Formulating and ecosystem approach to environmental protection
FORUM
Formulating an Ecosystem Approach to
Environmental Protection
ecosystem as a volumetric unit delineated by climatic and
O T T O a. G O N Z A L E Z 1
Environmental Results Branch, Office of Policy, Planning and landscape features is suggested. Following this definition,
ecosystems are organized hierarchically, from
Evaluation,
United States Environmental Protection Agency megaecosystems, which exist on a continental scale
401 M Street, S.W. (e.g., Great Lakes), to small local ecosystems.
Washington, DC 20460, USA
Threats to ecosystems can generally be categorized as:
(1) ecosystem degradation (occurs mainly through pollution)
ABSTRACT / The U.S. Environmental Protection Agency (2) ecosystem alteration (physical changes such as water
(EPA) has embraced a new strategy of environmental diversion), and (3) ecosystem removal (e.g., conversion of
protection that is place-driven rather than program-driven. wetlands or forest to urban or agricultural lands). Level of
This new approach focuses on the protection of entire threat (i.e., how imminent), and distance from desired future
ecosystems. To develop an effective strategy of ecosystem condition are also important in evaluating threats to
protection, however, EPA will need to: (1) determine how to ecosystems. Category of threat, level of threat, and
define and delineate ecosystems and (2) categorize threats to "distance" from desired future condition can be combined
individual ecosystems and priority rank ecosystems at risk. into a three-dimensional ranking system for ecosystems at
Current definitions of ecosystem in use at EPA are risk. The purpose of the proposed ranking system is to
inadequate for meaningful use in a management or regulatory suggest a preliminary framework for agencies such as EPA
context. A landscape-based definition that describes an to prioritize responses to ecosystems at risk.
Over the past several years there has b e e n a growing a n d the framework for ranking ecosystems at risk intro-
awareness in the US E n v i r o n m e n t a l Protection Agency d u c e d h e r e i n can be used to help i m p l e m e n t a policy
(EPA) that c o m p l i a n c e with media-based (e.g., air, wa- of focusing on ecosystems, r a t h e r than distinct media,
ter, solid waste) regulations can n o t ensure p r o t e c t i o n in p r o g r a m m i n g e n v i r o n m e n t a l protection.
of entire ecosystems (EPA 1987, 1990a, 1994). Over the
past two years the EPA has e m b r a c e d a shift away from
How to Define and Delineate Ecosystems
a media-based program-driven focus for the agency to
one that is place-driven a n d ecosystem-based (EPA
"Ecosystem" is a familiar term to many p e o p l e , yet
1994). State agencies with responsibilities similar to
its m e a n i n g varies d e p e n d i n g on the user. This is true
EPA's may also wish to develop an ecosystem-based ap-
even for those working in the e n v i r o n m e n t a l arena. A
p r o a c h for e n v i r o n m e n t a l protection. To develop an
small marsh n e a r a city, a large forest stand, a g r o u p of
effective strategy o f ecosystem protection, however, the
sand d u n e s o n Lake Michigan, or the entire G r e a t Lakes
EPA a n d o t h e r agencies will n e e d to: (1) d e t e r m i n e
can all be c o n s i d e r e d ecosystems, a l t h o u g h they differ
how to define a n d delineate ecosystems, a n d (2) rank
by a few to several orders of m a g n i t u d e in size. A place-
threats to individual ecosystems at risk as a m e a n s o f
based or ecosystem a p p r o a c h to e n v i r o n m e n t a l protec-
prioritizing agency response. In this p a p e r I suggest how
tion will r e q u i r e a definition of ecosystem that is b o t h
to m e e t these two needs. T h e definition of ecosystem
scientifically defensible a n d administratively practical.
However, most s t a n d a r d ecology texts (e.g., O d u m 1971,
Ricklefs 1983, Begon a n d others 1990) p r e s e n t ecosys-
KEY WORDS: Ecosystemapproach; Ecologicalrisk assessment; Envi-
tem as a vague concept, r a t h e r than as a definable,
ronmental protection; EPA
measurable construct on the g r o u n d . A c u r r e n t defini-
tion of ecosystem in use at EPA is: "a dynamic c o m p l e x
~Current address: US Agency for International Development, Center
for Environment, Room 509, SA-18,Washington, DC 20523-1812, USA. of plant, animal, a n d micro-organism c o m m u n i t i e s a n d
Environmental Management Vol. 20, No. 5, pp. 597-605 © 1996 Springer-Verlag New York Inc.
598 o.J. Gonzalez
their non-living e n v i r o n m e n t interacting in a functional ecosystem as a '~¢olume o f l a n d a n d air" on the earth's
unit" (EPA 1994). This definition is typical o f those surface. Rowe a n d S h e a r d (1981) d e s c r i b e d the land-
f o u n d in standard ecology text books. However, such scape as a hierarchy of ecosystems, large a n d small,
definitions are inadequate, imprecise, a n d n o t workable nested within o n e another. Acknowledging the contri-
in a m a n a g e m e n t or regulatory context. To derive a butions o f these authors, I p r o p o s e the following defini-
better definition, we must first consider an ecosystem tion o f ecosystem:
as a place.
A volume of land, air, and water with natural boundaries, delineated
primarily by landscape features and climatic factors. It encompasses
Concept of an Ecosystem as a Place:
a set of natural ecological processes, organisms, and anthropogenic
A Landscape-Centered View
processes that function within a nested hierarchy of volumes.
W h e t h e r o n e considers an ecosystem a place depends
on o n e ' s general c o n c e p t o f ecosystem. Is the ecosystem T h e advantages of this definition over most others com-
conceived as a function o f the organism o r a function monly used is that it is: (1) functional within a spatial a n d
of the environment? For example, to a wildlife biologist, t e m p o r a l hierarchy o f ecosystems a n d (2) landscape-
a grizzly bear's ecosystem is d e f i n e d by all the land, based, thus b o u n d a r i e s can be d e l i m i t e d in the field
water, plant, a n d animal resources used by the bear. If a n d on maps with a fair d e g r e e of p e r m a n e n c e . As such,
the spatial distribution of those resources were to ecosystems are conceived as places, large a n d small,
change (e.g., high densities of prey animals shifting to nested, a n d functional within o n e a n o t h e r in a hierarchy
new locations), then so would the b o u n d a r i e s of the o f spatial sizes.
ecosystem. In this organism-centered view o f the ecosys-
Hierarchy of Ecosystems
tem, b o u n d a r i e s are drawn a r o u n d the areas used by
the organism(s) (usually animal) o f interest. This is Ecosystem is a term a p p l i e d across a wide variety o f
c o n g r u e n t with the p o p u l a t i o n - c o m m u n i t y view o f eco- spatial scales. For e x a m p l e some ecosystems may be
systems described by O ' N e i l l a n d others (1986), the 10,000 sq km or larger (e.g., Greater Yellowstone ecosys-
naturalist view described by Surer a n d Bartell (1993), tem), while others (e.g., a small patch o f forest) may
be only 1 sq km o r smaller. Functionally, as well as
a n d the bioecologist view described by Rowe a n d Barnes
spatially, ecosystems exist in a nested hierarchy (Figure
(1994). If the needs a n d habits o f the animal were to
1). Watersheds can be used as a c o n v e n i e n t illustration
change, so would the b o u n d a r i e s of the ecosystem.
o f this concept. F o r example, the Great Lakes, which
Thus, the ecosystem is conceived as a function o f the
e x t e n d into seven US states a n d one C a n a d i a n province,
animal.
