. This means
that species with finer-scale landscape requirements
would not necessarily benefit from protecting the habitat
of species with broader-scale landscape requirements.
The maps of watershed ranks for each window size
(Figure 3.29) capture the regional patterns. It is clear that
the complex spatial pattern in the urban areas to the
north and west of the Chesapeake Bay provide extensive
edge habitat. In some cases, the rank of a watershed
changes for different window sizes, which indicates
scale-dependent pattern differences between water-
sheds. The charts demonstrate that there are fewer
watersheds with suitable forest edge habitat for species
with large area requirements.
The amount of forest edge habitat in each
watershed was estimated by using a spatial filter
to map forest edges with three different window
sizes. The proportion of total watershed area
above a threshold forest edge value was used as
the index.
Figure 3.28.
Surface maps of forest edge habitat, shown in olive, at (a) 7 hectare,
(b) 65 hectare, and (c) 600 hectare scale in the mid-Atlantic region
(see text for explanation).
10 20 30 40 SO 60 70 80 SO 100
Indicator Value
Figure 3.29.
Proportion of watershed with suitable forest edge habitat at (a) 7, (b) 65, and (c)
600 hectare scale in the mid-Atlantic region.
-------
<0.01
0.01 - 0.20
0.21 -1.99
4m 2.00-6.92
5*> 6.92
1 10
Indicator Value
10 20 30 40 50 60 70 80 90 100
Indicator Value
-------
Interior Forest Habitat
In contrast to edge species a variety of wildlife species,
require nearly the opposite type of habitat — large tracts
of continuous forest cover. Interior forest habitats are
relatively rare and easily lost, so forest interior species
sometimes become the focal point for debates over
human activities such as road-building. In the mid-
Atlantic region, interior forest is most likely to be found
where the percentage of forest is high and fragmentation
is low. But like edge habitat, interior forest exists at many
scales. Salamanders need different amounts of forest
habitat than bears, and differences such as these call for
a multiple-scale analysis.
Species with smaller area requirements should enjoy a
region-wide distribution based on the habitat map for the
smallest window size, whereas species with larger area
requirements should have more restricted and patchy
distributions (Figure 3.30). Suitable large-window habitat
is concentrated and relatively well-connected in moun-
tainous areas, with little or none appearing in the most
urbanized areas. Although there are some differences in
watershed rankings of interior forest habitat proportions
for the three window sizes (Figure 3.31), overall patterns
are similar.
With such a high threshold value for "suitability" (90%
forest in a window), the proportion of watershed area that
is suitable decreases rapidly with increasing area require-
ments. The charts illustrate that in most of the water-
sheds, at least half of the area is considered suitable by
the small-window analysis. The number of watersheds
with half of the area in large-window interior blocks is
much smaller.
The amount of interior forest habitat in each
watershed was estimated by using a spatial
filter to map forest density with three different
window sizes The proportion of total watershed
area above a threshold forest density value
was used as an index.
Figure 3.30
Surface map of interior forest habitat, shown in green, at (a) 7 hectare,
(b) 65 hectare, and (c) 600 hectare scale in the mid-Atlantic region
(see text for explanation).
1B<23.8
2H 23.8 - 39.5
3H 39.5-49.7
4| 49.7-65.7
5B>65.7
10 20 30 40 50 60 70 80 90 100
Indicator Value
Figure 3.31
Proportion of watershed with suitable interior forest habitat at (a) 7, (b) 65, and
(c) 600 hectare scale in the mid-Atlantic region.
-------
• <*£' :.r ' s,
.ri^:!:^>y!'-'->' *• V
.4' V.... .•" "/rV
2E 10.7-25.9
3H 25.9 - 36.4
4B 36.4- 53.7
5B>53.7
10 20 30 40 50 60 70 80 90 100
Indicator Value
10 20 30 40 50 60 70 80 90 100
Indicator Value
-------
Another possible indicator combines information from all
three scales of analysis. A watershed with adequate
interior habitat at all three window sizes might have a
greater diversity of interior forest species. Watersheds
that have a higher proportion of area which supports
more scales of habitat are identified in Figure 3.32.
These are generally the watersheds identified in the
single-scale analysis using the largest window size.
The Largest Forest Patch in Relation to the
Amount of Forest Land Cover
About 30 years ago, A.W. Kuchler made maps of poten-
tial natural vegetation, or vegetation that would occur if
influenced by only natural processes such as weather
This index was derived by overlaying the
three maps of interior forest habitat shown
earlier. The index value is the proportion of
area in a watershed that was above the
threshold value for all three scales.
and fire. In the mid-Atlantic region, Kuchler's maps
show that the potential natural vegetation is almost
exclusively forest, and areas with other cover types
represent departures from natural conditions.
Previous discussion introduced the concept of forest
fragmentation (Figure 3.27). Consider a watershed with a
certain amount of forest cover. If the forest is in one
continuous patch, then the area of the largest forest
patch equals the total forest area. If the largest patch is
smaller than this expected value, then fragmentation has
occurred and the remaining forest cover is discontinuous.
Qulntile
1
2
3
4
5
Data Range
(Percent)
Flgura 3.32
Proportion of watershed with suitable interior forest habitat at three scales in
the mid-Atlantic region.
-------
-TV
*^sr' •
- ~ — •• +* *-,^*~ ^^ „ vy J &** £J
Figure 3.33 shows the results of plotting the proportion of
each watershed in the largest forest patch versus the
proportion of urban and agriculture (non-forest, or
anthropogenic land cover) in each watershed. The
double-hatched line in Figure 3.33 is the expected value
without forest fragmentation. The figure indicates that
the size of the largest forest patch becomes less than
expected (that is, fragmentation becomes more important
at a watershed scale) when human-altered land cover
occupies about 25% of the watershed. A curve has
been drawn through the points in an effort to more clearly
show the observed relationship. Figure 3.34 ranks all
1 j 1 1 1 r 1 f , r—1 -r f..
0.05 0.10 0.15 0.20 0.25 0.30 035 OM OA5 0.50 0.65 0.60 0.65 070 0.75 OflO OB5 030 095 1.00
Proportion of Watershed In Anthropogenic Land Cover
Figure 3.33
Proportion of the watershed In the largest forest patch In relation to the proportion
in anthropogenic (non-forest) land cover (see text for explanation).
watersheds based on their departure from the expected
values given in Figure 3.33.
Landscape Change (1975-1990)
A common perception is that patterns of forested, agricul-
tural and urban areas remain constant over time. In fact,
land cover changes occur all the time. In this section we
present patterns of vegetation change measured by
comparing satellite images from 1975 and 1990. The
change is determined by using a vegetation measure
called the Normalized Difference Vegetation Index or
NDVI (see Appendix) which was calculated for each pixel
on each of the two dates. NDVI is a measure of the
relative greenness of an area. NDVI values range
between 0 and 1; high values usually indicate presence
of forest, whereas low values generally indicate bare
ground, pavement, or a water body. Positive changes in
NDVI indicate a greening up of an area (for example,
reestablishment of forests, maturing of lawns), whereas
negative changes indicate losses in greenness (for
example, clearing of a forest for development, forest
dieback caused by an insect infestation). However,
observed changes are usually more difficult to interpret
(see discussion below). When the NDVI values are
essentially the same at both dates, then there has been
no change. When the value is greater in 1975 than
1990, we interpret this as vegetation loss during that 15-
year period. When the value in 1975 is less than 1990,
we interpret this as vegetation gain. Total vegetation
change is taken to be the sum of loss and gain on an
area basis.
Comparison of temporal changes in reflectance mea-
sures from satellites, such as NDVI, can be useful for
gaining insight into land cover changes when land cover
maps from two different dates are not available. Inter-
preting the measurements relative to land cover change
is not simple, because some changes in reflectance are
not changes in land cover. Crop rotation is a good
example. Change in NDVI measurements may be the
result of seeing a field in production on one date and
fallow on the other. Interpretation of these measure-
ments for actual land cover change requires a lot of
additional work beyond calculating their difference over
time. Because of the additional work needed to interpret
actual land cover change and because NDVI data were
not available over the entire region, these indicators were
not used in the synthesis which appears in the next
chapter.
-------
Despite the complications, the amount and spatial pat-
tern of NDVI change is important. For example, many of
the decreases in NDVI turn out to be associated with
road improvements, new residential developments,
urbanization projects, and construction of reservoirs. A
good example is the vegetation loss associated with the
construction of Interstate 295 east of Richmond which is
illustrated in Figure 3.35. In the central Pennsylvania
Mountains, some large blocks of vegetation gain suggest
recovery from a gypsy moth infestation. Other gains in
NDVI appeared to be the result of maturing vegetation in
residential developments. Gains in NDVI appear to be
associated with both natural and anthropogenic pro-
cesses, whereas non-crop rotation NDVI losses appear
to be more consistently associated with anthropogenic
activities.
These examples show that, after calibration, NDVI
changes over time can help answer several ecologically-
important questions, such as how much change has
occurred, whether or not change is evenly distributed
over all the watersheds in the region, and whether or not
vegetation change concentrated in the headwater re-
This index was calculated as one minus
the ratio of the area of the largest forest
patch over total forested area in a
watershed.
Qulnlile
1|
2I
3 I
< 1.5
1.5- 5.0
5.0- 11.3
4 H 11-3-23-°
5 | 23.0 - 49.2
BJI Not included in analysis
0.05 0.1 0.15 0.2 025 0.3 0.35 0.4 0.45 0.5
Indicator Value
Figure 3.34,
Departure of the largest forest patch from the maximum possible for a given
amount of anthropogenic cover in the mid-Atlantic region.
-------
gions of streams. Regional-scale differences among
watersheds can be large. For example, Figure 3.36
shows vegetation change for three watersheds in the
region. In one of the watersheds, about 3% of the
surface area shows change, and in another two the value
is about 35%. The next section describes some regional
patterns in NDVI change, recognizing that more work is
needed for confident interpretations of land cover
change.
t^St^HS^^tS^Mi^^':^
tM**fw-^K.' *£'. 'W/ ^ - r'ur'^S^
A:X.----r^i*rf*»x'i!r%rii/S.:.wjV'3F>..p.»*<;"i:
Figure 3.35.
Vegetation change east of Richmond, VA. The NDVI loss east of Richmond
(box at lower left) corresponds to the construction of Interstate 295 between
1975 and 1990. The box at lower right illustrates differences in NDVI that are
associated with periodic exposure of aquatic vegetation in a tidal marsh. Source:
North American Landscape Characteristics Program, Landsat Multispectral
Scanner image.
-------
Figure 3.36,
NDVI changes In three watersheds In the mid-Atlantic region. Yellow and red
Indicate decreases In NDVI, green Indicates increases, and gray indicates no
dlscemable change, (a) A north-central Pennsylvania watershed, (b) A south-
east Pennsylvania watershed, (c) A southeast Virginia watershed. Source:
North American Landscape Characteristics Program, Landsat Multispectral
Scanner images.
-------
-------
Vegetation Change Among Watersheds
Figures 3.37 through 3.39 show the rank ordering of
watersheds in terms of vegetation loss, gain, and total
change. Vegetation gain and loss have a regional
pattern with the highest rates of change along the east-
ern seaboard and decreasing westward. But there are
some exceptions. On the vegetation loss map (Figure
3.37), there are moderate to high rates of loss in the
Appalachian Mountains, and low rates of loss along the
western edge of the Coastal Plain in Virginia. On the
vegetation gain map (Figure 3.38), there are some high
amounts of gain scattered throughout the western por-
tion of the region and some low amounts of gain on the
coastal plain. The color pattern on the vegetation gain
map is similar to the color pattern on the population map
shown earlier in this chapter (Figure 3.6).
There appears to be some correlation between vegetation
gain and loss. In other words, gain and loss appear to be
high together, or low together, on average. This correlation
pattern is evident in the total change map (Figure 3.39).
Areas of loss and gain may be correlated because the
areas of highest loss (initial clearing of forests) and gain
(maturing lawns) tend to be near each other within expand-
ing suburban areas surrounding cities. Figure 3.39 shows
the clearest pattern of high rates of change along the coast
and decreasing westward.
Quintfe
1
2
3
4
F!gut93.37
Decreases In the NDVI from 1975 to 1990 in the mid-Atlantic region (see text for
explanation).
5.0- 8.5
8.5-12.7
12.7-15.5
Data Range
Quintile (Percent)
5H 15.5-26.1
H not included
D no data
Figure 3.38
Increases in the NDVI from 1975 to 1990 in the mid-Atlantic region (see text for
explanation).
-------
Vegetation Change Within Watersheds
We have seen that vegetation change is not uniformly
distributed over the region. Some watersheds show high
rates of vegetation change, while others are low. What
about changes within individual watersheds? Is vegeta-
tion change uniformly distributed within a watershed?
To answer this question, we divided each watershed into
two sections: first-order stream regions, and all higher-
order stream regions. First-order streams are small
streams at the top of the watershed. A second-order
stream is formed at the confluence of two first-order
streams, a third-order stream is formed at the confluence
of two second-order streams, and so on. The first-order
stream region in a watershed is the area that drains into
all first-order streams in that watershed. This area is
usually the steepest portion of the watershed and there-
fore can be-more impacted by the loss of vegetation. We
compared the observed change in the first-order region
with the expected value if the change were evenly
distributed (without regard to stream order) throughout
the watershed.
Data Range
Quintile (Percent)
1 • 2.2-12.5
2B 12.5-17.0
3 3 17.0-23.0
4 • 23.0-30.4
Data Range
Quintile (Percent)
SB 30.4-45.9
H not included
D no data
Figure 3.39
Total change in the NDVI from 1975 to 1990 in the mid-Atlantic region (see text
for explanation).
-------
Figures 3.40 through 3.42 show the difference between
observed and expected values for vegetation loss, gain,
and total change in first-order stream regions. When
vegetation loss is high across the whole watershed
(orange or red in Figure 3.37), the loss in the first-order
region tends to be higher than expected (red in Figure
3.40). The opposite seems to be the case for vegetation
gain. When vegetation gain is high across the whole
watershed (red in Figure 3.38), it tends not to occur in
the first-order region (green in Figure 3.41). This pattern
suggests that some portion of vegetation gain is associ-
ated with human activities. People tend to avoid devel-
oping land in the first-order region where slopes are
steep. Thus, we would expect that vegetation gain
associated with human activity would be concentrated in
the higher-order stream regions of the watershed.