Alternatively, the ecosystem can be conceived as a constitute a megaecosystem c o m p r i s e d o f m a n y smaller
ecosystems. At a lower level in scale, the Lake Erie water-
function of the environment. In this landscape-centered
shed could be c o n s i d e r e d a large regional ecosystem,
view, ecosystems are fixed places on the landscape en-
compassing physical, chemical, a n d biological resources within which is nested the Detroit River watershed, a
small regional ecosystem. Smallest is the tiny Rouge
a n d processes, along with various organisms. This con-
River, which flows t h r o u g h n e i g h b o r h o o d s n e a r the city
cept of the ecosystem as a fixed place is p r o b a b l y most
of Detroit a n d connects to the larger Detroit River. Each
familiar to geologists, hydrologists, a n d landscape ecolo-
o f these ecosystems is a place, with smaller ones nested
gists a n d is well described by Rowe a n d Barnes (1994)
as the geoecologist view. It is similar to a material cycling within larger ones, f o r m i n g a spatial hierarchy. A func-
tional hierarchy also exists because activities at a h i g h e r
perspective (Suter a n d Bartell 1993), or a process-func-
level in the hierarchy affect ecosystems at the lower
tional view of ecosystems (O'Neill a n d others 1986).
levels. Conversely, i m p r o v e m e n t o f e n v i r o n m e n t a l qual-
Foresters are also familiar with the c o n c e p t o f the forest
ity at an u p p e r level of the hierarchy is often a function
site, which identifies the ecosystem as a physical location
of success in the ecosystems comprising the lower levels.
(Spurr a n d Barnes 1980). T h e l a n d s c a p e - c e n t e r e d view
In a nested hierarchy of ecosystems, h i g h e r levels
o f ecosystems is also consistent with a place-driven ap-
contain a n d are c o m p o s e d of all the ecosystems at lower
p r o a c h to e n v i r o n m e n t a l p r o t e c t i o n because the ecosys-
levels (O'Neill a n d others 1986). Boundaries o f ecosys-
tem has a definite location.
tems may be b o t h structural a n d functional (Mien a n d
Workable Definition of Ecosystem Starr 1982). If the differences f o u n d between one side
of a b o u n d a r y a n d the o t h e r are significant, than the
A broad, place-driven strategy of e n v i r o n m e n t a l pro-
b o u n d a r y is true, or natural. If the differences are n o t
tection that can be i m p l e m e n t e d a n d a g r e e d u p o n by
significant, than the b o u n d a r y is artificial (Allen a n d
all relevant parties must begin with a s o u n d definition
Starr 1982) a n d may n o t define separate ecosystems.
o f ecosystem. Rowe (1961) i n t r o d u c e d the n o t i o n o f an
599
Formulating an Ecosystem Approach
I
M1 of these factors could be used in delineating eco-
( Megaecosyst
m systems.
Delineating Boundaries for Ecosystems
There are a n u m b e r of ways one might reasonably
delineate boundaries for ecosystems. The appropriate-
Large regional ness of one way over another depends on the ecological
ecosystem questions one wants to ask. For example, both the ecore-
gions of the United States delineated by Bailey (1983)
and Omernik (1987) represent divisions of the US land-
Small regional
ecosystem scape into regions, each relatively h o m o g e n e o u s inter-
nally in landform, soil, and other characteristics. Such
delineations can be useful in describing the potential
Local ecosystem
natural vegetation of an area. Another important use
Figure 1. Nested hierarchy of ecosystems. is in predicting the response of a site to m a n a g e m e n t
practices or other h u m a n impacts based on the response
of other sites in the same ecoregion (Bailey 1983, Omer-
nik 1987).
For example, a small wildlife refuge may be designated
However, while ecosystems can be delineated based
inside a larger wetland area. Yet, ecologically there may
on homogeneity, some of ]EPA's interests might be bet-
be no difference between the refuge and the rest of the
ter served using another basis of delineation. This is
wetland not designated as a refuge; thus the refuge
because in employing a place-driven approach to envi-
border would be an artificial boundary. Natural bound-
ronmental protection, a regulatory agency such as EPA
aries should be used for most delineations of ecosystems.
needs to determine how h u m a n activities affect air and
Nevertheless, sometimes artificial boundaries (e.g., po-
water and the places to which the air and water are
litical borders such as county lines) must be used to
naturally transported. To understand how a particular
b o u n d ecosystems into administratively practical units.
place is influenced by activities, one must recognize
According to hierarchy theory, hierarchical systems
the functional linkage between the condition of one
should be "nearly decomposable" (Simon 1962), mean-
location and activities in others. Air and water often
ing they can be divided into subsystems such that interac-
provide the conduit for these functional linkages. Thus,
tions within a subsystem are both more n u m e r o u s and
for EPA and similar agencies, ecosystem delineations
stronger than interactions between subsystems (Platt
will be most useful if they are based on functionality.
1969, Allen and Starr 1982). For example, a megaecosys-
U n d e r an emphasis on functionality, patterns of wa-
tern such as the Great Lakes can be readily decomposed
terflow and airflow functionally linked to an area on
to its five constituent lake watersheds (large regional
the landscape would be used to help delineate a volume
ecosystems), each of which can be decomposed to
of land, air, and water as an ecosystem.
smaller subwatersheds (small regional ecosystems). The
For example, at g r o u n d level, a large regional ecosys-
relative strength and n u m b e r of interactions can be
tem, such as a watershed, is b o u n d e d by landform. How-
used conceptually to help determine where one ecosys-
ever, high above ground, air patterns may transport
tem ends and another begins. Ecological interactions
particulates to and from areas beyond the boundary of
are both constrained and fostered by boundaries
the watershed. Similarly, below ground, an aquifer may
such as:
extend beyond the ground-level boundary of an ecosys-
tem. If activities in areas beyond the ecosystem's land-
(1) landform (e.g., hills, mountains, valleys, eskers,
surface borders affect the aquifer, then they also affect
kettles, kames, river floodplains),
the ecosystem. Functionally, these other areas are part
(2) air patterns (speed, direction, and temporal qual-
of an ecosystem where the land boundaries may be
ity of winds),
smaller. Thus, land, air, and subsurface boundaries of
(3) patterns of precipitation and temperature,
an ecosystem need not be congruent (Figure 2).
(4) Land u s e / l a n d cover (e.g., agriculture, urban, for-
Ideally, the higher-level boundaries selected for eco-
est, grassland, wetland), and
(5) chemical and physical traits (e.g., concentrations systems should be fairly p e r m a n e n t and be relevant to
of certain chemicals in air, water or soil; tempera- EPA's traditional authorities over air and water quality.
ture of stream water). The suggested scheme for delineating ecosystems is:
600 o.J. Gonzalez
f
J
f
f
/*
/
Upper air layer
J
J
f
/
Lower air layer
Land and Water
(surface)
(underground)
Groundwater
Aquifer ~__ _
7
f
Figure 2. Conceptual boundaries of an
J
\
ecosystem.
distribution of solar energy. In addition, landform is
Climate (considered at macro and local scales)
the most stable c o m p o n e n t of an ecosystem, and thus
Hydrology (watersheds, subwatersheds)
provides a basis for ecosystem delineation within a cli-
Land U s e / L a n d cover (agriculture, industry, forest,
matic regime (Rowe and Sheard 1981). Therefore, land-
wetland, etc.)
form boundaries can be useful as boundaries for pro-
Political boundaries (town lines, county lines,
cesses. At a fine scale within a watershed, land use and
etc.)
land cover will be useful for making practical delinea-
tions of ecosystems.
Climate (as it relates to wind patterns and patterns of
wet and dry deposition) and hydrology (as it relates to
Implications for Monitoring of Ecosystems
drainage basins, watersheds, and surface and groundwa-
Monitoring programs designed to detect environ-
ter flow) are the two most important factors to use
mental problems will need to be scale-specific. This is
in delineating ecosystem boundaries. General climate
because many ecosystem properties are scale-depen-
trends and wind patterns are essentially permanent, as
dent. In moving vertically through a nested hierarchy
are the landscape features that delimit watersheds. An-
of ecosystems from a megaecosystem down to local eco-
other factor to be used to further delineate ecosystem
systems, changes occur in ecosystem properties such as
boundaries is d o m i n a n t land use (as it relates to the
size, process rates, permanence, stability of boundaries,
practicalities of regulation--urban, agricultural, indus-
and rate of change in condition (Figure 3). Ocean beach
trial, forest, grassland, wetland, etc.). Although artificial,
ecosystems can be used to illustrate how the hierarchy
political boundaries (as they relate to jurisdictional au-
of ecosystem processes and properties are related to
thority) may also need to be considered when delineat-
ecosystem size. For example, some of the most ephem-
ing ecosystems. Nevertheless, ecosystems should primar-
eral ecosystems are tidal pools, which can disappear and
ily be delineated so that their boundaries are the true
reappear within a day. However, changes in coastal sand
borders of the ecological processes of interest.
dune size, shape, and location may occur over a period
Major ecosystem processes are climate-driven, gov-
of years or decades, while wide-scale changes might only
erned by broad regimes of temperature and precipita-
be detectable over centuries or perhaps millennia. An-
tion. Within a climatic regime, landform exerts the main
other example of a scale-dependent property is ground-
influences over mesoclimate and ecosystem processes
level ozone concentrations. In urban areas, the ozone
(Rowe and Sheard 1981, Barnes and others 1982, Bailey
concentration may fluctuate more rapidly on a local
1985, 1987, Albert and others 1986, Swanson and others
level than on a regional level.
1988). Landforms bind ecosystems both structurally and
O'Neill and others (1986) suggest that higher levels
functionally. Landform influences water flow, moisture
in the hierarchy reflect "only the averaged and inte-
availability, local wind patterns, and the reception and
601
Formulating an Ecosystem Approach
Stability of Changes in
Ecosystem Size Process Permanence
boundaries condition
hierarchy rates
Slower
Higher Larger Longer More
Slower
level
Faste r
Lower Smaller Sho~er Less
Faster
level
Figure 3. Relationship between ecosystem hierarchy levels and ecosystem properties.
grated responses of the components." The effects of Implications for Ranking Ecosystems at Risk
local heterogeneity are averaged out at the broader Just as monitoring of ecological problems must be
scales of higher levels in the ecosystem hierarchy (Wiens sensitive to scale, so must the determining of threats to
1989, King 1993). Thus, there may be significant disrup- ecosystems and the ranking of ecosystems for agency
tion of a c o m p o n e n t ecosystem at a lower level in the response. As ecosystems at risk are c o m p a r e d for the
hierarchy that does not perceptibly affect the higher- purpose of setting priorities for action by EPA, it is
level ecosystem. Yet, the smoothing of fine-scale variabil- important that comparisons only be made a m o n g eco-
ity at broad scales is useful because it removes some of systems at the same level in the ecosystem hierarchy.
the noise from observations, making it possible to detect For example, when determining which ecosystems are
broad-scale trends (e.g., rise in atmospheric CO2 and a high priority for action by EPA, small regional ecosys-
other greenhouse gases). Nevertheless, this smoothing tems at risk would not be c o m p a r e d with large regional
may mask fine-scale signals that may be indicative of ecosystems at risk, since the scale of the problems (and
the corrective actions required) would differ signifi-
emerging environmental problems (King 1993). There-
cantly. Thus, it is important for ecosystems to be prop-
fore, one may need to use a "zoom-lens" approach,
erly delineated so that ecological threats can be cor-
moving through the nested hierarchy and back again
rectly assessed.
in order to more clearly see at which scale monitoring
is appropriate.
Categorizing and Ranking
O'Neill and others (1986) further suggest that an
Threats to Ecosystems
ecosystem cannot be defined arbitrarily in space and
time, but must be "defined relative to the scale of the
Threats to ecosystems vary in type, severity, extent,
problem being addressed." This further emphasizes the
and imminence. As EPA embraces the goal of protecting
importance of choosing the proper level within a nested
entire ecosystems, it will need to rank ecosystems at risk
hierarchy of ecosystems when monitoring environmen-
in order to set priorities for agency action. At present,
tal problems. According to hierarchy theory, each level
ecological risk assessment primarily involves estimating
of an ecosystem hierarchy operates at a relatively distinct
risks to indicator organisms from exposure to certain
temporal and spatial scale (O'Neill and others 1989).
chemical agents introduced into the ecosystem (EPA
The most rapid response to environmental changes can
1992, Surer 1993). However, ecosystems contain a multi-
be f o u n d in the lower levels of the ecosystem hierarchy tude of species exposed to multiple chemical agents
(Klijn and Udo de Haes 1994). The response of a nested over various periods of time. The problem of assessing
hierarchy of ecosystems to a certain stress may be signifi- ecological risk is further complicated by differences in
cant at a lower level in the hierarchy, but appear only ecosystem size and n u m b e r of organisms, often varying
as a minor one at a higher level (Overton 1977 as cited over a few orders of magnitude. Moreover, threats to
in O'Neill and others 1986). Thus, if the scale of moni- ecosystems are not limited to point discharges of pollut-
toring is inappropriate, a significant response could be ants but include other activities, such as alterations to
missed (Overton 1977 as cited in O'Neill and others the physical structure of the ecosystem, which may de-
1986). grade ecosystem quality.
602 o.J. Gonzalez
As an early step towards formulating a strategy for Level of Threat to Ecosystems
ecosystem protection, I propose a preliminary ranking I propose the following four levels of threat to eco-
system for ecosystems at risk. By using ecosystem ranking systems:
solely to prioritize agency response, a fairly qualitative
ranking system can be employed. This system would
Class l--without intervention, the ecosystem's status will
allow greater flexibility in the use of EPA scientific exper-
be largely unchanged five years from now.
tise (as well as that of partner agencies or organizations)
Class 2--without intervention, the ecosystem's status will
to make recommendations regarding important threats
have declined somewhat five years from now.
to ecosystems.
Class 3--without intervention, the ecosystem's status will
The proposed ranking system for ecosystems-at-risk
have dramatically declined, perhaps resulting in eco-
has three main parts:
system disappearance five years from now.
Class 4 collapse or disappearance of the ecosystem is
(1) category of threat, imminent (less than two years).
(2) level or class of threat, and
(3) distance from desired future condition (i.e., dis- EPA scientists could work with scientists from partner
tance from the goal). agencies and organizations. A high degree of inter-
agency cooperation at various scales will be required
for an ecosystem approach to be workable and successful
A description of each of these parts, and the way in
(MacKenzie 1993, Grumbine 1994). Together (within
which they may be combined conceptually in devel-
states and EPA regions), EPA and partner agency or
oping response strategies, is discussed below.
organization scientific staffs could review relevant data
and information to determine the appropriate category
Category of Threat to Ecosystems
and level of threat to an ecosystem. Five years is a com-
I suggest the following three broad categories of
monly used planning horizon in many institutions and
threat:
agencies. It is a reasonable period for attempting to
estimate future conditions of an ecosystem following
certain actions. It also would probably be more difficult
(1) Ecosystem degradation--occurs mainly through
pollution, but could also be from selective removal to achieve a consensus opinion if longer planning hori-
zons were used.
of species (e.g., overfishing, overhunting, etc.);
(2) Ecosystem alteration--major physical changes
Distance from Desired Future Condition
(such as dredging, water diversion) and major re-
moval of species (i.e., extinction); and Beyond achieving regulation compliance, EPA, work-
(3) Ecosystem removal--highest level of alteration ing with other agencies and stakeholders, may set goals
(e.g., destruction of wetlands due to urbanization, or form a consensus for the desired future condition
conversion of forest to cropland, etc.) of an ecosystem. The desired future condition or goal
for an ecosystem in an industrial area may differ from
one near a recreation or wilderness area. Both scientists
Different types of threats will require different types
and stakeholders would qualitatively determine how
of response from EPA. For example, in many cases,
close to or far from (i.e., "distance") the desired future
ecosystem degradation, the threat most within EPA's
condition an ecosystem was.
traditional authority, might require more of a regulatory
The "distance" from desired future condition scale
response. Alternatively, responses to other threats might
is simple, consisting of four distances:
require more interagency policy leadership or facilita-
tion among many stakeholders.
(1) close,
Furthermore, as EPA's five-year strategic plan (EPA
(2) moderate,
1994) clearly sets forth, responses to environmental
(3) far,
threats cannot be solely regulatory. Response at the
(4) very far.
ecosystem level provides an opportunity to work with
stakeholders of all types (e.g., corporate environmental
education initiatives, ecosystem-wide pollution preven-
Three-Dimensional Ranking
tion programs, initiatives to reduce chemicals in agricul-
Category of threat, level or class of threat, and dis-
tural runoff) in a particular place towards achieving
tance from desired future condition comprise the three
improved environmental quality.