Comparison of the expected versus observed map for
total vegetation change in first-order streams (Figure
3.41) with that across the whole watershed (Figure 3.39)
does not show a clear pattern, which might be expected
because of the opposite trends seen earlier for vegeta-
tion loss and gain.
• Observed < Expected H Not included
H Observed = Expected D No data
|| Observed > Expected
Figure 3.40
Differences in observed and expected decreases
in the NDVI from 1975 to 1990 in first-order
stream regions in the mid-Atlantic region (see
text for explanation).
-------
^
• Observed < Expected @ Not included
H Observed = Expected Q No data
• Observed > Expected
I Observed < Expected
I Observed = Expected
I Observed > Expected
Not included
No data
Figure 3.41
Differences in observed and expected increases
in the NDVI from 1975 to 1990 in first-order
stream regions in the mid-Atlantic region (see
text for explanation).
Figure 3.42
Differences in observed and expected total
change in the NDVI from 1975 to 1990 in first-
order stream regions in the mid-Atlantic region
(see text for explanation).
-------
Vegetation Loss on Steep Slopes
When vegetation is removed, the soil surface is exposed
to erosion. The steepgr the slope, the greater the poten-
tial erosion. Figures" S!*43 and 3.44 show the pattern of
vegetation loss'bn slopes greater than 3%. Not surpris-
ingly, there is little problem on the Coastal Plain, where
land is generally flat, nor is there much in the Piedmont
of south-central Virginia. Important areas of vegetation
loss on steep slopes include eastern Pennsylvania,
extending south to the Chesapeake Bay and west along
the Maryland Panhandle into western Pennsylvania,
central Virginia in the Ridge and Valley Physiographic
Province, and southwestern Virginia in
the southern Appalachian Mountains.
Figure 3.43
Surface map of decreases in the NDVI from 1975 to 1990 on slopes greater than
three percent in the mid-Atlantic region (see text for explanation).
1 • 0.02- 1.5
2 • 1.5 - 2.5
3 B 2.5 - 3.3
4 H 3.3 - 5.7
5 • 5.7 -16.0
H not included
D no data
Figure 3.44
Proportion of watershed with decreases in the NDVI from 1975 to 1990 on
slopes greater than three percent in the mid-Atlantic region (see text for
explanation).
-------
f > s£ «*< < < "a*"" js """%. ^
^- a- -*-' - fr - f-
Comparative Assessment of Mid-
Atlantic Watershed Conditions
This chapter summarizes the indicators presented in
Chapter 3, in order to identify changing environmental
conditions across the region. A simple way to do this can
be adapted from the magazine, Consumer Reports.
When rating products, Consumer Reports provides
relative scores on performance or features to help the
consumer decide which brand to purchase. As applied
here, each indicator from Chapter 3 could be considered
as a guide to relative watershed "performance" for an
indicator, and that could help in determining which water-
shed to "buy."
The watershed ranks for each indicator are presented in a
Consumer Reports format in Table 4.1 (see discussion on
Data Interpretation, inset on page 79). It is possible to get
an idea of the relative condition of a watershed by reading
across a row and counting the number of boxes of a
particular color. Watersheds dominated by green and
khaki colors are in better relative condition (more desir-
able) than those dominated by red and orange (less
desirable). Table 4.1 can be used by readers to identify
conditions of watersheds relative to a particular question
or interest. For example, the reader may want to know
how their individual watershed rates relative to other
watersheds in the region with regard to a set of forest
habitat indicators. The table provides a way for readers to
explore different combinations of indicators for the water-
shed they live in, and to compare their area to neighboring
watersheds. Because watershed ranks are shown, each
column contains an equal number of the different colors
(recall that there is an equal number of watersheds in
each of the 5 groups).
Of course, an atlas is about maps. Maps show the spatial
distributions that cannot be seen in tables. A simple
summary of the data in Table 4.1 is shown on the maps in
Figure 4.1. The map at the upper left shows the number
of indicators for which a given watershed was ranked in
the top 20% (most desirable) of all watersheds. The map
at the bottom right shows the number of indicators for
which a given watershed was ranked in the lowest 20%
(least desirable) of all watersheds.
Comparisons of the maps suggests some general conclu-
sions. Relative to all other watersheds in the mid-Atlantic
region, the watersheds in southeastern Pennsylvania and
the northern end of the Chesapeake Bay have consistently
lower values for all of the landscape indicators. Conversely,
there are a few watersheds in southwestern and north-
central portions of the region that have consistently higher
scores across all the indicators.
Although the maps are valuable for their simplicity and
ability to show watersheds that have consistently higher or
lower scores, a more sophisticated technique is needed to
group all of the watersheds into categories of environmen-
tal quality. Cluster analysis is a statistical technique that is
often used to find groups based on the similarity of data
values. One familiar example of the clustering technique
is the assignment of new hospital patients to groups of
similar risk, based on factors such as age, medical history,
and other factors.
-------
A cluster analysis was done using nine of the 32 indica-
tors from Chapter 3 (see inset below). These nine indica-
tors were selected because they represent a broad range
of environmental condition measurements and were not
highly correlated with each other overall. The procedure
identified nine groups of watersheds (Figure 4.2). The
mean indicator scores for each cluster (Table 4.2) can be
inspected to see how the groups are different.
. Ca"nonlcaTaTscfiminanr :
j combinations of yanabfe_s that explain the variability in *
»7feiermmeT!rtere ts gooo* separaflSfTDetweencfusfers "J
i_pfcis of canontca! scores fffhe clusters identified were not ^
"i canonical space flienTl Ts Tikefy that there is no real difference
ivSefe"used In the cluster analysis because it is better to use
»rsTnal aronot sirongjy correfa&d the nine Indicators used were 1) 1990
wo density 2J pogufalori cfiange^'roa^'deri'^r^rpropbffion^ 1
Jwjglri wffl^roods wrffiln 30 meters^S)proportion[_oFwaFersheff wftfj~" ~"~ ^J
fufeon Bpes greaterlhan friree'percen!. 6) pFopbTGorTorvralersfied j
* gBl vwBTa^jacenl forest cover,_7) proportion of watershed supporting _ _' J
"" ""^""lalatIfiree scales S^average afmospReric suTfate'wet'"" ~ '""" "~"|
"numterof"wafertnipounoWienfe~perjfiSO^tr^OTTfflomi!etB."'"~ "I
rtnei tut^FA "(!&«•«« WT^r* ? TpjnE™?1.. ajfi £., ,* i,*if\ nacac Pnni rlWfinn anrl *
of Iha_land5capeffidicators u^ed>{-^
*" ^6*n P^ulalonpres^ulBs'afe'highrSgrTculfure on sfeep |
nd streamSngfli w (h adjacent fbresf were mode?afely iriverseiy ~"" |
-were kgX becauw (heymeasuTe different aspects of the'" ~ *~\
® Brnw Is an indcajorof erosion p_otBntjal_and_the latter js an^^j
...quaHE/ ariSTTabTlal" Because of the"stron^corTela!!ori"li:>eSJeerr"""|
akm and" road density, only roads were mcludedlri assigning relative
c! scores fs%5 Table 4 25 ~~
Cluster means were used to rank the groups according
to relative cumulative impact. The ranking was done by
reading down the columns and marking (red in Table 4.2)
the three cluster means with the most extreme values.
The three highest values were marked for population
density, population change, road density, roads by
streams, agriculture on steep slopes, atmospheric sulfate
deposition and impoundment density; higher values for
these indicators suggest potential negative impact. The
three lowest values were marked for forests by streams
and proportion of the watershed supporting interior forest
habitat at three scales; lower values for these indicators
suggest potential negative impact.
After identifying the three values in each column, the
number of red values for each cluster was counted in
each row. This count is interpreted as a measure of
relative cumulative environmental impact for that group of
watersheds. The score was then used to rank the water-
shed groups. Higher scores suggest greater adverse
impact than lower scores. These numbers were used to
interpret watershed conditions as specified on page 84.
IJJflTfillf
, and
t tf the/irst and second canonical scores showea mat clusters 1,^,0,0, ana a .
|groups, and that clusters j> and 6 werei_we[l^^ separated from.duster^ |
msunS separation of clusters 5 and 6 and clusters 4_and_7 were evictenj^ J
fjtWTsTind'Srd and 1st and 4th canonicaTscores,_respectively These __ J
(ooe&Tviat the clusters are grda1)Iy_ differentjrom each other Examma- si
«SC"innr5^ca!or~vaTuesBy watersrie3"grou"ps (cfusfersfsuppofls Ms
s regional boundary "were" noTincfuded in" iHe
r area Is outside the mid-Atlantic Region
fU€lI emciiy&l*> l/oC9US@ TOOSt Of utcu a\*sa ia uuiaiuc u ic n nvi-nncii iuu i x&yiui i »
i "edae" watersheds often had unrealistic, extreme values for one or more _ _ J
*Trs,"lnd e^irema values have undesireabie consequences on cluster — ^
~ ' He^encompasslng Phlladelphkfcpde 2040202 in figure 36) .
causeofls exUeme'yaijes.jTJgs't ofthgwfte'rsried, however, 1
n duster 6
-------
"7!4
*?&
Hi i
Data Interpretation
he Stoljjrig'inforrnatidn*. should help explain howTo interpret Table 4 1 Boxes
gded in gray indicate that no data was available for that indicator This problem
*" cWnfineo! tojhe vegetation change indicators for about 40 watersheds Also,
nostpyijelndicatorsTn Chapter 3, a relatively high or low value has a clear
jningTfelafive to landscape condition For example, high values of soil loss
gest greater erosiorj problems But the interpretation is less clear for other
jajprs, such as; forestjedge, the vegetation increase indicators, and impound-
it density A high proportion of forest edge suggests more clearing of forests,
hich many people would consider to be a negative impact Increasing edge
fjtbftat Jjowever, can increase the abundance of some wildlife game species,
"^ jjgrjjpany people would see as a positive impact
dTtion, many of'the indicators shown in Chapter 3 are correlated with each
, jfjndicaiqcs are positively correlated, then they tend to increase or decrease
fie? If tfiey'are negatively correlated, then one tends to decrease as the other
Ceases In both cases the^value of one indicator can tie predicted from the
ue of the othe|, and so there js no "new" information to be gained from the
x>nd one; For example, wfie're forest density is high, the amount of forest edge
|jne,cessarily fow, and where atmospheric nitrate wet deposition is high,
Imospberic sulfate wet deposifion also tends to be high The reason for
SfesSntmg correlated indicators in Chapters 3 and 4 is to provide information in
different terms ifonly a small set of uncorrelated measures were presented, then
puch^of the reaj-world meaning of landscape patterns would be lost — it would be
""1 to the* reader to know which indicators that did not appear were correlated with
j~pnesjhat,wgre presented For example, farmers may be more interested in
nifrate.deposition, instead of its correlate sulfate deposition, because nitrogen is a
.component of fertilizer Likewise, deer hunters may bejnore interested in the
Ooca§ofi,pf. edge habitat, not in an inversely correlated measure of high forest
jjjgnsify By studying maps of all the indicators for all watersheds, the reader is
i able to compare and see which are correlated, and choose which ones to consider
" jjased on hisjsr her own perspectives Finally, the fact that some of the Indicators
are correlated should be viewed as an opportunity and not a problem, by restoring
^alues for a given indicator it is likely that other indicators will also change
-------
3010102
3010104
3010105
3010106
3010201
3010204
2050202
2050203
2050205
5050005
5050006
5050007
5050009
5070101
5070102
5070201
5070202
5090102
2050201
2050204
2050206
2050304
2070001
2070002
2070003
2070005
2070006
2080201
2080202
2080203
3010103
5010001
5010003
5010005
5020001
5020004
5030201
5030203
5050002
5050003
5050004
6010101
6010205
2070011
2080102
2080104
2080105
2080106
2080107
2080109
2080206
2080207
3010202
3010205
2040207
2060002
2060004
POPDENS
POPCHG
RDDENS
mam
m m m
m m n
D • •
D • a
o a a
fl B fl
• a •
• • •
• a •
O Q D
• • a
• a a
B B a
UINDEX
a
a
a
•
n
•
•
•
•
•
D
•
g
•
•
•
•
D
a
•
Q
a
•
•
N03DEP
B
B
fl
U
a
a
a
•
•
•
a
S04DEP
•
•
•
U
a
a
a
•
•
a
a
OZAVG
•
•
a
B
LJ
a
LJ
fl
•
a
•
a
•
CO
Q
a
•
u
a
•
y
a
a
•
B
B
•
a
CROPSL
a
a
B
a
•
•
•
•
•
•
a
•
LJ
u
a
•
u
a
•
•j
1
•
a
B
a
B
•
•
n
a
Q
n
u
•
a
•
y
a
•
fA
<£ Q
oi fe
gR
K
a •
• •
a B
a B
a •
a •
a •
m •
• •
• m
• Q
H EH
0 O O rj H g
0 , oH^t-ssissgsaobii
§^ri^El§la88eillil|g|SS
HGKc/3^^'i^'^OQQQOMMMQ^r^r^'o 'S
So?coe2Sg55E,WHHK2zggS22S^
• aaaaflfliiaflflflaafliiflnflfls 2
maammmmmaammmamammmmi i
maaammmammmmmmmmmmmmj 2
• OgUHflflflflQElflflflflCiB H a • 4 4
• ••(^•••••••••[jaH a m m 2 2
nBBBBBBBBBBBBB[]ElElB[]B4 0
BaaaaaaaHaaanaBBoaB6i2 4
BHHIIHQOIillBBBaBflBflaBC]4 4
BaaaBDyaBnaaaneaiiBBiie 3
BBBHBnilJIB[JtlHaflflBHBBB6 5
aBBaaaaaDDBBBaaaaaaas 4
Table 4.1
Summary of watershed rankings by indicator. The color scheme of green, khaki, yellow, orange, and red
represents the first- through fifth-ranked groups of watersheds, respectively. This table includes all watersheds
used in the cluster analysis.