603
Formulating an Ecosystem Approach
L iL
"Distance" f r o m desired condition
V e r y Far - - "Distance" f r o m desired condition
V e r y Far - -
• Ecosystem # 2
Far - Far Ecosystem # 2
/
/
Moderate - Moderate - /
/
Ecosystem#1 Ecosystem #1
Close - - Close - -
/
% I i_ I I=
///
'%y
o~ ~" ~ Category o f t h r e a t Category o f t h r e a t
Level o f t h r e a t of threat
Level
Figure 5. Three dimensional ranking of ecosystems at risk
Figure 4. Three dimensional ranking of ecosystems at risk
(ecosystem 1). Note: angles are modified for ease of illus- (ecosystem 2) Note: angles are modified for ease of illustration.
tration.
4 level of threat would now be further away from the
dimensions with which a r a n k for an ecosystem at risk
origin, thereby increasing its priority rank.
may be derived. Graphically, each of these dimensions
is an axis, a n d a rank is an x - y - z c o o r d i n a t e of the three
Using the Three-Dimensional Ranking System
axes (Figures 4 a n d 5). Thus, the priority ranking o f an
ecosystem is a function of the category a n d class of T h e r e are three basic types o f r a n k i n g methods: ne-
threat, as well as the distance from the desired future gotiated consensus, voting, a n d formulas (EPA 1993).
condition. An ecosystem is r e p r e s e n t e d as a p o i n t in T h e three-dimensional r a n k i n g system p r o p o s e d in this
t h r e e - d i m e n s i o n a l space, a c o o r d i n a t e of all three axes. p a p e r uses n e g o t i a t e d consensus along with a simple
T h e f u r t h e r away a p o i n t is from the origin o f the three additive formula. N e g o t i a t e d consensus would be used
axes, the m o r e that ecosystem requires p r o t e c t i o n rela- to d e t e r m i n e where an ecosystem should fall in each
tive to o t h e r r a n k e d ecosystems. axis. T h e distance covered on each axis could be quanti-
F o r example, in Figure 4 the two dots r e p r e s e n t two fied with a n u m e r i c a l score (e.g., o n the category o f
different ecosystems. Ecosystem 1 is close to its desired threat axis, removal would get a h i g h e r n u m b e r than
future condition; however, its d i s a p p e a r a n c e d u e to eco- alteration). Thus, a value would be p l a c e d on each o f
system removal (category o f threat) is i m m i n e n t (class the categories, classes, a n d distances o f their respective
4 level o f threat). In Figure 5, ecosystem 2 is shown to axes. O n c e the values for an ecosystem have b e e n a d d e d
be t h r e a t e n e d by d e g r a d a t i o n a n d far from its desired together, the sum can be c o m p a r e d to o t h e r r a t e d eco-
future condition. However, without intervention, the systems a n d be priority ranked.
ecosystem will have d e c l i n e d within five years (class 2 Decisions on two o f the three dimensions, category
level of threat), b u t does n o t face i m m i n e n t disappear- o f threat a n d level o f threat, could be m a d e by a scien-
ance as in the case of ecosystem 1. In c o m p a r i n g the tific p a n e l c o m p o s e d o f representatives from EPA, con-
two ecosystems, ecosystem 1, which faces i m m i n e n t re- servation organizations, s t a k e h o l d e r groups, a n d o t h e r
moval, is further away from the origin of the axes than a p p r o p r i a t e agencies (federal, state, o r local). M t h o u g h
ecosystem 2. Thus, ecosystem 1 is a h i g h e r priority for category o f threat a n d level o f threat are mainly qualita-
action than ecosystem 2. T h e further away a d o t is from tive d e t e r m i n a t i o n s by the panel, they would be based
the origin o f the t h r e e axes, the h i g h e r priority the on the p a n e l ' s review o f quantitative information. These
ecosystem it represents is for intervention. d e t e r m i n a t i o n s would r e p r e s e n t a n e g o t i a t e d consensus
In d e v e l o p i n g this r a n k i n g system, the distances of of e x p e r t j u d g m e n t . T h e p a n e l should emphasize the
tic marks o n the axes can be m o d i f i e d to increase their use o f site-specific data. These data are often f o u n d
relative weights in the rank. F o r example, if it was de- in studies c o n d u c t e d by local universities, conservation
c i d e d that a class 4 level o f threat should be a c c o r d e d organizations, a n d state a n d county agencies. However,
m o r e i m p o r t a n c e , the class 4 tic can be m o v e d f u r t h e r such i n f o r m a t i o n is often n o t available on a national
o u t on the level of t h r e a t axis. An ecosystem with a class basis. Thus, a place-driven ecosystem a p p r o a c h seeks
604 o.J. Gonzalez
and uses as much information as possible linked to the about how ecosystems are defined, and how problems
ecosystem (i.e., location) of interest. and solutions are framed. In summary:
Judgment regarding distance from desired future
condition, the third dimension, should be the purview • Ecosystems are places, large and small, nested in a
of an expanded panel with heavy stakeholder involve- spatial, temporal, and functional hierarchy.
ment. To reach consensus, the desires of the community • Ecosystem delineations must be scientifically defen-
regarding the standard an ecosystem acheives or main- sible and administratively practical.
tains must be recognized. The distance (i.e., time and • Boundaries for ecosystems are climatic factors and
effort required) from that desired state, however, is landscape features.
more a scientific question and would probably be han- • Ecosystem delineations should emphasize func-
dled best by the scientific panel. tionality.
Each of the three axes, category of threat, level of • Ecosystem scale has implications for monitoring
threat, and distance from desired future condition, is methods.
a continuum. The farther out a point is on the level of • Category of threat, level of threat, and "distance"
threat axis, the higher the threat class. Similarly, the from desired future condition can be combined to
farther out on the distance from the desired future rank ecosystems at risk.
condition axis, the greater the effort and time needed • Ranks should be based on a review of quantitative
for ecosystem recovery. Category of threat can be consid- information by a scientific panel with stakeholder
ered a continuum of reversibility, with ecosystem re- participation.
moval being the least reversible effect. Conceptually, • Ranks are determined using negotiated consensus
the further away from the origin the x-y-z coordinate and summing values from the three ranking di-
point is, the higher priority that ecosystem is for mensions.
agency response. • Ranks can be used to plan and prioritize EPA action
for ecosystems at risk.
Comparison with Other Ranking Systems
Acknowledgments
Other ranking systemsfor ecological risk consider a
variety of threats (for examples, see EPA 1990b, TNC This paper was written while I was a 1994 American
1994, EPA Region III no date). Some common disadvan- Association for the Advancement of Science (AAAS)
tages of a number of other ranking systems is that they: Environmental Science and Engineering Fellow in the
(1) may not be able to separate problems occurring at US Environmental Protection Agency's Office of Policy,
different scales, (2) are not place-specific, (3) do not Planning and Evaluation, Environmental Results
consider consequences of action or inaction within a Branch. I thank AAAS for its generous support. I also
certain time period (e.g., without intervention ecosys- thank Stephen Nelson, Claudia Sturges, Celia McEna-
tem will have declined five years from now), and (4) hey, Kristin Raab, and Chris McPhaul at AAAS, and
do not ascertain and incorporate a desired future condi- Karen Morehouse at EPA, for their fine organization
tion. An advantage of the proposed three-dimensional and stewardship of the fellowship program. At EPA
ranking system is that it does include these points. How- headquarters, Kim Devonald provided many useful in-
ever, it does not explicitly consider the rarity of an sights as my fellowship mentor. Also at EPA Headquar-
ecosystem or its resilience, which are included in some ters, I thank Wayne Davis, Sidney Draggan, Bill Painter,
other ecological ranking schemes. Peter Truitt, Nathan Wilkes, Margaret Saxton, and oth-
The proposed ranking system provides a framework ers for their suggestions and comments during the de-
for involving EPA, partner agencies, and stakeholders velopment of this paper. A special thanks to Tom De-
in determining threats to ecosystems at risk and de- Moss, Greene Jones, and Randy Pomponio, of EPA
termining priorities for agency response. The details of Region III, for their strong interest in the ideas ex-
both ecosystem delineation and the three-dimensional pressed herein. ! also thank Professor Burton V. Barnes,
ranking system must be developed and refined through University of Michigan, for his clarity and enthusiasm
testing in the field. in presenting the landscape ecosystem concept. This
paper benefited from the very constuctive comments of
R. G. Bailey, D. L. DeAngelis, and J. S. Rowe, whose
Conclusions
efforts I appreciate. The views expressed herein are
entirely the author's and do not represent official policy
An ecosystem approach to environmental protection
by the EPA or other agencies will require new thinking of either the EPA or AAAS.
605
Formulating an Ecosystem Approach
19-46 in S. Woodley,J. Kay, and G. Francis (eds.), Ecological
Literature Cited
integrity and the management of ecosystems. St. Lucie Press,
220 pp.
Albert D. A., S. R. Denton, and B. V. Barnes. 1986. Regional
landscape ecosystems of Michigan. School of Natural Re- Kiijn, F., and H. A. Udo de Haes. 1994. A hierarchical approach
sources, University of Michigan, Ann Arbor, 32 pp. to ecosystems and its implications for ecological land classifi-
cation. Landscape Ecology 9:89-104.
Allen T. F. H., and T. B. Starr. 1982. Hierarchy: Perspectives for
MacKenzie, S. 1994. Great Lakes intergovernmental coopera-
ecological complexity. University of Chicago Press, Chicago,
tion: A framework for endangered species conservation. En-
310 pp.
dangered Species Update 10 (3&4) :48-51.
Bailey, R. G. 1983. Delineation of ecosystem regions. Environ-
Odum, E. P. 1971. Fundamentals of ecology. W. B. Saunders,
mental Management 7:365-373.
Philadelphia, 574 pp.
Bailey, R. G. 1985. The factor of scale in ecosystem mapping.
Omernik, J. M. 1987. Ecoregions of the coterminous United
Environmental Management 9:271-276.
States. Annals of the Association of American Geographers 77:
Bailey, R. G. 1987. Suggested hierarchy of criteria for multiscale
118-125.
ecosystem mapping. Landscape and Urban Planning 14:
O'Neill, R. V., D. L. DeAngelis, T. F. H. Allen, andJ. B. Waide.
313-319.
1986. A hierarchical concept of ecosystems. Monographs in
Barnes, B. V., K. S. Pregitzer, T. A. Spies, and V. H. Spooner. population biology. Princeton University Press, Princeton,
1982. Ecological forest site classification. Journal of Forestry NJ, 272 pp.
80:493-498.
O'Neill, R. V., A. R.Johnson, and A. W. King. 1989. A hierarchi-
Begon, M. E.,J. L. Harper, and C. R. Townsend. 1990. Ecology: cal framework for the analysis of scale. Landscape Ecology 3:
Individuals, populations and communities. Blackwell Scien- 193-205,
tific Publications, Boston, 945 pp.
Platt,J. 1969. Theorems on boundaries in hierarchical systems.
EPA (US EnvironmentalProtection Agency). 1987. Unfinished Pages 201-214 in L. L. Whyte, A. G. Wilson, and D. Wilson
business: A comparative assessment of environmental prob- (eds.), Hierarchical structures. American Elsevier, New
lems. 230/2-87-025A. February 1987. Office of Policy, Plan- York, 322 pp.
ning and Evaluation. Washington, DC, 100 pp.
Ricklefs, R. E. 1983. The economy of nature. Chiron Press,
EPA (US EnvironmentalProtection Agency). 1990a. Reducing New York, 510 pp.
risk: Setting priorities and strategies for environmental pro- Rowe,J. S. 1961. The level-of-integrationconcept and ecology.
tection. SAB-EC-90-021. September 1990. Science Advisory Ecology 42:420-427.
Board, Washington, DC.
Rowe, J. S., and B. V. Barnes 1994. Geo-ecosystems and bio-
EPA (US Environmental Protection Agency). 1990b. The re- ecosystems. Bulletin of the Ecological Society ofAmerica 75:36-38.
port of the ecology and welfare subcommittee; reducing
Rowe, J. S., andJ. W. Sheard. 1981. Ecological land classifica-
risk, appendix A. EPA SAB-EC-90-021A. September 1990.
tion: A survey approach. Environmental Management 5:
Science Advisory Board. Washington, DC, 77 pp.
451-464.
EPA (US EnvironmentalProtection Agency). 1992. Framework Simon, H. A. 1962. The architecture of complexity. Proceedings
for ecological risk assessment. EPA/630/R-92/001. February of the American Philosophical Society 106:467-482.
1992. Risk assessment forum. Washington, DC, 41 pp.
Spurt, S. H., and B.V. Barnes, 1980. Forest ecology. John
EPA (US Environmental Protection Agency). 1993. A guide- Wiley & Sons, New York, 687 pp.
book to comparing risks and setting environmental priori-
Surer, G. W. 1993. Ecological risk assessment. Lewis Publishers,
ties. EPA 230-B-93-003. September 1993. Office of Policy,
Chelsea, Michigan, 538 pp.
Planning and Evaluation, Washington, DC, 199 pp.
Suter, G. W., and S. Bartell. 1993. Ecosystem-level effects. Pages
EPA (US Environmental Protection Agency). 1994. The new 275-308 in G. W. Suter (ed.), Ecological risk assessment.
generation of environmentalprotection: EPA's five year stra- Lewis Publishers, Chelsea, Michigan, 538 pp.
tegic plan. EPA 200-B-94-002.July 1994. Office of the Admin-
Swanson, F.J., T. K. Kratz, N. Caine, and R. G. Woodmansee.
istrator, Washington, DC, 167 pp.
1988. Landform effects on ecosystem patterns and processes.
EPA (US Environmental Protection Agency). Region III. No BioScience 38:92-98.
date. Comparative risk project: A risk-based assessment of
TNC (The Nature Conservancy) 1994. The conservation of
environmentalproblems. Region III, US EnvironmentalPro-
biological diversity in the Great Lakes ecosystem: issues and
tection Agency, Philadelphia, 46 pp.
opportunities. The Nature Conservancy Great Lakes Pro-
Grumbine, R. E. 1994. What is ecosystem management? Conser- gram. The Nature Conservancy, Chicago, 118 pp.
vation Biology 8:27-38. Wiens, J. A. 1989. Spatial scaling in ecology. Functional Ecol-
King, A. W. 1993. Considerations of scale and hierarchy. Pages ogy 3:385-397.
Formulating an Ecosystem Approach to
Environmental Protection
ecosystem as a volumetric unit delineated by climatic and
O T T O a. G O N Z A L E Z 1
Environmental Results Branch, Office of Policy, Planning and landscape features is suggested. Following this definition,
ecosystems are organized hierarchically, from
Evaluation,
United States Environmental Protection Agency megaecosystems, which exist on a continental scale
401 M Street, S.W. (e.g., Great Lakes), to small local ecosystems.
Washington, DC 20460, USA
Threats to ecosystems can generally be categorized as:
(1) ecosystem degradation (occurs mainly through pollution)
ABSTRACT / The U.S. Environmental Protection Agency (2) ecosystem alteration (physical changes such as water
(EPA) has embraced a new strategy of environmental diversion), and (3) ecosystem removal (e.g., conversion of
protection that is place-driven rather than program-driven. wetlands or forest to urban or agricultural lands). Level of
This new approach focuses on the protection of entire threat (i.e., how imminent), and distance from desired future
ecosystems. To develop an effective strategy of ecosystem condition are also important in evaluating threats to
protection, however, EPA will need to: (1) determine how to ecosystems. Category of threat, level of threat, and
define and delineate ecosystems and (2) categorize threats to "distance" from desired future condition can be combined
individual ecosystems and priority rank ecosystems at risk. into a three-dimensional ranking system for ecosystems at
Current definitions of ecosystem in use at EPA are risk. The purpose of the proposed ranking system is to
inadequate for meaningful use in a management or regulatory suggest a preliminary framework for agencies such as EPA
context. A landscape-based definition that describes an to prioritize responses to ecosystems at risk.