-------
ss o g
W M
2060005
2060007
2060008
2060009
2060010
2080110
2040101
2040103
2040104
2040105
2040106
2040203
2050101
2050107
2080204
2080205
5020003
2050104
2050106
2050301
2050302
2050303
2050305
2070004
2070007
2080103
3010101
5010006
5010007
5010008
5020002
5020006
5030106
5030202
5050001
5050008
6010102
6010206
2050306
2060003
2060006
2070008
2070009
4120101
5010004
5030102
5030105
2040202
2040201
2040205
2070010
2080108
2080208
5010009
5020005
5030101
5030104
a a
H a a
a a
a a H
a a
a a a
na
a a
a a
a a
aam
mam
aaa
a a a a
• a B a
BBB
Table 4.1 continued
-------
Lowest
Quintile
Highest Quintile
• No measurements
in top 20 percent
• 1 to 6 measurements
in top 20 percent
II 7 to 12 measurements
in top 20 percent
• 13 to 18 measurements
in top 20 percent
• Greater than 18 measurements
in top 20 percent
H Not included
Lowest Quintile
| No measurements
in bottom 20 percent
H 1 to 6 measurements
in bottom 20 percent
|f 7 to 12 measurements
in bottom 20 percent
| 13 to 18 measurements
in bottom 20 percent
| Greater than 18 measurements
in bottom 20 percent
O Not included
Figure 4.1
Ranking of watersheds by their
occurrence in the highest- and
lowest-ranked groups
(see text for explanation).
-------
Figure 4.2:
Results of the cluster analysis and ranking of watersheds based
on indicator values. The ranking is based on the column of relative
cumulative impact scores that is shown in Table 4.2.
I Rank 1 (Cluster 4)
I Rank 2 (Cluster 9)
I Rank 2 (Cluster 8)
I Rank 3 (Cluster 7)
I Rank 3 (Cluster 2)
I Rank 4 (Cluster 1)
[ Rank 5 (Cluster 3)
| Rank 6 (Cluster 5)
I Rank 7 (Cluster 6)
I Not included
Table 4.2
Cluster mean indicator values and relative cumulative impact score (see text for explanation).
Mean values by Indicator1
Cluster
1
2
3
4
5
6
7
8
9
A*
175
111
88
28
325
982
78
38
28
B
21.5
31.6
7.5
-3.5
10.9
-2.6
66.7
10.1
-4.0
C
2.46
1.68
1.89
1.66
2.44
4.06
1.54
1.32
1.13
D
6.34
2.54
6.94
1.69
4.78
7.11
1.88
5.48
9.93
E
8.92
0.33
15.24
6.03
20.29
7.19
1.12
9.66
3.28
F
84.48
82.55
72.63
93.60
70.82
73.64
88.65
83.85
89.97
G
20.18
0.59
16.33
5.79
1.29
0.78
6.6
37.46
68.97
H
2272
2435
2468
1877
2825
2607
2056
2377
2290
I
41.96
0.18
8.67
11.99
2.85
15.67
19.27
4.43
4.26
RCI
3
2
4
0
5
6
2
1
1
'Indicator Codes:
A.1990 population density; B. Population change (1970 - 1990); C. Road density; D. Proportion of watershed streamlength that had roads within 30 meters; E. Proportion of
watershed with cropland and pasture on slopes > 3 percent; F. Proportion of watershed streamlength with adjacent forest; G. Proportion of watershed supporting interior
forest habitat at three scales; H. Average annual atmospheric sulfate wet deposition (1987 and 1993); I. Number of impoundments per 1,000 kilometers of stream length
RCI.Relative cumulative impact (the number of red values in a given row)
*Not included in RCI scoring - see discussion of cluster analysis.
Wote; The correlations with other indicators in Chapter 3 that are greater than + 0.55 are as follows.
A. Forest edge habitat in 7 hectare, 65 hectare, and 600 hectare scales, and the forest fragmentation index; B. None; C. None; D. None; E. Crop land cover on slopes > 3
percent, vegetation loss on slopes > 3 percent; F. Percent forest land cover, crop land cover on slopes > 3 percent (inversely), forest edge in 7 and 65 hectare windows
(inversely), and the forest fragmentation index; G. U-index, percent forest land cover, soil loss index, total vegetation change (inversely), total vegetation change (inversely),
nitrogen and phosphorous export from watershed, vegetation loss and total vegetation change in first-order stream regions (inversely), interior forest habitat at 7, 65, and 600
hectare scale; forest edge habitat at 7 hectare scale (inversely), forest fragmentation index (inversely); H. Average annual nitrate wet deposition; I. None
-------
Cluster 4 (Rank 1)
Watersheds in this group are found along the south-
central portion of the region, along the border with North
Carolina. None of the indicator means for this group
were colored red. The relative cumulative impact score is
0, the highest condition ranking among the nine clusters.
Population pressures, road density, and atmospheric
sulfate deposition are low. Values for agriculture on steep
slopes and impoundment density are moderate. The
score for riparian vegetation is the highest among the
nine groups. The biggest adverse impact is the low value
for the proportion of the watershed supporting interior
forest habitat at three scales.
Cluster 9 (Rank 2)
Watersheds in this group are located in the southwestern
portion of the region and north-central Pennsylvania.
This group has the highest score for roads adjacent to
streams, and thus a relative cumulative impact score of 1.
Although road density is generally lower in these water-
sheds, the roads that do occur are often adjacent to
streams because the watersheds are on the Appalachian
Plateau. In this area, most land is on steep slopes, and
so not only is stream density higher but also the roads
that do appear tend to follow valleys. The watersheds in
this group have the highest amounts of forest and ripar-
ian forest cover, and impacts from population, roads,
agriculture, and impoundments are relatively low.
Cluster 8 (Rank 2)
Watersheds in this group are located mostly in the
Ridge-and-Valley region and in northern Pennsylvania.
The principal adverse impact is a relatively high amount
of agriculture on steep slopes, which gave this group a
relative cumulative impact score of 1.
Cluster 7 (Rank 3)
Watersheds in this group are located in southeastern
Virginia on the Coastal Plain. The principal adverse
impacts are high scores for impoundment density and
population change, which gave this group a relative
cumulative impact score of 2. Also, forests in these
watersheds tend to be more fragmented than in other
watersheds in the region.
Cluster 2 (Rank 3)
Watersheds in this group are largely restricted to the
Delmarva Peninsula. The principal adverse impacts are a
high score for population change and a low score for
proportion of the watersheds supporting forests at three
scales, which gave this group a relative cumulative
impact score of 2.
Cluster 1 (Rank 4)
Watersheds in this group are located mainly in eastern
Pennsylvania, with others scattered throughout the
region. The principal adverse impacts for watersheds in
this group are high scores for population density and
change, road density, and number of impoundments per
1,000 stream kilometers, resulting in a relative cumulative
impact score of 3. Interestingly, the score for the propor-
tion of the watershed supporting interior forest habitat at
three scales is in the upper third of all groups.
Cluster 3 (Rank 5)
Watersheds in this group are scattered throughout the
Ridge-and-Valley and Appalachian Plateau Physi-
ographic Provinces. The principal adverse impacts for
watersheds in this group are high scores for roads near
streams, agriculture on steep slopes, sulfate deposition,
and a low score for riparian forest cover, resulting in a
relative cumulative impact score of 4.
Cluster 5 (Rank 6)
Watersheds in this group are found along the northwest-
ern margin of the Chesapeake Bay and in northwestern
Pennsylvania. The adverse impacts for these watersheds
are high scores for population density, road density,
agriculture on steep slopes, and sulfate deposition, and
low scores for riparian vegetation, and proportion of the
watersheds supporting interior forest habitat at three
scales. Watersheds in this group have a relative cumula-
tive impact score of 5.
Cluster 6 (Rank 7)
Watersheds in this group are in the most urbanized areas
of the region, including the Pittsburgh, Philadelphia,
Washington, and Norfolk metropolitan areas. The princi-
pal adverse impacts for watersheds in this group are high
scores for population density, road density, amount of
roads near streams, sulfate deposition, and impound-
ment density, and low scores for riparian vegetation and
proportion of the watersheds supporting interior forest
habitat at three scales. Their relative cumulative impact
score of 6 is the highest of all clusters.
-------
•?
,** &£
i -f
- " /S5'
iWWBiW-l C
In addition to interpreting the indicators presented in
Chapter 3, the information in this chapter might serve
other purposes, including identifying watersheds that
currently are the most impacted, and by which stresses;
identifying areas that currently are in desirable condition,
but that might be vulnerable to adverse changes because
of human-induced stresses; identifying areas that are in
desirable condition for some resources or services but not
for others; and identifying watersheds where restoration
and risk reduction efforts might be most effective.
The discussion of watershed clusters (above) suggests
which watersheds are the least- and most-impacted
today in terms of the indicators used. The watersheds in
clusters 4, Q, and 8 generally have desirable scores for all
indicators, with only one or two exceptions. In contrast,
the watersheds in clusters 5 and 6 are in the least desir-
able condition, because human populations are high and
the values for habitat indicators are low.
The human population indicators could also be used to
identify vulnerable watersheds. The watersheds in cluster
1 now have relatively desirable conditions for water and
habitat, but population density is relatively high and is
increasing faster than most other watersheds. Environ-
mental conditions related to water and habitat in these
watersheds may be more vulnerable than in other water-
sheds with less population pressure. For example, cluster
8 is similar to cluster 1 in terms of habitat indicators, but
population density in cluster 8 is only about one-fifth of
cluster 1, and is increasing at about half the rate as in
cluster 1.
Some watersheds appear to be in a desirable condition
for one environmental resource, but in a less-desirable
condition relative to another. For example, the water-
sheds in cluster 2 are in better relative condition from a
water quality perspective, but provide little interior forest
habitat. The watersheds in cluster 3 have an opposite
pattern, with relatively more interior forest habitat but less-
desirable values for water-related indicators such as the
amount of crop land cover on steep slopes.
The results of the cluster analysis might be used to guide
restoration or preventative "best management practices"
(BMPs). BMPs are site-specific approaches to minimiz-
ing environmental damage or controlling pollution associ-
ated with intensive land uses. Examples include the
timely establishment of vegetative cover and storm water
detention ponds in areas cleared for residential or com-
mercial development. An example of restoration is the
creation of artificial wetlands to replace natural wetlands
lost in developed areas. Used together, BMPs and
restoration efforts can address a variety of environmental
concerns.
In general, most ecological restoration efforts and man-
agement practices are site-specific, requiring more
detailed information than the landscape indicators can
provide. The information in this atlas, however, can be
used to guide broad-scale restoration efforts, and to
identify areas where more intensive study and restoration
may be needed. Consider the watersheds in cluster 4.
These watersheds have a fairly high percentage of forest
(about 70%) but the amount of forest in large, contiguous
blocks (as measured by the proportion of the watershed
supporting interior forest habitat at three scales) is low
(about 6%). Field studies in the mid-Atlantic region have
shown that some bird species require large tracts of
continuous forest to survive. If the forest land in those
-------
watersheds were configured in large blocks, more of
these area-sensitive species might find suitable habitat
there. This type of information can help land managers to
focus their efforts where restoration and management are
most needed.
These are only a few examples of how relative conditions
can be interpreted regarding overall impacts, conditions
for different environmental resources, watershed vulner-
ability, and ecological restoration. Many other interpreta-
tions are possible because the indicators used in the
analysis can be related to several different aspects of
environmental condition. The reader is invited to use the
information in this atlas to make his or her own interpreta-
tions of landscape conditions in the mid-Atlantic region.
-------
-------
-------
Glossary
Anthropogenic Cover
Land cover associated with human activities, such as
agricultural fields, rock quarries, and urban areas. Liter-
ally, "land cover created by humans."
Bar Chart
A graphic representation of the frequency of different
' data values using rectangles with heights proportional to
the frequencies.
Cluster Analysis
A statistical procedure which groups members of a
population into similar categories (clusters) on the basis
of more than one ecological indicator.
Coarse Scale see Scale
Comparative Assessment
An analysis of environmental characteristics which
proceeds by evaluating members of a population relative
to other members (as opposed to an analysis of charac-
teristics relative to a standard or preferred condition).
Conceptual Model
An abstract framework used to organize ideas and
information into a form that is more easily examined.
These models are often helpful when searching for
commonalties between apparently unrelated phenom-
ena, or when defining the scope of inquiry when organiz-
ing and interpreting measurements of biological
conditions.
Cumulative Environmental Impact
The net result of more than one stress applied to a given
unit of the landscape.
Digital Map
An electronic representation of a portion of the earth's
surface that stores both the geographic location of an
object and descriptive data about the object.
Ecological Indicator
A characteristic of the environment that is measured to
provide evidence of the biological condition of a resource
(Hunsaker and Carpenter 1990). Ecological indicators
can be measured at different levels including organism,
population, community, or ecosystem. The indicators in
this volume are measures of ecosystem-level character-
istics.
Fine Scale see Scale
HUC
Hydrologic Unit Code, used by the U.S. Geological
Survey to reference hydrologic accounting units through-
out the United States. In this atlas, used interchangeably
with watershed.
Index Value
The realized measurement of an indicator for a given
landscape unit.
Landscape Conditions
The apparent status or characteristics of a landscape unit
as measured by one or more landscape indicators.
Landscape Ecology
The study of the distribution patterns of communities and
ecosystems, the processes that affect those patterns,
and changes in pattern and process over time (Forman
and Godron 1986).
Landscape Indicator
A characteristic of the environment that is measured to
provide evidence of the biological condition of one or
more resources at the ecosystem level. See also "eco-
logical indicator" and "landscape ecology".
Landscape Unit
Designed to identify repeating patterns associated with
dominant land uses in an area, and defined by the
relative proportions of forest, agriculture, and developed
(urban) land cover contained in that area.
Model
A representation of reality used to simulate a process,
understand a situation, predict an outcome, or analyze a
problem. A model is structured as a set of rules and
procedures, including spatial modeling tools that relate to
locations on the earth's surface.