Over the past several years there has b e e n a growing a n d the framework for ranking ecosystems at risk intro-
awareness in the US E n v i r o n m e n t a l Protection Agency d u c e d h e r e i n can be used to help i m p l e m e n t a policy
(EPA) that c o m p l i a n c e with media-based (e.g., air, wa- of focusing on ecosystems, r a t h e r than distinct media,
ter, solid waste) regulations can n o t ensure p r o t e c t i o n in p r o g r a m m i n g e n v i r o n m e n t a l protection.
of entire ecosystems (EPA 1987, 1990a, 1994). Over the
past two years the EPA has e m b r a c e d a shift away from
How to Define and Delineate Ecosystems
a media-based program-driven focus for the agency to
one that is place-driven a n d ecosystem-based (EPA
"Ecosystem" is a familiar term to many p e o p l e , yet
1994). State agencies with responsibilities similar to
its m e a n i n g varies d e p e n d i n g on the user. This is true
EPA's may also wish to develop an ecosystem-based ap-
even for those working in the e n v i r o n m e n t a l arena. A
p r o a c h for e n v i r o n m e n t a l protection. To develop an
small marsh n e a r a city, a large forest stand, a g r o u p of
effective strategy o f ecosystem protection, however, the
sand d u n e s o n Lake Michigan, or the entire G r e a t Lakes
EPA a n d o t h e r agencies will n e e d to: (1) d e t e r m i n e
can all be c o n s i d e r e d ecosystems, a l t h o u g h they differ
how to define a n d delineate ecosystems, a n d (2) rank
by a few to several orders of m a g n i t u d e in size. A place-
threats to individual ecosystems at risk as a m e a n s o f
based or ecosystem a p p r o a c h to e n v i r o n m e n t a l protec-
prioritizing agency response. In this p a p e r I suggest how
tion will r e q u i r e a definition of ecosystem that is b o t h
to m e e t these two needs. T h e definition of ecosystem
scientifically defensible a n d administratively practical.
However, most s t a n d a r d ecology texts (e.g., O d u m 1971,
Ricklefs 1983, Begon a n d others 1990) p r e s e n t ecosys-
KEY WORDS: Ecosystemapproach; Ecologicalrisk assessment; Envi-
tem as a vague concept, r a t h e r than as a definable,
ronmental protection; EPA
measurable construct on the g r o u n d . A c u r r e n t defini-
tion of ecosystem in use at EPA is: "a dynamic c o m p l e x
~Current address: US Agency for International Development, Center
for Environment, Room 509, SA-18,Washington, DC 20523-1812, USA. of plant, animal, a n d micro-organism c o m m u n i t i e s a n d
Environmental Management Vol. 20, No. 5, pp. 597-605 © 1996 Springer-Verlag New York Inc.
598 o.J. Gonzalez
their non-living e n v i r o n m e n t interacting in a functional ecosystem as a '~¢olume o f l a n d a n d air" on the earth's
unit" (EPA 1994). This definition is typical o f those surface. Rowe a n d S h e a r d (1981) d e s c r i b e d the land-
f o u n d in standard ecology text books. However, such scape as a hierarchy of ecosystems, large a n d small,
definitions are inadequate, imprecise, a n d n o t workable nested within o n e another. Acknowledging the contri-
in a m a n a g e m e n t or regulatory context. To derive a butions o f these authors, I p r o p o s e the following defini-
better definition, we must first consider an ecosystem tion o f ecosystem:
as a place.
A volume of land, air, and water with natural boundaries, delineated
primarily by landscape features and climatic factors. It encompasses
Concept of an Ecosystem as a Place:
a set of natural ecological processes, organisms, and anthropogenic
A Landscape-Centered View
processes that function within a nested hierarchy of volumes.
W h e t h e r o n e considers an ecosystem a place depends
on o n e ' s general c o n c e p t o f ecosystem. Is the ecosystem T h e advantages of this definition over most others com-
conceived as a function o f the organism o r a function monly used is that it is: (1) functional within a spatial a n d
of the environment? For example, to a wildlife biologist, t e m p o r a l hierarchy o f ecosystems a n d (2) landscape-
a grizzly bear's ecosystem is d e f i n e d by all the land, based, thus b o u n d a r i e s can be d e l i m i t e d in the field
water, plant, a n d animal resources used by the bear. If a n d on maps with a fair d e g r e e of p e r m a n e n c e . As such,
the spatial distribution of those resources were to ecosystems are conceived as places, large a n d small,
change (e.g., high densities of prey animals shifting to nested, a n d functional within o n e a n o t h e r in a hierarchy
new locations), then so would the b o u n d a r i e s of the o f spatial sizes.
ecosystem. In this organism-centered view o f the ecosys-
Hierarchy of Ecosystems
tem, b o u n d a r i e s are drawn a r o u n d the areas used by
the organism(s) (usually animal) o f interest. This is Ecosystem is a term a p p l i e d across a wide variety o f
c o n g r u e n t with the p o p u l a t i o n - c o m m u n i t y view o f eco- spatial scales. For e x a m p l e some ecosystems may be
systems described by O ' N e i l l a n d others (1986), the 10,000 sq km or larger (e.g., Greater Yellowstone ecosys-
naturalist view described by Surer a n d Bartell (1993), tem), while others (e.g., a small patch o f forest) may
be only 1 sq km o r smaller. Functionally, as well as
a n d the bioecologist view described by Rowe a n d Barnes
spatially, ecosystems exist in a nested hierarchy (Figure
(1994). If the needs a n d habits o f the animal were to
1). Watersheds can be used as a c o n v e n i e n t illustration
change, so would the b o u n d a r i e s of the ecosystem.
o f this concept. F o r example, the Great Lakes, which
Thus, the ecosystem is conceived as a function o f the
e x t e n d into seven US states a n d one C a n a d i a n province,
animal.
Alternatively, the ecosystem can be conceived as a constitute a megaecosystem c o m p r i s e d o f m a n y smaller
ecosystems. At a lower level in scale, the Lake Erie water-
function of the environment. In this landscape-centered
shed could be c o n s i d e r e d a large regional ecosystem,
view, ecosystems are fixed places on the landscape en-
compassing physical, chemical, a n d biological resources within which is nested the Detroit River watershed, a
small regional ecosystem. Smallest is the tiny Rouge
a n d processes, along with various organisms. This con-
River, which flows t h r o u g h n e i g h b o r h o o d s n e a r the city
cept of the ecosystem as a fixed place is p r o b a b l y most
of Detroit a n d connects to the larger Detroit River. Each
familiar to geologists, hydrologists, a n d landscape ecolo-
o f these ecosystems is a place, with smaller ones nested
gists a n d is well described by Rowe a n d Barnes (1994)
as the geoecologist view. It is similar to a material cycling within larger ones, f o r m i n g a spatial hierarchy. A func-
tional hierarchy also exists because activities at a h i g h e r
perspective (Suter a n d Bartell 1993), or a process-func-
level in the hierarchy affect ecosystems at the lower
tional view of ecosystems (O'Neill a n d others 1986).
levels. Conversely, i m p r o v e m e n t o f e n v i r o n m e n t a l qual-
Foresters are also familiar with the c o n c e p t o f the forest
ity at an u p p e r level of the hierarchy is often a function
site, which identifies the ecosystem as a physical location
of success in the ecosystems comprising the lower levels.
(Spurr a n d Barnes 1980). T h e l a n d s c a p e - c e n t e r e d view
In a nested hierarchy of ecosystems, h i g h e r levels
o f ecosystems is also consistent with a place-driven ap-
contain a n d are c o m p o s e d of all the ecosystems at lower
p r o a c h to e n v i r o n m e n t a l p r o t e c t i o n because the ecosys-
levels (O'Neill a n d others 1986). Boundaries o f ecosys-
tem has a definite location.
tems may be b o t h structural a n d functional (Mien a n d
Workable Definition of Ecosystem Starr 1982). If the differences f o u n d between one side
of a b o u n d a r y a n d the o t h e r are significant, than the
A broad, place-driven strategy of e n v i r o n m e n t a l pro-
b o u n d a r y is true, or natural. If the differences are n o t
tection that can be i m p l e m e n t e d a n d a g r e e d u p o n by
significant, than the b o u n d a r y is artificial (Allen a n d
all relevant parties must begin with a s o u n d definition
Starr 1982) a n d may n o t define separate ecosystems.
o f ecosystem. Rowe (1961) i n t r o d u c e d the n o t i o n o f an
599
Formulating an Ecosystem Approach
I
M1 of these factors could be used in delineating eco-
( Megaecosyst
m systems.
Delineating Boundaries for Ecosystems
There are a n u m b e r of ways one might reasonably
delineate boundaries for ecosystems. The appropriate-
Large regional ness of one way over another depends on the ecological
ecosystem questions one wants to ask. For example, both the ecore-
gions of the United States delineated by Bailey (1983)
and Omernik (1987) represent divisions of the US land-
Small regional
ecosystem scape into regions, each relatively h o m o g e n e o u s inter-
nally in landform, soil, and other characteristics. Such
delineations can be useful in describing the potential
Local ecosystem
natural vegetation of an area. Another important use
Figure 1. Nested hierarchy of ecosystems. is in predicting the response of a site to m a n a g e m e n t
practices or other h u m a n impacts based on the response
of other sites in the same ecoregion (Bailey 1983, Omer-
nik 1987).
For example, a small wildlife refuge may be designated
However, while ecosystems can be delineated based
inside a larger wetland area. Yet, ecologically there may
on homogeneity, some of ]EPA's interests might be bet-
be no difference between the refuge and the rest of the
ter served using another basis of delineation. This is
wetland not designated as a refuge; thus the refuge
because in employing a place-driven approach to envi-
border would be an artificial boundary. Natural bound-
ronmental protection, a regulatory agency such as EPA
aries should be used for most delineations of ecosystems.
needs to determine how h u m a n activities affect air and
Nevertheless, sometimes artificial boundaries (e.g., po-
water and the places to which the air and water are
litical borders such as county lines) must be used to
naturally transported. To understand how a particular
b o u n d ecosystems into administratively practical units.
place is influenced by activities, one must recognize
According to hierarchy theory, hierarchical systems
the functional linkage between the condition of one
should be "nearly decomposable" (Simon 1962), mean-
location and activities in others. Air and water often
ing they can be divided into subsystems such that interac-
provide the conduit for these functional linkages. Thus,
tions within a subsystem are both more n u m e r o u s and
for EPA and similar agencies, ecosystem delineations
stronger than interactions between subsystems (Platt
will be most useful if they are based on functionality.
1969, Allen and Starr 1982). For example, a megaecosys-
U n d e r an emphasis on functionality, patterns of wa-
tern such as the Great Lakes can be readily decomposed
terflow and airflow functionally linked to an area on
to its five constituent lake watersheds (large regional
the landscape would be used to help delineate a volume
ecosystems), each of which can be decomposed to
of land, air, and water as an ecosystem.
smaller subwatersheds (small regional ecosystems). The
For example, at g r o u n d level, a large regional ecosys-
relative strength and n u m b e r of interactions can be
tem, such as a watershed, is b o u n d e d by landform. How-
used conceptually to help determine where one ecosys-
ever, high above ground, air patterns may transport
tem ends and another begins. Ecological interactions
particulates to and from areas beyond the boundary of
are both constrained and fostered by boundaries
the watershed. Similarly, below ground, an aquifer may
such as:
extend beyond the ground-level boundary of an ecosys-
tem. If activities in areas beyond the ecosystem's land-
(1) landform (e.g., hills, mountains, valleys, eskers,
surface borders affect the aquifer, then they also affect
kettles, kames, river floodplains),
the ecosystem. Functionally, these other areas are part
(2) air patterns (speed, direction, and temporal qual-
of an ecosystem where the land boundaries may be
ity of winds),
smaller. Thus, land, air, and subsurface boundaries of
(3) patterns of precipitation and temperature,
an ecosystem need not be congruent (Figure 2).
(4) Land u s e / l a n d cover (e.g., agriculture, urban, for-
Ideally, the higher-level boundaries selected for eco-
est, grassland, wetland), and
(5) chemical and physical traits (e.g., concentrations systems should be fairly p e r m a n e n t and be relevant to
of certain chemicals in air, water or soil; tempera- EPA's traditional authorities over air and water quality.
ture of stream water). The suggested scheme for delineating ecosystems is:
600 o.J. Gonzalez
f
J
f
f
/*
/
Upper air layer
J
J
f
/
Lower air layer
Land and Water
(surface)
(underground)
Groundwater
Aquifer ~__ _
7
f
Figure 2. Conceptual boundaries of an
J
\
ecosystem.
distribution of solar energy. In addition, landform is
Climate (considered at macro and local scales)
the most stable c o m p o n e n t of an ecosystem, and thus
Hydrology (watersheds, subwatersheds)
provides a basis for ecosystem delineation within a cli-
Land U s e / L a n d cover (agriculture, industry, forest,
matic regime (Rowe and Sheard 1981). Therefore, land-
wetland, etc.)
form boundaries can be useful as boundaries for pro-
Political boundaries (town lines, county lines,
cesses. At a fine scale within a watershed, land use and
etc.)
land cover will be useful for making practical delinea-
tions of ecosystems.
Climate (as it relates to wind patterns and patterns of
wet and dry deposition) and hydrology (as it relates to
Implications for Monitoring of Ecosystems
drainage basins, watersheds, and surface and groundwa-
Monitoring programs designed to detect environ-
ter flow) are the two most important factors to use
mental problems will need to be scale-specific. This is
in delineating ecosystem boundaries. General climate
because many ecosystem properties are scale-depen-
trends and wind patterns are essentially permanent, as
dent. In moving vertically through a nested hierarchy
are the landscape features that delimit watersheds. An-
of ecosystems from a megaecosystem down to local eco-
other factor to be used to further delineate ecosystem
systems, changes occur in ecosystem properties such as
boundaries is d o m i n a n t land use (as it relates to the
size, process rates, permanence, stability of boundaries,
practicalities of regulation--urban, agricultural, indus-
and rate of change in condition (Figure 3). Ocean beach
trial, forest, grassland, wetland, etc.). Although artificial,
ecosystems can be used to illustrate how the hierarchy
political boundaries (as they relate to jurisdictional au-
of ecosystem processes and properties are related to
thority) may also need to be considered when delineat-
ecosystem size. For example, some of the most ephem-
ing ecosystems. Nevertheless, ecosystems should primar-
eral ecosystems are tidal pools, which can disappear and
ily be delineated so that their boundaries are the true
reappear within a day. However, changes in coastal sand
borders of the ecological processes of interest.
dune size, shape, and location may occur over a period
Major ecosystem processes are climate-driven, gov-
of years or decades, while wide-scale changes might only
erned by broad regimes of temperature and precipita-
be detectable over centuries or perhaps millennia. An-
tion. Within a climatic regime, landform exerts the main
other example of a scale-dependent property is ground-
influences over mesoclimate and ecosystem processes
level ozone concentrations. In urban areas, the ozone
(Rowe and Sheard 1981, Barnes and others 1982, Bailey
concentration may fluctuate more rapidly on a local
1985, 1987, Albert and others 1986, Swanson and others
level than on a regional level.
1988). Landforms bind ecosystems both structurally and
O'Neill and others (1986) suggest that higher levels
functionally. Landform influences water flow, moisture
in the hierarchy reflect "only the averaged and inte-
availability, local wind patterns, and the reception and
601
Formulating an Ecosystem Approach
Stability of Changes in
Ecosystem Size Process Permanence
boundaries condition
hierarchy rates
Slower
Higher Larger Longer More
Slower
level
Faste r
Lower Smaller Sho~er Less
Faster
level
Figure 3. Relationship between ecosystem hierarchy levels and ecosystem properties.
grated responses of the components." The effects of Implications for Ranking Ecosystems at Risk
local heterogeneity are averaged out at the broader Just as monitoring of ecological problems must be
scales of higher levels in the ecosystem hierarchy (Wiens sensitive to scale, so must the determining of threats to
1989, King 1993). Thus, there may be significant disrup- ecosystems and the ranking of ecosystems for agency
tion of a c o m p o n e n t ecosystem at a lower level in the response. As ecosystems at risk are c o m p a r e d for the
hierarchy that does not perceptibly affect the higher- purpose of setting priorities for action by EPA, it is
level ecosystem. Yet, the smoothing of fine-scale variabil- important that comparisons only be made a m o n g eco-
ity at broad scales is useful because it removes some of systems at the same level in the ecosystem hierarchy.
the noise from observations, making it possible to detect For example, when determining which ecosystems are
broad-scale trends (e.g., rise in atmospheric CO2 and a high priority for action by EPA, small regional ecosys-
other greenhouse gases). Nevertheless, this smoothing tems at risk would not be c o m p a r e d with large regional
may mask fine-scale signals that may be indicative of ecosystems at risk, since the scale of the problems (and
the corrective actions required) would differ signifi-
emerging environmental problems (King 1993). There-
cantly. Thus, it is important for ecosystems to be prop-
fore, one may need to use a "zoom-lens" approach,
erly delineated so that ecological threats can be cor-
moving through the nested hierarchy and back again
rectly assessed.
in order to more clearly see at which scale monitoring
is appropriate.