Net Primary Productivity
A measure of carbon flux over a given landscape unit,
roughly, the actual amount of organic matter created by
green plants, whether it accumulates in plants, is eaten
by animals, or becomes dead material over a fixed time
interval (after Waring and Schlesinger 1985).
-------
Pixel
A contraction of the phrase "picture element". The
smallest unit of information in an image or raster map.
Referred to as a cell in an image or grid.
Quintile
Any of the four values that divide the items of a fre-
quency distribution into five classes with each containing
one fifth of the total population. For example, one-fifth of
the watersheds in a population have indicator measure-
ments less than the first quintile. In this atlas, a "quintile"
also refers to one of the five groups formed by the
dividing values.
Riparian Zone
The area of vegetation located on the bank of a natural
watercourse, such as a river, where the flows of energy,
matter, and species are most closely related to water
dynamics. In this volume, the "riparian zone" specifically
refers to the linear corridors associated with streams and
stream-side vegetation.
Scale
1. The spatial or temporal dimension over which an
object or process can be said to exist, as in, for ex-
ample, "the scale of forest habitat".
2. The spatial, attribute, and temporal parameters associ-
ated with making an observation or measurement,
usually including resolution, extent, window size,
classification system (nomenclature), and lag. Impor-
tant because measured values often change with the
"scale of measurement".
3. The way in which objects, parts of objects, or pro-
cesses are related as the scale of measurement
changes. For example, fractal models are used to
describe some types of "scaling behavior".
4. The amount of information or detail about an area. For
example, "coarse-scale" maps have less detailed
information than "fine-scale" maps. Related terms
include "broad-scale" (covering a large area). The
cartographic terms "large-scale" and "small-scale" are
(contrary to expectation) equivalent to "fine-scale" and
"coarse-scale", respectively.
Spatial database
A collection of information that contains data on the
phenomenon of interest, such as forest condition or
stream pollution, and the location of the phenomenon on
the earth's surface.
Spatial Pattern
Generally, the way things are arranged on a map. For
example, the pattern of forest patches can be described
by their number, size, shape, distance between patches,
etc. The spatial pattern exhibited by a map can also be
described in terms of its overall texture, complexity, and
other indicators.
Sediment Loading
The solid material transported by a stream, expressed as
the dry weight of all sediment that passes a given point in
a given period of time.
Watershed
A region or area bounded by ridge lines or other physical
divides and draining ultimately to a particular watercourse
or body of water.
-------
.
Appendix: Additional Information about the
Indicators in Chapter 3
This appendix describes the methods that were used to
create the maps and charts of the indicators that are
shown in Chapter 3. The information is organized by
indicator. The reader may also refer to Chapter 1 for a
description of some common computer operations such
as overlaying, cookie-cutting, spatial filtering, and water-
shed ranking.
Table A.1 shows all the indicator values obtained for all
watersheds in the mid-Atlantic region. The listed "HUC"
number can be used to locate individual watersheds by
using Figure 3.3. A few watersheds, indicated by itali-
cized "HUC" numbers in the table, were on the edge of
the region. Most of the area of these watersheds lies
outside of the region, so the values obtained for these
watersheds are probably less reliable than others. The
unreliable values were not used in the cluster analysis in
Chapter 4.
Population density and change (POPDENS and
POPCHG).
The United States Census Bureau compiles population
statistics by sampling units that are not watersheds, and
so it was necessary to convert these (county-level)
statistics to a per-watershed basis. The procedure was
based on differences in road density across the region,
assuming that the population is distributed in proportion
to road occurrence. A map of local road densities in 1
km2 windows across the region was prepared by using
the U.S. Geological Survey Digital Line Graph map of all
roads for the entire region. The total of all these window
scores was then calculated for each county. The popula-
tion within a given window was then estimated by divid-
ing the road density for that window by the total road
density score for the county that the window was in, and
multiplying the result by the total population for the
county.
The population for a watershed was then estimated by
overlaying the map of watershed boundaries on the
derived map of population, and summing the population
estimates in that watershed. Population change was
derived by repeating the procedure for Census data
taken in 1970 and 1990, subtracting the 1970 per-
watershed estimates from the 1990 estimates, and, for
each watershed, expressing the result as a percentage
of the 1970 estimate.
Human use index (UINDEX)
Two different methods were used to create the two maps
of the human use index. The surface map for the mid-
Atlantic region was produced by using a spatial filter. The
window size was about 65 hectares and contained 729
pixels in a 27x27 pixel window. The window was moved
one pixel at a time across the land cover map. At each
step, the number of pixels that had agriculture or urban
land cover were counted. Dividing this sum by the
number of pixels in the window (729) yielded the index
value which was then mapped at the location corre-
sponding to the center of the window. A second spatial
filter was then applied to "smooth" the surface map. The
smoothing filter found the median index value in 9x9 pixel
windows (about 7 hectares). The final map is shown at
7-hectare resolution.
The watershed map was produced by using a cookie-
cutter procedure to extract the land cover information for
each watershed separately. The number of pixels with
agriculture or urban land cover was then counted in each
watershed, and the total was divided by the total number
of pixels for a given watershed to yield the per-water-
shed index value.
Road density (RDDENS)
The United States Geological Survey road maps are very
detailed maps which are available as digital line draw-
ings. To create the surface map of relative road density,
the line drawings were first converted to raster images (or
bitmaps) with a resolution of 90 meters. That is, each
90-meter by 90-meter square in the region that con-
tained at least one road segment was coded as contain-
ing a road. Then a spatial filter was applied to this
90-meter resolution map. The window size was approxi-
mately 1 km2 and contained 121 pixels in a 11x11 pixel
window. The window was moved one pixel at a time
across the land cover map. At each step, the number of
pixels that were coded as containing at least one road
segment were counted. The road density score was
obtained by dividing this sum by the number of pixels in
the window (121), and this score was then mapped at
the location corresponding to the center of the window.
This procedure tends to emphasize the importance of the
first occurrence of a road in a given location, and to give
less weight to subsequent occurrences.
-------
To create the watershed map of road density, a different
procedure was used. The line drawings representing all
roads were clipped using the watershed boundaries, so
that a per-watershed value could be calculated. The
total length of roads in each watershed was divided by
the total area of the watershed. The resulting value
represents road density as road length (kilometers) per
unit area (square kilometers).
Air pollution (NO3DEP, SO4DEP, OZAVG)
The source maps are based on regional-scale models
(J. Lynch, Penn State University; A. LeFohn, Asle and
Associates) which extrapolate measurements made at a
set of sampling stations across the eastern United
States. Nitrate and sulfate wet deposition estimates (in
kg/ha) are from 1987 and 1993. The ozone models
predict the W126 indicator which is a measure of cumula-
tive annual exposure above a critical threshold value.
Values for 1988 and 1989 were used. All of these
source maps were resampled to 90-meter resolution for
our analyses.
Landscape units
The land cover map was analyzed using the spatial
filtering technique. The window size was about 590
hectares and contained 6,561 pixels. The window was
moved one pixel at a time across the land cover map. At
each step, counts were made of the number of pixels
that were forest, agriculture, and urban in the window.
Then, a landscape unit type was assigned to the location
at the center of the window by using rules which are
described in the text. To simplify the resulting map, a
majority-rule spatial filter was applied and the resolution
was reduced to about 7 hectares. The majority-rule filter
examined all landscape pattern types within 7-hectare
windows and assigned the most common type to the
whole window.
Forest and agriculture land cover along streams
(RIPFOR, RIPAG)
Maps of forest and agriculture land cover along streams
were created by using the overlay technique. The map of
streams was converted to a raster format with 30-meter
pixels. This version of the streams was overlaid on the
land cover map to determine the stream length that
flowed through forest and agriculture land cover. The
length of streams flowing through forest and agriculture
land cover, respectively, was divided by the total length of
streams in each watershed to arrive at the index value.
A 30 meter pixel size was used because it was consis-
tent with the pixel size of the land cover map. The pro-
portions would change with different stream pixel sizes,
depending on the amount of forest and agriculture land
cover in the riparian zone defined by the pixel size.
Roads along streams (STRD)
The procedure was similar to that used for the preceding
indicators. Road and stream maps were converted to a
raster format with 30 meter pixels, and then overlaid.
The number of pixels where both a road and a stream
occurred was divided by the total number of stream
pixels in the watershed.
Impoundment density (DAMS)
The U.S. Geological Survey defines large dams as those
that are able to store at least 5,000 acre-feet of water.
The source data were converted into a map of point
locations, and overlaid on the watershed map. The
number of dams in each watershed was then divided by
the total stream length for the watershed to estimate the
density of impoundments. The density estimate is
expressed as the number of dams per 1,000 kilometers
of streams.
Crop land and agriculture land on steep slopes
(CROPSL, AGSL)
Agriculture on steep slopes was mapped by overlaying
the slope map and the land cover map. Percent slope is
calculated from the U.S. Geological Survey digital eleva-
tion model (DEM) as the vertical rise in elevation per
horizontal distance traveled. After overlaying the two
maps, the proportion of watershed area that was crop, or
agriculture, on slopes greater than three percent was
found by using the cookie-cutting technique.
The three percent threshold value was taken from U. S.
Department of Agriculture studies that classified slopes
into six categories. Based on this classification, slopes
greater than or equal to three percent have a greater risk
of soil erosion.
-------
Potential nitrogen and phosphorus loadings to
streams (STNL, STPL)
A literature survey of North American nutrient export
studies (Young and others, 1996, in the Journal of
Environmental Management) provided coefficients for
estimated export (kg/ha/yr) for nitrogen and phosphorus
under different types of land uses. To estimate total
nutrient export potential on a per-watershed basis, the
reported median coefficients for comparable agricultural
uses were multiplied by the amount of land cover in the
agriculture land cover classes. The coefficient-times-
land use model was developed in 1980 for the United
States Environmental Protection Agency by Rechow and
others (US EPA 440/5-80-011, Washington, DC). The
coefficients reported for nitrogen varied from 2.6 to 6.2
kg/ha/yr, with a median value of 3.9 kg/ha/yr. The
values reported for phosphorous ranged from 0.3 to 1.5
kg/ha/yr, with a median value of 0.7 kg/ha/yr.
Soil loss potential (PSOIL)
The Universal Soil Loss Equation (USLE) estimates soil
erosion from agricultural lands as a function of rainfall,
soil type, slope, and land cover characteristics. The
basic equation is:
A=R*K*LS*C*R where A is long-term average
annual soil loss (tons/acre/year), R is the long-term .
erosive potential of rainfall, K is the soil erodibility factor,
LS is the length-slope factor, C is cover and manage-
ment factor, and P is the support management factor
(e.g., strip cropping, buffer-strip cropping). Representa-
tive values for the mid-Atlantic region would be R=200,
K=0.37, LS=0.93, C=0.12, and P=0.5. An R value of
200 (tons/acre/year) is typical of the eastern seaboard in
the northern part of the region (e.g., Philadelphia). A K
value of 0.37 is representative of a loamy soil (e.g.,
Hagerstown silty clay loam at the USDA research station
in State College, PA), A LS factor of 0.93 would be found
on a 6% slope that extended for 200 feet. A C value of
0.12 is representative of corn with a ground cover of
residue (e.g., dead) vegetation, and a P value of 0.5 is
representative of farming practice of tilling and planting
along contours. Under these conditions the estimated
soil loss is 4.1 tons/acre/year. In contrast, if a bare soil
surface was exposed to typical rainfall patterns, that
surface would lose 200 tons/acre/year of soil, or about 50
times as much as the representative example. The
difference is due to soil type (percent sand, silt, clay),
slope, the type of cover, and the type of management.
We created a map for each parameter in the model
(rainfall erosive potential, soil erodibility, length-slope,
cover and support management), and simply multiplied
the values in each map on a pixel-by-pixel basis. R
factor values were taken from USDA Agricultural Hand-
book 537, soil erodiblity was taken from USDA, Soil
Conservation Service digital soil maps, and length-slope
values were taken from USGS DEM data. The C factor
was the median value for corn under all reported crop
residue conditions. The C factors for corn were used
because it is the most common crop in the region, and
the land cover data available did not distinguish between
different agricultural crops. The support management
factor was set to that for contour tillage and planting
because most farmers plant crops along contours, not
perpendicular to them, and more detailed information
was not available.
The area of each watershed with the potential for soil
losses greater than 1 ton per acre per year was then
found by summing the number of pixels in each water-
shed that exceeded this threshold value. The indicator is
the proportion of the watershed above that threshold
value.
Forest land cover (FOR%)
The cookie-cutter procedure to extract the land cover
information was applied to each watershed separately.
The number of pixels with forest land cover was then
counted in each watershed, and the total was divided by
the total number of pixels for a given watershed to yield
the per-watershed index value.
Forest fragmentation (FORFRAG)
Forest fragmentation was assessed here at a resolution
of about one-tenth hectare by using a version of the land
cover map which had only two lumped categories, forest
and non-forest. The fragmentation statistic measures the
probability that a randomly selected forested spot in a
watershed is not adjacent to another forested spot.
Higher values indicate higher fragmentation. The statistic
was calculated separately for the forest cover within each
watershed in the mid-Atlantic region, rather than using a
sliding window technique.
-------
Forest edge habitat (EDGE7, EDGE65, EDGE600)
Forest edge habitat differences among watersheds were
assessed by the sliding window technique. The fragmen-
tation indicator described above was used, but calculated
in a small window that was placed within a watershed. If
the calculated indicator value exceeded one-half the
maximum value for that amount of forest, then the center
of the window was marked as suitable edge habitat.
After moving the calculation window throughout the
watershed, the proportion of the watershed that was
labeled suitable was calculated and used as the indicator
value. The exact window sizes used were 7.29ha,
65.61 ha, and 590.49ha.
Forest interior habitat {INT7, INT65, INT600, INTALL)
The sliding window technique was used to assess interior
forest habitat. The proportion of forest cover was calcu-
lated within a window that was placed within a water-
shed. If the proportion of forest exceeded a threshold
value of 90%, then that place in the watershed was
considered to be suitable interior habitat. After placing
the calculation window everywhere in a watershed, the
proportion of the watershed that was suitable habitat was
determined. These proportions were then used to rank
the watersheds.
The proportion of watershed area supporting three scales
of interior forest habitat was calculated as the proportion
of pixels in a watershed that exceeded the threshold
value for all three window sizes (7.29ha, 65.61 ha, and
590.49ha).
Departure of the largest forest patch size from the
maximum possible for a given amount of anthropo-
genic cover (FORDIF)
Each forest patch was determined with a routine that
finds all adjacent pixels of the same cover type and then
assigns them a unique value, and also retains the origi-
nal land cover value. There are as many unique values
as there are patches in the watershed. From these data,
we created a file of forest patches and sorted it to find
the largest forest patch. A proportion was calculated
using the watershed area as the denominator. The.
proportion was then subtracted from 1.0 minus the U-
index to derive the indicator value.
Calculation of NDVI and its change (NDVIDEC,
NDVIINC, NDVITOT)
NDVI is calculated from satellite spectral reflectance data
in the red and infrared wavelengths, using the equation:
NDVI = (infrared - red) / (infrared + red). The reason
that NDVI in particular, and all vegetation indices in
general, are able to distinguish plants from all other
surface features is that vegetation reflectance jumps
dramatically in the infrared region of light, and is strongly
absorbing (not reflective) in the red region. For vegeta-
tion, typical infrared and red values might be 0.8 and 0.1
respectively, giving an NDVI value of 0.78. For other
earth surface features, the infrared and red reflectance
values are more similar. Because of this, the numerator
tends to be close to zero (or even slightly negative) while
the denominator tends to double. NDVI values for
nonvegetative surfaces are typically close to or less than
zero. Once the NDVI maps are made for each date,
differences are calculated simply by subtraction. The
resulting differences range negative to positive, centered
on zero (0). Values close to zero indicate that there has
not been a change in land cover. Calculating the differ-
ence of temporal satellite images usually yields an
approximately normal distribution. For a normal distribu-
tion, about 70% of the values are within one standard
deviation of the mean, which is zero in this case. Previ-
ous research has shown that one standard deviation is
an accurate threshold to distinguish change from no
change. We chose one standard deviation as our
change/no change threshold.
NDVI change within watersheds (1STDEC, 1STINC,
1STTOT)
Observed values for the three aspects of vegetation
change come simply from the change that occurred in
the first order stream region. Expected values come from
the product of the change over the whole watershed
multiplied by the proportion of the watershed in the first
order stream region. It was necessary to choose a
threshold to decide if a calculated difference between
observed and expected was significant. We chose
±0.25% of the watershed area as the threshold. This
threshold is arbitrary but splits the observed versus
expected map for total vegetation change into approxi-
mately equal thirds.
-------
NDVI loss on steep slopes (NDVI3%)
Vegetation loss on slopes greater than 3% was created
using the overlay technique. The NDVI change data was
overlaid upon the USGS DEM data which were reclassi-
fied into percent slope. Proportional values were calcu-
lated by dividing the amount NDVI loss, gain, or total
change by total watershed area.
-------
HUC
£04010^
2040103
a0401Q4
2040105
^2040106
2040201
2040202
2040203
2040205
2040207
20501 01
2050103
2050104
2050105
2Q501Q6
2050107
£050201
2050202
-2050203
2050204
1050205,
2050206
;20S03Q1
2050302
tlZ0503Q3
2050304
?2050305
2050306
;2flBOQQ2
2060003
^2060004
2060005
-IQSOMfi.
2060007
troops
2060009
-H060010
2070001
2070(002
2070003
'2070004
2070005
2070006
2070007
2070008
2070009
2070010
20700JL
ijQaQlO,.
208010
2080104
208010
2080JOJ
208010
"208010
208010
POPDENS
51 72
114.53
76.16
625.06
488.13
291.56
3533.76
81635
596 15
171 27
6398
JL15.7
28
153.98
74.74
178.88
87.48
20.45
26.74
85.36
61.71
58.67
145.45
109.05
50.56
44.5
i 279.69
254.35
207,9*
707.08
695.28
55 52
86158
83,15
100.37
634
86.09
13.12
432*
2453
690*
4046
42.73
181.1
617 1'
24976
1154.6
J73.5
29.9
31.3
49 1
36 7
32 7
165.6
1898.0
49.7
POPCHG
54
35
81
25
13
20
,5
-1
15
?9
?1
8
4
1
0
0
1
19
4
10
R
I n
R
?R
71
an
3f
J2
fau
32
38
3/
.j45
47
3«
3n
c
37
31
c-
49
55
33
56
14
19'
19
84
_8_
3
UINDEX
3.08
5 25
6.54
[4.03
32.46
64.36
57.84
50.95
59 22
58.17
17.16
27.66
35.77
43.38
32.6
30.97
J2J1
3.21
5.19
29.48
10.95
30.12
40.54
29.89
31.5
28 5C
50.46
69.3'
63.37
61,3_1
38.56
60.24
-49,37
32.94
52.36
37.45
47 3(
18.22
16.71
14.&
49.4i
42.42
357'
57.6?
66.4
54.3
J30.8
24.0
38.3
33.5S
23.8*
28 6
23.6
60.9
1434
RDDENS
2.74
1.7
3.61
7.71
2.73
6.92
6.01
3.32
3.17
1.74
1.38
J10.13
2.62
10.22
1.45
2.19
1.67
0.89
0.98
1.47
0.99
1.54
1.77
1.51
1.3£
1.2£
2 25
2 5^
1.6£
3 71
3.51
1.7
2 3^
1.37
1.67
1.4J
1.6-
0.85
1.31
1.0!
1.6£
1.47
^L,38
1.89
1.92
1.9;
4.1
1.6
1.5
1.2
1.2£
1.2
1.3
1.7
5 7
I i!e
NO3DEP
634
570
536
629
677
696
792
850
H8V
588
594
_872
984
941
852
814
993
2098
2176
2052
21fa/
200S
2031
I91F
l/4b
1966
2041
2081
18VV
2071
1688
1645
1/44
1510
16/4
147f
14Hh
1516
KV>
14b
1695
1371
1286
1435
166<
200
160}
1430
1325
122
130
1V1*
116
129
134:-
|13/
SO4DEP
050
087
160
443
2435
2634
281 /
2799
2913
X4H3
961
:459
2815
2653
2518
2563
2882
3020
3182
2934
31/2
2886
2967
278C
262£
2872
3016
314t
2802
3168
2576
253C
2691
2324
2441
V3o:-
233C
242?
258*
V336
2605
2285
?14:
2309
VHH
305i
2532
224
2096
207;
2075
199
195
207
V1b
1 216
OAVZO
34V/ !
5.45
37.81 !
43.79 '
8.24
47.08
47.72
1.66
7.64
4911
I2.75
i7.16
37.23
38.48
34.91
I3.56
39.92
40.02
37.7
37.9S
36.67
33. 1£
36.01
39.37
4V VV
10.4;
41.1;
44.35
48.47
47.9?
4».1 /
46.6J
48.81
4o.o2
44.64
43.62
44V/
49.2;
46.1*
49.0"
48.0!
53.0'
53.6!
52.87
50.7E
48.83
49.8
48.0
43.9
53.8
46.6E
46.6*
46.2
42.6
40.6
1 43 7
RIPFOR
6.16
91.94
4V b4
7.06
83.24
1.17
56.29
39.06
3.35
86.21
84 bH
'5.89
56.73
59.03
74.34
'9.01
93.47
34.8
71 b
87.96
'4W
37.61
57.52
r0.4{
73.57
166.13
56.85
76.7J
34.96
39.4
82.02
75.81
31.6*
30.7!
76.7'
84.9;
88.9'
66.79
68.3;
7b /!•
70.6
65.5;
53.3'
70.6
83.8
86
70.8*
88.7
93.5
89.7
65.8
i 179.1 2
RIPCROP
3.04
72
2.82
0.95
12.89
17.36
10.27
6.48
9.83
4.98
1631
6.38
41.11
27.08
6.28
4.41
o.lo
4.56
27.31
1.78
25.1
(2.4€
30.92
30.57
27.9J
33.4
43.6J
26.51
26.0E
32.1*
1 15.46
25.9
1/ /b
17.6!
23.9
141
12.34
35.58
35.6'
30.0;
31.96
35.15
50.22
181.
10.4
33.4
9.6
6.6
! 11.43
4.O
91
Uu.u
1
CO
651 :
3.05
5.69
3.68
7.18
6.87
0.65
10 1b
6.68
2.71
6.57
bH/
5.63
5.48
8.32
3.91
5.57
/81
5.85
7.8£
7.8
8.22
Y4H
6.56
7.6*
8.22
2.77
7.14
2.8
2.92
•i -iR
2.99
2.32
2.8,
4.72
5.82
4.67
5.2;
5.6'
5.5£
4.O
3.8*
5.7'
5.3
2.1',
31:-
1.43
1.52
i 1.64
7.6;
:l 1.44
to
Q
733
Kfi 41
6.57
7.64
2.02
7.12
9.04
5.89
12.17
0
31.32
2.26
K AK
13 R1
2.85
24.46
4.87
2.81
2.44
3.23
411
9.27
11.8
3.89
3.42
8.9
6.43
0
0
I °
~0
1.31
9.0
12.26
19/
7.93
9.03
4.62
13.31
0.7:
12.5'
7.7
13.2
.14 Kb
23.8
20.1
1/8
55.5
CROPSL
5.99
5.59
175
3.68
7.5
5.48
2.11
9.46
5.4
0.06
8.97
3.84
7 4A
20.31
5.62
0.27
6.29
1-19
1.43
10.8
4.4
9.32
12.46
11.65
12.63
9.32
6.28
17.72
2.23
15.0'
o 07
0.11
8.95
Q
~0
0
0.02
91'
6.68
6.26
9.48
7.59
7.7*
11 H1
15.15
4.54
2.5
6.3£
1.5
13*
1.4
n 9
0.0
0.1
*
^
8.12
7.72
2.08
5.04
1.02
6.98
318
5.15
0.33
0.09
2.28
8.56
94 RR
26.65
21.61
374
7.3
1 £J9
1.95
13.9
6.47
12.89
1 8.58
I4.67
IR1'
14.0*
10.52
28.4-
3.77
22.29
0.2
12.il
Q
"~o"
0
o.o;
11.21
8.2'
8.2c
15.6'
15.08
14.3*
1 23.5
27.37
7.0
3.5
n ^
11.9
2.3
2.0*
2.9
n ^
0.0
0.1
STNO3L
i?0
i30
?RO
490
•00
<5?n
480
\70
510
RVO
330
390
A^n
470
410
390
3?0
260
?70
390
3nn
400
44f
4nr
41 f
39f
4ar
57(
57f
5?f
420
570
490
520
5fiO
490
54f
34f
33f
3?l
470
4VO
40i
/on
490
540
4fif
39
360
1?"
40
36
37
36
~4T
47
6
W.9
50.5
35.2
06.4
73.8
138
98.3
97
)8.2
46.4
531
71.6
82 6
98
79.2
71.2
497
29 3
31 9
7?fi
418
74.3
88.3
74.8
78.6
69.6
98.2
28. /
30.5
11.7
79 6
128.4
103.7
114 8
25.6
103.5
?,1.'
55.f
50.2
48.fi
96.*
79'
73.5
07 ^
10?
117?
91 .f
72
60 6
77.E
73
61.
fi4
61
97.
96
-j
1
9.51
0.19
3.05
7.38
3.57
1.08
1.27
8.38
•1.71
1.06
3.19
22.14
30 16
34.35
26.59
9.73
9.95
1 7
2.23
23.44
8.12
21.48
31.74
23.82
24.46
24.51
43.67
53. 6£
58.24
39.5;
1875
56.14
34.06
26 94
48.67
34.02
38.77
14.9-
10
11.15
43.1*
39.0E
33.46
53 26
49.93
64.87
21.85
21 .8..
20 03
35.35
25.78
19.8-
24.0
14 8
11.0
42.0
S?
1
4.46
9.97
0^.6
4.73
35.79
4.13
31.12
8.01
.6.84
25.99
H.19
71.66
63 67
56.05
66.05
57.35
81.76
96 56
94.01
69.98
88.67
69.09
57.87
59.75
57.12
70.66
48.37
29.62
33.5*
36.87
5892
35.96
46.49
41 64
42.39
54.06
28.2'
81.25
82.43
84.79
49.8E
57.05
63.8*
43 78
41.28
32.85
43.33
62.27
72 1
61.3
57.5
73.0;
67.8
29.3
40.3*
FORFRAG
0.16
2.43
6.16
5.16
3.98
2.39
9.12
5.88
4.47
5.66
1.75
5.97
1.48
7.71
4.66
8.1
1 97
2.87
8.47
4.7
9.93
4.07
7.76
9.18
8.78
1 6.3*
30.06
27.45
30.39
21 42
25.26
24.5
16 69
16.95
14.19
22.47
6.6
7.49
6.09
15.67
10.6*
9.38
1793
25.65
28.4
27.7
14.6*
11 5
13.1
14.4
11.7
12.3
127
32.0
23.7
EDGE?
6.2b
7.87
3.1
4.07
24.85
58.94
59.78
43.65
52.29
42.66
9.44
6.82
2574
31.62
21.96
22.3
6.84
072
2.35
17.21
6.3S
20.15
29.45
18.12
18.08
16.77
40.67
57.23
44.42
51.32
38.09
36.55
24.15
19.78
34.68
8.07
6.32
5.51
33.73
25.39
20.35
39 32
4343
59.2
43.2
16.4
106
21.7
18.3*
9.3
12.8
134
56.6
31.1
EDGE65
U.4/
1.18
2.04
5.78
2.571
6.17
8.13
8.65
1.05i
4.18
1.46?
3.88
7.61!
3.49
1:47
0.01
0.87;
4.79
6.54
22.74
4.03
1 1 .87
10.55
39.6
50.92
30.78
39.38
20.33
22.86
15.16
7.91
22.24
3.16
1.46.
0.83
30.95
18.77
14.56
32.05
28.62
53.77
30.81
5.77
2.55
10.15
9.72
1.43
3.78
7.17
50.23
14.27
-------
HUC
2080110
2080201
2080202
2080203
2080204
2080205
2080206
2080207
2080208
3010101
3010102
3010103
3010104
3010105
3010106
3010201
3010202
3010203
3010204
3010205
3040101
4120101
4130002
5010001
5010002
5010003
5010004
5010005
5010006
5010007
5010008
5010009
5020001
5020002
5020003
5020004
5020005
5020006
5030101
5030102
5030103
5030104
5030105
5030106
5030201
5030202
5030203
5050001
5050002
5050003
5050004
5050005
5050006
5050007
5050008
5050009
POPDENS
39.7
12.45
34.04
37.12
75.53
258.15
274.03
66.38
1205.09
64.49
26.38
49.27
91.15
31.27
31.39
11.28
44.47
78.41
17.99
318.68
111.23
242.11
27.22
73.13
102.53
79.32
124.42
33.25
72.35
165.53
365.61
1469.23
32.63
62.83
183.11
38.85
551.28
422.35
626.81
135.94
344.62
825.99
546.17
84.21
41.53
68.42
19.07
27.29
32.1
13.29
75.54
30.07
197.63
58.43
111.07
113.11
POPCHG
1
30
21
35
bO
32
33
/4
35
24
b
9
1
-6
0
-10
18
432
-b
120
12
4
1
0
-b
0
3
3
b
-4
0
-12
10
-4
2
5
-12
-11
-1b
-4
-10
-10
-11
-9
-2
3
2
42
1
9
4
3
-/
-6
-1
-4
UINDEX
33,78
12.69
24.14
18.69
24.97
24.65
34.64
24.28
60.58
28.62
27.29
17.44
24.61
29.41
29.16
25.76
34.79
41.23
28.13
37.26
13.62
47.21
24.75
11.65
26.03
13.99
40.42
15.1
28.89
26.65
29.56
35.93
17.66
26.11
22.7
11.12
38.93
31
38.73
55.13
52.2
39.84
43.17
28.82
12.81
29.06
13.62
36.53
21.97
19.63
13.3
8.73
8.65
5.84
20.89
6.66
RDDENS
1.18
0.89
1.1
1.33
1.59
1.75
2.55
1.45
4.26
1.74
1.34
1.83
2.06
1.63
1.91
1.45
1.4
2.26
1.68
1.36
3.5
3.74
13.05
2.16
9.74
1.31
1.65
1.32
1.94
1.92
2.16
3.63
1.25
1.62
1.98
0.98
2.75
1.91
4.54
2.74
42.52
3.05
2.16
2.98
2.58
3.11
1.22
1.54
1.19
0.9
1.54
0.88
1.3
1.15
1.76
0.98
NO3DEP
1379
1304
1308
122b
1233
1115
1193
1150
1286
1102
1118
1006
1045
1041
1129
1122
1158
1137
1119
1326
928
2122
2033
2167
2125
2192
2084
2127
2085
1943
2032
2071
1731
1631
1714
1899
1853
1987
1822
1957
1689
1950
2093
1731
1602
1498
1569
1085
1094
1406
1212
1463
1359
1581
1457
1251
SO4DEP
2184
2139
2159
2086
2107
1931
1961
197S
2082
1869
1928
1726
1794
1790
1948
1903
1916
1896
1909
2185
1632
2529
2810
3009
2781
2949
2709
2966
2922
2849
2929
2952
2649
2588
2656
2860
2811
2938
2749
2667
2430
2795
2923
2787
2690
2501
2585
1844
1802
2243
1949
2298
2223
2525
2409
2038
OZAVG
43.88
51.11
49.47
51.15
51.54
46.93
41.79
46.31
40,04
50.25
48.4
48.82
48.81
49.3
44.66
42.9
41.25
40.3S
43.47
39.91
48.06
41.69
38.96
40.98
41.41
42.74
43.43
41.64
40.22
38.97
38.24
38.17
44.23
40.49
39.19
45.14
38.03
41.24
38.07
41.81
41.64
39.19
39.75
36.75
38.28
39.58
41.06
48.85
51.96
51.97
49.59
48.37
43.21
43.8
40.05
42.62
RIPFOR
84.96
85.95
72.23
87.63
83.87
91.64
86.14
92.62
79.56
83.05
93.36
90.32
93.1
93.1
93.45
94.5
90.12
87.53
93.95
84.22
89.42
80.73
77.97
87.5S
85.32
93.63
77.2
93.99
85.21
87.59
81.42
80
83.37
64.56
76.38
94.26
65.76
81.4
73.24
70.66
73.02
80.12
77.94
77.37
76.49
61.85
76.87
68.51
77.6
85.59
91.35
93
86.14
89.31
73.43
89.72
RIPCROP
15.83
15.34
23.82
13.65
18.87
8.44
7.8
7.15
11.52
18.64
8.91
10.22
7.41
8.68
7.37
6.18
10.73
14.57
7.6
15.45
10.54
20.72
24.3
12.48
16.48
5.73
24.72
5.13
14.89
11.04
17.45
11.78
17,16
34.57
20.95
5.72
29.49
17.1
17.1
28.77
20.14
12.03
21.26
21.4
22.77
36.87
23.62
32.64
23.38
15.33
7.07
6.18
5.6
9.39
24.37
7.24
1
to
2.16
3.72
4:37
3.44
3:29
2.18
3.1£
1.59
3.92
4.5£
1.17
4.17
1.89
1.73
1.56
2.16
2.11
2.48
1.78
2.75
4.87
4.5
5.07
4.54
5.2S
3.27
3.67
3.68
6.75
6.35
8.51
9.67
5.42
8.53
10.11
3.58
11.67
6.24
8.12
4.22
4.56
6.7?
4.89
8.8
6.99
7.01
10.69
7.29
7.65
3.84
6.06
4.17
12.91
8.85
11.52
11.79
to
§
6:9
2.54
3.82
12.8£
30.34
36.32
16.93
24.81
20:78
6.37
•9.46
10.9
19.2
13.73
10.1?
13.62
17.7
17.47
7.52
6.56
0
1.67
0
1.58
0
2.6
4.61
2
5.74
13.49
12.47
10.23
2.85
15.9
30.81
3.95
11.99
9.3
14.92
5.98
0 ..
5.9
5.15
10.82
3.22
11.73
3.19
4.01
8.56
2.34
5.63
2.55
5.8
2.31
6.58
5.33
CROPSL
0.02
5
7.59
2.99
3.69
1.42
0.29
1.4
0.16
6.42
4.23
5.68
5.91
8.65
5.07
0.99
0.28
0.28
2.9
0.04
1.81
6.06
10.61
3.61
7.54
2.81
7.96
5.14
13.64
9.76
11.41
7.62
5.87
5.72
6.02
4.62
9.75
11.21
5.92
8.21
8.96
6.85
8.27
5.44
2.19
6.11
3.72
11.7
9.16
8
4.18
2.74
2.23
2.48
5.51
1.9
ti
0
^
0.05
6.85
13.9
5.78
7.58
2.74
0.5
2.81
0.3
11.01
6.62
7.77
7.19
11.26
7.65
1.6
0.36
0.32
4.38
0.04
9.3
13.1?
17.37
6.05
16.12
5.87
19.74
6.62
16.91
12.14
16.2
12.48
8.58
9.13
9.41
6.2
16.14
15.57
13.1
21.2
19.2
13.39
18.12
14.71
4.07
13.5
6.46
22.43
14.43
11.41
6.71
4.15
2.79
3.77
10.12
2.79
STNO3L
500
310
350
3?0
350
350
390
340
520
37C
370
330
370
400
380
370
430
480
390
ROO
?90
450
370
300
360
310
4?0
320
390
380
390
390
330
360
350
300
410
400
390
470
450
400
4?0
360
310
370
310
410
350
340
300
?80
?ao
?70
340
270
to
104.5
44.8
57.8
476
56.4
55.9
71.4
551
113.9
64
65
51.7
65.1
73.7
688
661
84.6
1037
70
14? 1
37.5
879
6?4
41?
584
42.5
77
47.4
7?, 7
66.9
701
71 ?
51.6
6?
57?
41.9
78.1
73.5
698
934
899
71 9
793
58.7
4?.8
61.5
43.9
747
57.5
55.?
4?1
36
33.6
33
52.5
32
-J
I
40.75
11.06
22.79
14.98
21.21
17.93
18.6
19.17
25.65
22.97
23.94
13.78
21.12
27.66
25.2
21.41
31.45
36.48
23.39
28.55
11.35
36.89
19.53
8.56
22.93
10.96
37.99
9.61
22,19
19.45
24
18.76
14.83
21.23
16.98
8.19
28.11
24.8
23.29
47.66
41.75
23.8
33.25
25.04
10.43
25.38
11.29
32.76
19.18
16.9
8.36
5.62
2.08
3.7
15.48
2.51
xp
OS
s
23.42
86.63
75.43
80.38
74.21
73.47
59.6
74.39
30.4
69.88
69.36
81.68
74.11
69.94
67.14
73.32
64.39
56.79
71.02
51.28
86.05
52.32
75.16
87.42
73.75
85.27
58.84
84.57
70.59
72.62
69.99
63.3
81.57
73.14
76.26
87.64
60.5
68.13
59.71
42.73
46.39
59.78
56.29
69.54
85.06
67.51
85.98
62.75
77.42
79.78
85.59
90.18
90.21
93.57
78.09
92.86
FORFRAG
26.56
4.48
6.46
8.07
11.01
10.6
15.21
10.75
24.63
12.31
12.47
10.08
13.88
13.68
13.89
10.96
12.71
14.51
11.67
9.73
10.59
?1?9
13.72
5.17
10.09
5.9
16.81
6.54
13.06
11.03
13.37
19.54
8.05
13.53
11.66
5.53
19.14
11.96
19.74
25.47
25.94
20.62
19.94
13.68
7.17
13.25
7?4
13.23
8.2
6.36
7.81
4.88
5.25
4.07
9.79
4.8
EDGE7
39.3
5.37
10.62
7.16
10.5
10.64
19.21
8.4
45.85
13.03
8.21
4.36
7.25
7.54
11.47
8.51
12.3
15.25
9.9
16.19
1.09
26.03
10.39
4.51
11.11
5.13
17.7
6.01
11.58
12.34
14.37
19.93
6.72
10.84
8.47
3.48
20.31
16.55
22.36
30.5
28.53
21.79
22.1
9.53
3.17
12.77
3.47
16.22
8.42
6.99
4.5
2.36
3.46
1.14
8.89
1.8
EDGE65
22.74
1.3
4.34
1.55
2.71
4.12
10.38
1.55
40.22
5.28
1.12
0.61
1.42
0.73
1.94*
0.92
1.49
1.41
1.42
9.22
0
11.11
1.44
0.61
1.76
0.64
2.78
1.09
1.84
3.35
5.08
8.42
1.66
2.2
2.56
0.47
7.26
6.47
10.03
9.99
7.89
6.3
5.6
2.33
1.21
6.01
0.89
7.78
2.68
3.38
1.17
0.33
0.96
0.13
5.01
0.54
-------
HUC
J5070101
5070102
S870201
5070202
5070204
5090101
5090102
6010101
6010102
6010205
6010206
POPDENS
46.89
89.46
45.03
29.52
156.39
256.26
40.7
34.4
55.96
40.31
45.63
POPCHG
-1
-2
-14
1
10
-2
1
2
10
11
17
UINDEX
4.96
11.25
4.79
5.74
21.75
27.73
8.47
24.27
42.1
25.27
27.38
RDDENS
1.06
1.34
1.44
1.57
7.21
10.33
1.16
1.02
2.61
1.44
1.82
NO3DEP
1093
1352
1118
1071
1356
1467
1321
967
892
1023
1115
SO4DEP
1776
2206
1813
1724
2149
2380
2128
1626
1508
1701
1857
§
§
45.37
41.48
44.04
42.91
41.64
39.98
42.08
43.73
45.1
42.3
41.3
RIPFOR
91.18
76.25
89.51
89.89
67.11
63.35
83.85
73.95
58.16
73.3
73.13
RIPCROP
4.63
24.92
5.39
7.25
29.72
30.67
17.93
28.9
39.9
26.25
25.79
1
«o
13.12
15.16
15.91
17.76
8.46
6.08
13.68
9.87
10.01
12.62
9.85
1
10.57
3.57
9,88
3.54
0
2.73
2.68
1.33
2.67
3.72
4.27
CROPSL
1.66
3.42
1.58
2.06
3.91
6.46
1.81
10.87
12.«b
10.47
10.62
*
•*
2
6.2
2.04
2.66
14.1
12.72
6.21
17.78
25.7
17.86
17.87
STNO3L
2/0
290
270
2/0
330
3/0
2«U
360
430
360
3/0
£
62.3
33.9
60.9
80.«
60.3
63. /
-j
!
1.44
7.54
1.43
2.02
17.94
20.14
5.69
22.65
39.61
22.06
23.19
S?
1
94.58
88.27
93.86
93.65
72.55
65.78
90.62
75.2
57.39
74.35
72.24
FORFRAG
4.04
6.73
4.07
4.79
12.78
13.45
5.95
9.07
13.75
10.38
1-1.23
EDGE7
0.7
4.22
0.66
0.74
::ii,.48
12.89
: 2.08
10.92
26.24
11.68
12.31
EDGE65
0.07
1.93
0.05
0.04
5.16
11.44
0.47;
2.74
15.38
4.91
3.99
-------
HUC
2040101
2040103
2040104
2040105
2040106
2040201
2040202
2040203
2040205
2040207
2050101
2050103
2050104
2050105
2050106
2050107
2050201
2050202
2050203
2050204
2050205
2050206
2050301
2050302
2050303
2050304
2050305
2050306
2060002
2060003
2060004
2060005
2060006
2060007
2060008
2060009
2060010
2070001
2070002
2070003
2070004
2070005
2070006
2070007
2070008
2070009
2070010
207001 1
2080102
2080103
2080104
2080105
2080106
2080107
2080108
2080109
EDGE600
0
0
0.14
9.76
14;43
26.42
43.52
24.7
19.32
9.47
0
0.81
0.54
2.15
0.39
4.48
0.03
0
0.11
5.51
0.14
9.62
7.41
2.76
2.4
2.03
29.21
35.28
10.2
20.41
2.96
2.11
6.92
5.65
2.33
1.5
12.82
0.48
0.04
0.03
21.35
6.21
5.26
16.68
8.66
37.89
13.34
0.83
0.2
1.4
3.51
0
0.23
3.33
40.91
4.21
fc
62.69
54.61
77.92
23.87
45.83
3.08
5.41
22.3
8.13
6.99
57.64
43.02
40.59
26.52
37.38
44.31
63.62
92.46
88.1
56.51
79.36
54.26
39.86
57.02
51.45
55.69
33.89
10.84
10.75
11.6
27.46
11.19
17.44
21.7
19
31.02
10.26
67.14
65.74
70.66
32.09
43.11
49.5
25.69
16.72
15.47
15.67
36.16
48.65
39.85
33.6
46.99
42.67
33.94
9.37
13.65
28;9
5.21
4.23
4.06
14.16
2.3
6.64
10.66
7.07
15.96
2.94
57.23
:6SM,i
60.05
24.13
38.02
43.02
20.17
9.03
11.12
7.38
21.41
31:48
27.29
19.55
28.37
25.97
18.94
4.83
3.75
INT600
36.04
26.89
69.84
9.3
31,15
0
0.07
7.44
0,
0.02
22.75
4.04
19.69
1.03
13.38
22.96
39.67
91.07
81.4
32.95
71:3 A
41.31
19;99
34.95
23.64
29.66
,21.51
1.68
0.87
0.23
2.8
0.13
0.63
0.97
0.61
4.23
0.26
42.4
37.31^
46.91
14.71
33.32
36,29
15.6
2.49
7.9
2.08
7.88
13,05
15.09
7.21
8.09
8.57
4.75
0.87
0.11
INTALL
:• 30.52
22.07
61.22
8.51
27.7
0
0.07
6.77
.•:-:-.'.-0 ••-•;'••••
0.01
18.92
3.36
17.48
0.6
12,34
20.54
35,72
85.92
78.08
30.97
66.1
37.99
•18.64 ;
32.45
21:18
27.32
20.56
1.6
: 0,77
0.17
•2.4 -••"
0.12
0.49
0.87
0.56 '.
3.59
0.2 i
38.02
31.94
41.14
13.43
31.93
33.5
14.4
2.39
7.31
1.71
6.74
11.45
14.27
5.94
6.78
7.06
3.89
0.84
0.05
FORDIF
0.0155
0.0152
0.0152
0.0767
0.1048
0.1835
0.1493
0.3117
• 0.2651
0.2807
0:0125
0.0533
0.4922
0.0884
0.0784
0.07
0.0124
0.0009
0.0384
0.0636
0.02i2
0.1993
: 0.1601
0.3166
:;0.2825:
0.067
'0.1743
0.2377
0.2897 ;
0.2996
0.2567;
0.3541
-0.3457
0.3306
0.3035
0.1198
0,4436
0.1305
0.0104
0.0061
0.1475
0.3419
0.0852
0.166
0.2607
0.1898
0.2979
0.194
0.0682
0.0638
0.1415
0.0276
0.045
0.0561
0.2398
0.3904
NDVI DEC
: 4.7
6.46
. -r^---;
18.3
14.35
26.11
18.91
16.96
V27.3 :,,
—
-.-'.••.7.62-'
7.66
3.62 •:
2.8
7.59
10.41
9.11
0.94
1.15
3.26
;1.22
7.74
6.16 ;
5.48
•: 10.49 ':•:
4.05
12.04-
24.86
• 30.52 -y
18.93
Til. 99
—
:14;01
—
.'•'. -^— :-f :-•:.»
—
.-_— - .-••-.•
2.01
8.42 :
7.9
16.79
10.43
21.07
—
:13;64
19.11
12.08
10.09
.^_ •-.,
5.34
15.42
13.5
10.77
13.01
:
NDVIINC
13.07
15.42
- •—£--.••>.
22.81
17.9
15.88
6.61
21.95
:. 12;95:
—
2.8:
4.84
8.12
13.76
8.04
8.19
•-;-"7,76'.
1.22
1.8
11.6
3.61
10.06
22.09
7.77
3.88
13.46
17.02
10.98
-15.38
17.63
14.39-
—
20.38
—
'•: •'•'•"• .. -•.
—
.•__--. .- •.
13.49
.9.33
3.4
7.43
13.15
4.74 >
—
22.01
13.3
26. 16
20.27
.'lii. - •'
19.22
16.07
12.05
14.48
12.81
— ...
—
1-
g
S
i
17.77
21.87
T '-it" -----
41.11
32.25
42
25.52
38.92
40.25
—
10.42
12.5
11.74
16.56
15.64:
18.6
16.88
2.17
2:95 ;
14.86
4.83
17.8
28.25
13.25
14.37
17.51
29.06
35.83
-45.9
36.55
26.39
34.39
—
: ::^1_.'-: -. .
-'• i - -
15.49
17.74
11.3
24:23
23.58
-. 25.81
—
35.65
32.41
38.24
30.36
• ^^ -
24.56
31.49
25.55
25.26
25.82
• — . . ..-.„.
—
1ST INC "
11.11
15.63
•. :- -•:-:'..
20.39
17.08
13.78
5.95
21.32
9.49
2.59
4.42
••-' 7.74
12.77
7.04
6.29
: &A7
1.16
1.16
10.58
3.94
9.55
21.69
6.2
• 3.39
12.1
16.76 ,
10.14
12.01
12.89
13.83
19.69
11.94
8.36
2.4
5.52
9.33
3.44
19.76
12.51
23.93
16.44
15.65
7,9
11.55
13.82
10.15
—
J.OJ. 1SL
16.19
21.71
.• . •
38.88
31.36
41.34
25.38
38.56
32.14
9.93
12.42
10.98
15.25
"13.43
14.68
16.76
2.02
1.91
13.39
; 5.03
17.11
27.58
11.18
12.28
15.63
28.68
35.93
37.87
27.43
24.72 V
33.78
__._ • -
13.67
16.27
8.87
18.35
17.31
19.98
33.54
30.89
36.16
25.27
20.48
16.23
25.95
24.65
21.97
—
1ST DEC
5.08
6.08
18.49
14.28
27.56
19.43
17.24
22.65
7.34
8
3.24
2.48
6.39
8.39
- 8.59
0.86
0.75
2.81
1.09
7.56
5.89
4.98
;••;• 8.89
3.53
11:92
25.79
25.86
14.54
10.89
14.09 -
1.73
7.91
6.47
12.83
7.98
16.54
13.78
18.38
12 23
8.83
-.'.'•'
4.83
8.33
14.4
10.83
11.82
NDVI 3%
2.83 !
3.13
529
5.81 !
3.9
2.69
5.65
7.3 • i
5.65
5.26
2.34 I
1.48
4.89
6.15
6.12 >
0.68
0.62 i
1.53
0.79 '
348
3 22 '
3.19
6.25 -'
2.41
2 7 ''
10.04
1.71 "-•
8.7
2 29 '"-•'
3.58 ]
. • .......
1.54
6.76 i
6.26
6.05 I
7 82
16.16 i
4.76 ]
7.14
3.18 I
1.31
• :•••-"!
1.59
- '1.53 T
0.91
0.71 !
0.38
.. . . ,(
-------
HUC
2080110
2080201
2080202
2080203
S2080204
2080205
h2080206
2080207
;2080208
3010101
=3010102
3010103
^3010104
3010105
3FIOT06
3010201
13010202
3010203
3010204
3010205
3040101
4120101
13130002
5010001
H010002
5010003
S010004
5010005
15010006
5010007
screws
5010009
P020001
5020002
S020003
5020004
10020005
5020006
1030101
5030102
ilj03Q102
5030104
1030105
5030106
gQ3020l
5030202
=§63020':
5050001
6050002
5050003
5050005
80S0006
5050007
iOBOGQE
5050009
EDGE600
15.67
0.01
0.31
0.21
0.3
1.5
3.85
0.16
31.24
2.32
0.01
0
0.21
0.02
0
0
0.03
0
0
2.72
0
5.82
0
0
0
0
0
0.04
0.03
0.08
0.54
2
0.08
0.04
0.54
0
2.38
1.19
2.88
1.19
0.06
0.1
0.06
0.13
0.03
1.41
0.16
2.12
0.31
0.42
0,28
0
0.06
0.01
0.61
0.01
|
5.57
76.55
61.71
62.04
50.62
50.61
33.96
49.66
11.51
45.59
40.49
57.67
41.28
37.9
37.96
47.98
37.22
27.86
44.8
35.31
60.67
19.82
47.84
76.18
51.79
71.56
28.38
70.06
44.12
50.76
44.11
29.9
63.21
44.65
49.69
74.06
27.37
46.1
28.32
10.02
11.38
25.5
23.79
39.52
68.27
41.77
68.59
38.77
59.44
65.77
67.96
78.63
80.31
84.73
57.63
83.41
§
0.56
69.05
53.48
49.98
36.44
34.37
20.57
32.8
4.69
32.64
22.25
44.12
23.17
19.92
19.78
30.31
18.84
8.76
27.8
27.07
49.03
6.26
32.74
67.35
35.22
62.53
11.56
61.51
27.17
37.46
29.65
15.34
51.2
26.22
31.51
65.63
12.41
34.6
13.74
1.22
1.75
9.46
9.89
21.92
57.06
26.32
58.65
26.82
47.4
56.66
57.26
72.19
76.16
80.66
45.7
80.34
1
0
58.63
44.89
36.17
21.03
16.37
8.22
16.28
0.62
19.57
5.61
31.31
7.14
3.5
5.43
11.43
4.94
0
11.23
18.57
31.11
1.55
18.68
55.74
15.49
55.53
1-56_,
53.3
9.57
23.23
15.29
3
41.13
9.3
17.21
58.37
3.15
23.72
2.75
0
0
0.15
1.96
9.19
47.82
10.25
48.55
16.19
32.46
47.25
47.88
67.43
71.8
80.63
31.3
81.19
INTALL
0
55.34
43.15
32.21
19.05
13.26
7
13.08
0.6
17.84
4.49
26.14
5.7
2.72
4.39
9.45
4.24
0
9.7
18.15
25.15
1.31
15.68
51.58
13.37
50.1
1.38
49.42
8.16
20.8
13.96
2.04
36.21
7.35
13.66
52.02
2.73
20.98
2.13
0
0
0.15
1.65
6.56
37.49
8.43
39.75
14.96
29.34
42.89
39.43
59.85
63.56
70.8
25.78
70.46
FORDIF
0.3528
0.0552
0.0493
0.0662
0.017
0.0307
0.1967
0.0274
0.2171
0.0352
0.0666
0.0158
0.0365
0.0157
0.075
0.018
0.0319
0.02
0.0187
0.2227
0.0026
0.194
0.0119
0.0108
0.0294
0.0211
0.207
0.0135
0.024
0.038
0.3209
0.0516
0.0247
0.0202
0.0777
0.009
0.1839
0.133
0.093?
0.3218
0.134
0.1971
0.1131
0.0372
0.0117
0.2004
0.0099
0.08
0.0236
0.0208
0.0142
0.0144
0.0067
0.0025
0.2334
0.0027
NDVI DEC
—
2.04
1.44
4.25
—
8.06
16.84
11.33
—
6.54
8.91
3.84
5.76
7.1
9.28
19.73
27.46
25.83
10.52
—
2.73
—
—
—
—
—
—
—
10.82
8.63
'5.81
5.81
3.89
4.21
4.26
2.92
5.56
6.28
—
—
—
—
5.54
—
—
—
—
13.17
—
4.91
—
5.05
—
6.12
—
—
NDVI INC
-^ .
6.84
16.81
7.87
14.86
14.83
10.27
—
2
5.34
3.54
8.78
4.27
4.71
7.89
6.15
8.58
12.67
—
2.45
—
—
—
. : • '
9.12
9.19
9.53
8.21
8.87
14.51
12.24
5.31
7.24
10.28
— -
—
—
—
16.04
—
—
—
—
4.67
—
3.77
—
2.73
.. —
2.02
—
—
NDVI TOT
—
8.88
18.25
12.13
-^-- -
22.92
31.67
21.6
',:•_ — ..
8.55
14.24
7.38
14.54
11.37
13.99
27.62
33.61
34.41
23 2
—
5.18
—
. __^
—
•• —
—
. .._.„_
—
19.94
17.82
15.34
14.01
12.77
18.72
, ::16.5
8.23
12.8
16.56
.-_ - •
—
: ,— '
21.57
—
—
—
—
17.84
—
8.69
—
7.77
-. —
8.14
—
—
1ST INC
—
6.16
13.72
3.97
•'-; — ' - ••
13.75
12.25
10.21
.-— •
1.62
4.94
3.49
9.17
4.26
4.79
8.22
6.27
8.32
14.19
—
2.51
—
—
—
—
—
—
—
6.5
9.34
:9.64
8.79
•••"•" 8.48
13.99
: 10.77
5.11
- 5.41 •
10.87
:""•— • •- -
—
• : /~i,J':'
—
13.08
—
—
—
—
4.17
—
3.32
—
2.17
—
1.19
—
— <•
1ST TOT
: — •••••
7.72
14.73
6.34
— -•
21.91
28.56
22.07
—
7.16
14.43
7.08
15.17
12.11
14.6
29.04
34.09
33.75
25.65
—
4.85
—
—
—
—
—
—
—
:,,14:17
18.06
15.43
14.51
12.05
17.66
14.73
7.99
.V9.33
17.34
*--—'-••:• ',-
—
•• • -—/-•• ",
—
17.46
—
—
—
15.37
—
8.59
—
6.48
—
4.83
—
—
1ST DEC
—
1.56
1.01
2.37
••.— • ••
8.16
16.31
11.86
' — -••
5.54
9.49
3.59
6
7.85
9.81
20.82
27.82
25.43
.11.46
—
; 2.34
—
—
—
—
—
—
—
, 7.67
8.72
:-. 5.79
5.72
3.57
3.67
3.96
2.88
,3.92
6.47
... .— -•• ••••••
—
.•-,;. -rr.'..:-..-:
—
: 4.38
—
—
—
—
11.2
—
5.27
— .
4.31
T-
3.64
—
—
NDVI 3%
• " • 1
1.26
0,85 •
1.37
1.02
• 0.31 I
1.34
— : I
3.16
2.03 ;
1.89
1.76 .;
2.48
2.31 !
0.88
0.18 1
0.09
1.56
—
1,88
—
—
—
— <
—
— I
—
7.71
5.33
3.84 '
3.57
2.79 i
2.48
2.63 •
2.22
3.25 ;
3.84
"•,/,-rS-1,: - i
—
' — ••!
2.65 :
—
— 1
—
—
7.5
—
3.8
—
. 4.27
— *
5.69
— '
—
-------
HUC
5070101
5070102
5070201
5070202
5070204
5090101
5090102
6010101
6010102
6010205
6010206
EDGE600
0
0.24
0
0
0.61
5.25
0
0
3.73
0.27
0
fc
86.64
73.52
86.65
84.12
50.41
42.41
77.05
56.19
38.6
54.39
50.48
10
^
83.9
67.08
84.27
82.57
38.03
28.67
70.78
44.82
31.54
43.7
40.56
INT600
87.77
62.67
87.63
82.89
20.94
11.72
68.35
209
22.9
30.61
30.39
INTALL
76.09
52.92
76.84
72.68
16.39
8.97
58.04:
27.45
21.64
26.87
25.64
FORDIF
0.0011
0.0801
0,0117
0.001
0.1134
0.1194
0.0025
0.0545
0:2629
0.0516
0.0684
NDVIDEC
: ' ' — . • '
—
.- — ', ...
—
•_, •-. -
—
..>—•' ':
—
•-. —
—
.. v —'•'•-
NDVIINC
.- !_ , -•--•
'•• ' ..— ' ' '
'••••;— •'••
K.
e
s
i
1ST INC
J.O± J.SI-
1ST DEC
NDVI3%
'•'• . • :. - <
••••-.• >l
I
•
-------
-------
Suggested Further Reading
Appleton, E.L. 1995. A cross media approach to saving
the Chesapeake Bay. Environ. Sci. Tech. 29:550-555.
Bourgeron, RS. and M.E. Jensen. 1993. An overview
of ecological principles for ecosystem management.
Pp. 49-60 in Jensen, M.E., and RS. Bourgeron (eds.),
Eastside Forest Ecosystem Health Assessment, Vol II,
Ecosystem Management: Principles and Applications.
United States Forest Service, Missoula, Montana.
Cooper, S.R. 1995. Chesapeake Bay watershed histori-
cal land use: impact on water quality and diatom com-
munities. Ecol. Appl. 5:703-723.
Correll, D.L., I.E. Jordan, and D.E. Weller. 1994.
The Chesapeake Bay Watershed: effects of land use
and geology on dissolved nitrogen concentrations.
Pp. 639-648 in Toward a Sustainable Coastal
Watershed, Chesapeake Research Consortium
Publication No. 149.
EPA (Environmental Protection Agency). 1994.
Landscape monitoring and assessment research plan.
620/R-94/009, United States Environmental Protection
Agency, Office of Research and Development,
Washington, D.C.
EPA (Environmental Protection Agency). 1995.
Water quality functions of riparian forest buffer systems in
the Chesapeake Bay Watershed. Report of the Nutrient
Subcommittee of the Chesapeake Bay Program.
EPA 903-R-95-004, United States Environmental
Protection Agency, Annapolis, MD.
ESRI (Environmental Systems Research Institute). 1994.
ARC/INFO Version 7. ESRI, Inc. Redlands, California.
Forman, R. T. T, and M. Godron. 1986.
Landscape Ecology. John Wiley and Sons, NY.
Franklin, J. F. 1992. Scientific basis for new perspec-
tives in forests and streams. Pp. 25-72 in R. J. Naiman
(ed.). Watershed Management. Springer-Verlag, NY.
Frenzel, S.A., and R.B. Swanson. 1996. Relations of
fish community composition to environmental variables in
streams of central Nebraska, USA. Environ. Man.
20:689-795.
Golley, F.B. 1993. Development of landscape ecology
and its relationship to environmental management.
Pp. 37-^4 in Jensen, M.E., and RS. Bourgeron (eds.),
East side Forest Ecosystem Health Assessment, Vol II,
Ecosystem Management: Principles and Applications.
United States Forest Service, Missoula, Montana.
Hunsaker, C.T., and D.E. Carpenter (eds.). 1990.
Ecological indicators for the Environmental Monitoring
and Assessment Program. EPA 600/3-90/060. U.S.
Environmental Protection Agency, Office of Research
and Development, Research Triangle Park, NC.
Hunsaker, C.T., D.A. Levine, S.P. Timmins, B.L Jackson,
and R.V. O'Neill. 1992. Landscape characterization for
assessing regional water quality. Pp. 997-1006 in D.H.
McKenzie, D.E. Hyatt, and VJ. McDonald (eds.),
Ecological Indicators. Elsevier Appl. Sci., New York.
Hunsaker, C.T., and D.A. Levine. 1995. Hierarchical
approaches to the study of water quality in rivers.
Bioscience 45:193-203.
Jones, K.B., J. Walker, K.H. Riitters, J.D. Wickham, and
C. Nicoll. 1996. Indicators of landscape integrity.
Pp 155-168 in J. Walker and D.J. Reuter (eds.),
Indicators of catchment health: a technical perspective.
CSIRO Publishing, Melbourne, Australia.
Kepner, W.G., K.B. Jones, D.J. Chaloud, J.D. Wickham,
K.H. Riitters, and R.V. O'Neill. 1995. Mid-Atlantic
landscape indicators project plan. EPA 620/R-95/003,
U.S. Environmental Protection Agency, Office of Re-
search and Development, Washington, D.C.
Kuchler, A.H. 1964. Potential natural vegetation of the
conterminous United States. American Geogr. Soc.
Spec. Publ. No. 36, Washington, D.C.
Lapin, M., and B.V. Barnes. 1995. Using the landscape
ecosystem approach to assess species and ecosystem
diversity. Cons. Biol. 9:1148-1158.
Lowrance, R.R, R.L Todd, J. Fail, O. Hendrickson, R.
Leonard, and L.E. Asmussen. 1984. Riparian forests as
nutrient filters in agricultural watersheds. BioScience
34:374-377.
-------
Noss, R.F., and A.Y. Cooperrider. 1994.
Saving Nature's Legacy: Protecting and Restoring
Biodiversity. Island Press, Washington, D.C.
Ojima, D.S., K.A. Galvin, and B.L Turner II. 1994. The
global impact of land-use change. BioScience 44:300-
304.
O'Neill, R.V., D.L DeAngelis, J.B. Waide, and T.F.H.
Allen. 1986. A Hierarchical Concept of Ecosystems.
Princeton University Press, Princeton, NJ.
O'Neill, R.V., J.R. Krummel, R.H. Gardner, G. Sugihara,
B.Jackson, D.L DeAngelis, B.T. Milne, M.G. Turner, B.
Zygmunt, S.W. Christensen, V.H. Dale, and R.L. Graham.
1988. Indices of landscape pattern. Landscape Ecol.
1:153-162.
O'Neill, R.V., C.T. Hunsaker, K.B. Jones, K.H. Riitters,
J.D. Wickham, P. Schwarz, LA. Goodman, B. Jackson,
and W.S. Baillargeon. 1997. Monitoring environmental
quality at the landscape scale. BioScience 47(8): 513-
519.
Reckhow, K.H., M.N. Beaulac, and J.T. Simpson. 1980.
Modeling phosphorous loading and lake response under
uncertainty: A manual and compilation of export
coeffiecients. U.S. Environmental Protection Agency,
EPA 440/5-80-011.
Riitters, K.H, R.V. O'Neill, C.T. Hunsaker, J.D. Wickham,
D.H. Yankee, S.P. Timmins, K.B. Jones, and
B.L. Jackson. 1995. A factor analysis of landscape
pattern and structure metrics. Landscape Ecol. 10:23-29.
Riitters, K.H., J.D. Wickham, and K.B. Jones. 1996.
A landscape atlas of the Chesapeake Bay Watershed.
Second Edition. Tennessee Valley Authority, Norris, TN.
29pp.
Riitters, K.H., R.V. O'Neill, and K.B. Jones. 1997.
Assessing habitat suitability at multiple scales:
a landscape-level approach. Biol. Cons. 81:191-202.
Roth, N.E., J.D. Allan, and D.L. Erickson. 1996.
Landscape influences on stream biotic integrity assessed
at multiple scales. Landscape Ecology 11:141 -156.
Saunders, DA, R.J. Hobbs, and C.R. Margules. 1991.
Biological consequences of ecosystem fragmentation:
a review. Cons. Biol. 5:18-32.
Schlosser, I.J., and J.R. Karr. 1981. Water quality in
agricultural watersheds: impact of riparian vegetation
during base flow. Water Resources Bull. 17:233-240.
Sparks, R.E. 1995. Need for ecosystem management of
large rivers and their floodplains. BioScience 45:168-
182.
Turner, S. J., R. V. O'Neill, W. Conley, M. R. Conley, and
H. C. Humphries. 1991. Pattern and Scale: Statistics for
Landscape Ecology. Pp. 17-41 in M. G. Turner and R. H.
Gardner (eds.). Quantitative Methods in Landscape
Ecology. Springer-Verlag, NY.
Waring, R.H., and W.H. Schlesinger. 1985.
Forest Ecosystems, Concepts and Management.
Academic Press, Inc., Orlando.
Wickham, J.D., and D. Norton. 1994. Mapping and
analyzing landscape patterns. Landscape Ecol. 9:7-23.
Wischmeier, W. H., and D. D. Smith. 1978. Predicting
rainfall erosion loss: A guide to conservation planning.
Agricultural handbook 537. U. S. Department of
Agriculture, Washington, D.C.
Young, W.J., F.M.Marsten, and J.R. Davis. 1996 Nutri-
ent exports and land use in Australian catchments.
Journal of Environmental Management, 57: 165-183.
* U.S. GOVERNMENT PRINTING OFFICE: 1997 — 678-640
-------
Acknowledgements
Many people contributed to this atlas.
We thank Curt Edmonds, Karl Hermann, Bill Kepner,
Clay Lake, Bob Schonbrod, Sumner Crosby, Terry
Slonecker, Dan Heggem, Rick Kutz, Tom Pheiffer, Rick
Batiuk, Kent Mountford, Nita Sylvester, Kim Devonald,
Stu Kerzner, Laura Jackson, Brian Spavin, Ann Pitchford,
and Don Garofalo of the EPA, and Carolyn Hunsaker of
the Oak Ridge National Laboratory, for their assistance
and input during many phases of the project. We are
especially grateful to Rick Kutz and Tom Pheiffer for
organizing a series of review workshops with scientists
and stakeholders who live in the mid-Atlantic region, and
Rick Cooksey (U.S. Forest Service), Diane Eckles (U.S.
Fish and Wildlife Service), Larry Fogelson (Maryland
Office of Planning), Charlie Yuill (Univ. of West Virginia),
Jeff Waldon (Fish and Wildlife Exchange), Phil Hager
(Worester County Planning Dept), Carl Hershner (Vir-
ginia Institute of Marine Sciences), Cully Hession (Phila-
delphia Academy of Sciences), Bill Jenkins (Maryland
Dept. of Natural Resources), Rick McCorkle (U.S. Fish
and Wildlife Service), Sandi Patty (Maryland Dept. of
Natural Resources), Larry Pamotto (Delaware Dept. of
Natural Resources), Nancy Roth (Versar, Inc.), Bob
Shedlock (U.S. Geological Survey), Ann Sloan (Maryland
Dept. of Natural Resources), Tom Stockton (Maryland
Dept. of Natural Resources), Joe Tassone (Maryland
Office of Planning), Paul Tubach (James City County
Planning Dept), Lisa Wainger (Institute for Ecological
Economics, Univ. of Maryland), Oliver Weatherbee
(Center for Remote Sensing, Univ. of Delaware), and
John Wolf (Maryland Dept. of Natural Resources), who
participated in review workshops and provided comments
on various parts of the atlas. We thank Steve Paulsen,
Gil Veith, Ed Martinko and others in the Environmental
Monitoring and Assessment Program for their gracious
support over the years, and Denice Shaw of the EPA
whose hard work led to completion of several region-
wide databases used to calculate indicators in this atlas.
We also acknowledge Jim Vogelmann, June
Thormodsgard, John Dywer, and Tom Loveland of the
U.S. Geological Survey, EROS Data Center, whose
collective vision has made multiple-scale, regional
assessments possible, and Doug Norton (EPA) for his
efforts in helping EPA to adopt a broader, landscape view
of the environment. We thank Tom DeMoss, Greene
Jones, and Rick Linthurst of the EPA for providing the
institutional leadership for the development of and sup-
port for this and other projects in the mid-Atlantic region.
We are grateful to Kent Mountford, Rick Cooksey, Ron
Landy, Charles P. Nicholson, the TVA Clean Water
Initiative, and the BLM Phoenix Training Center (Phoenix,
AZ) for photographs used in this atlas. We thank Vern
Meentemeyer (Univ. of Georgia), Jim MacMahon (Utah
State Univ.), Ron Foresta (Univ. of Tennessee), Kent
Price (Univ. of Delaware), and Jianguo Wu (Arizona State
Univ. West) for technical reviews of the atlas. We are
grateful to the EPA Science Advisory Board, Ecological
Processes and Effects Committee, for suggestions about
the approaches used in this Atlas. We thank technical
editors Donna Sutton (Lockheed Martin) and Barbara B.
Smith. Atlas layout and graphic design by Garry Pyle
(Tennessee Valley Authority).
The EPA, through its Office of Research and Develop-
ment (ORD), partially funded and collaborated in the
research described here, in cooperation with the Ten-
nessee Valley Authority, and,the Oak Ridge National
Laboratory. It has been peer reviewed by the EPA and
approved for publication. Mention of trade names or
commercial products does not constitute endorsement or
recommendation by EPA for use.
-------
------- |