Categorizing and Ranking
O'Neill and others (1986) further suggest that an
Threats to Ecosystems
ecosystem cannot be defined arbitrarily in space and
time, but must be "defined relative to the scale of the
Threats to ecosystems vary in type, severity, extent,
problem being addressed." This further emphasizes the
and imminence. As EPA embraces the goal of protecting
importance of choosing the proper level within a nested
entire ecosystems, it will need to rank ecosystems at risk
hierarchy of ecosystems when monitoring environmen-
in order to set priorities for agency action. At present,
tal problems. According to hierarchy theory, each level
ecological risk assessment primarily involves estimating
of an ecosystem hierarchy operates at a relatively distinct
risks to indicator organisms from exposure to certain
temporal and spatial scale (O'Neill and others 1989).
chemical agents introduced into the ecosystem (EPA
The most rapid response to environmental changes can
1992, Surer 1993). However, ecosystems contain a multi-
be f o u n d in the lower levels of the ecosystem hierarchy tude of species exposed to multiple chemical agents
(Klijn and Udo de Haes 1994). The response of a nested over various periods of time. The problem of assessing
hierarchy of ecosystems to a certain stress may be signifi- ecological risk is further complicated by differences in
cant at a lower level in the hierarchy, but appear only ecosystem size and n u m b e r of organisms, often varying
as a minor one at a higher level (Overton 1977 as cited over a few orders of magnitude. Moreover, threats to
in O'Neill and others 1986). Thus, if the scale of moni- ecosystems are not limited to point discharges of pollut-
toring is inappropriate, a significant response could be ants but include other activities, such as alterations to
missed (Overton 1977 as cited in O'Neill and others the physical structure of the ecosystem, which may de-
1986). grade ecosystem quality.
602 o.J. Gonzalez
As an early step towards formulating a strategy for Level of Threat to Ecosystems
ecosystem protection, I propose a preliminary ranking I propose the following four levels of threat to eco-
system for ecosystems at risk. By using ecosystem ranking systems:
solely to prioritize agency response, a fairly qualitative
ranking system can be employed. This system would
Class l--without intervention, the ecosystem's status will
allow greater flexibility in the use of EPA scientific exper-
be largely unchanged five years from now.
tise (as well as that of partner agencies or organizations)
Class 2--without intervention, the ecosystem's status will
to make recommendations regarding important threats
have declined somewhat five years from now.
to ecosystems.
Class 3--without intervention, the ecosystem's status will
The proposed ranking system for ecosystems-at-risk
have dramatically declined, perhaps resulting in eco-
has three main parts:
system disappearance five years from now.
Class 4 collapse or disappearance of the ecosystem is
(1) category of threat, imminent (less than two years).
(2) level or class of threat, and
(3) distance from desired future condition (i.e., dis- EPA scientists could work with scientists from partner
tance from the goal). agencies and organizations. A high degree of inter-
agency cooperation at various scales will be required
for an ecosystem approach to be workable and successful
A description of each of these parts, and the way in
(MacKenzie 1993, Grumbine 1994). Together (within
which they may be combined conceptually in devel-
states and EPA regions), EPA and partner agency or
oping response strategies, is discussed below.
organization scientific staffs could review relevant data
and information to determine the appropriate category
Category of Threat to Ecosystems
and level of threat to an ecosystem. Five years is a com-
I suggest the following three broad categories of
monly used planning horizon in many institutions and
threat:
agencies. It is a reasonable period for attempting to
estimate future conditions of an ecosystem following
certain actions. It also would probably be more difficult
(1) Ecosystem degradation--occurs mainly through
pollution, but could also be from selective removal to achieve a consensus opinion if longer planning hori-
zons were used.
of species (e.g., overfishing, overhunting, etc.);
(2) Ecosystem alteration--major physical changes
Distance from Desired Future Condition
(such as dredging, water diversion) and major re-
moval of species (i.e., extinction); and Beyond achieving regulation compliance, EPA, work-
(3) Ecosystem removal--highest level of alteration ing with other agencies and stakeholders, may set goals
(e.g., destruction of wetlands due to urbanization, or form a consensus for the desired future condition
conversion of forest to cropland, etc.) of an ecosystem. The desired future condition or goal
for an ecosystem in an industrial area may differ from
one near a recreation or wilderness area. Both scientists
Different types of threats will require different types
and stakeholders would qualitatively determine how
of response from EPA. For example, in many cases,
close to or far from (i.e., "distance") the desired future
ecosystem degradation, the threat most within EPA's
condition an ecosystem was.
traditional authority, might require more of a regulatory
The "distance" from desired future condition scale
response. Alternatively, responses to other threats might
is simple, consisting of four distances:
require more interagency policy leadership or facilita-
tion among many stakeholders.
(1) close,
Furthermore, as EPA's five-year strategic plan (EPA
(2) moderate,
1994) clearly sets forth, responses to environmental
(3) far,
threats cannot be solely regulatory. Response at the
(4) very far.
ecosystem level provides an opportunity to work with
stakeholders of all types (e.g., corporate environmental
education initiatives, ecosystem-wide pollution preven-
Three-Dimensional Ranking
tion programs, initiatives to reduce chemicals in agricul-
Category of threat, level or class of threat, and dis-
tural runoff) in a particular place towards achieving
tance from desired future condition comprise the three
improved environmental quality.
603
Formulating an Ecosystem Approach
L iL
"Distance" f r o m desired condition
V e r y Far - - "Distance" f r o m desired condition
V e r y Far - -
• Ecosystem # 2
Far - Far Ecosystem # 2
/
/
Moderate - Moderate - /
/
Ecosystem#1 Ecosystem #1
Close - - Close - -
/
% I i_ I I=
///
'%y
o~ ~" ~ Category o f t h r e a t Category o f t h r e a t
Level o f t h r e a t of threat
Level
Figure 5. Three dimensional ranking of ecosystems at risk
Figure 4. Three dimensional ranking of ecosystems at risk
(ecosystem 1). Note: angles are modified for ease of illus- (ecosystem 2) Note: angles are modified for ease of illustration.
tration.
4 level of threat would now be further away from the
dimensions with which a r a n k for an ecosystem at risk
origin, thereby increasing its priority rank.
may be derived. Graphically, each of these dimensions
is an axis, a n d a rank is an x - y - z c o o r d i n a t e of the three
Using the Three-Dimensional Ranking System
axes (Figures 4 a n d 5). Thus, the priority ranking o f an
ecosystem is a function of the category a n d class of T h e r e are three basic types o f r a n k i n g methods: ne-
threat, as well as the distance from the desired future gotiated consensus, voting, a n d formulas (EPA 1993).
condition. An ecosystem is r e p r e s e n t e d as a p o i n t in T h e three-dimensional r a n k i n g system p r o p o s e d in this
t h r e e - d i m e n s i o n a l space, a c o o r d i n a t e of all three axes. p a p e r uses n e g o t i a t e d consensus along with a simple
T h e f u r t h e r away a p o i n t is from the origin o f the three additive formula. N e g o t i a t e d consensus would be used
axes, the m o r e that ecosystem requires p r o t e c t i o n rela- to d e t e r m i n e where an ecosystem should fall in each
tive to o t h e r r a n k e d ecosystems. axis. T h e distance covered on each axis could be quanti-
F o r example, in Figure 4 the two dots r e p r e s e n t two fied with a n u m e r i c a l score (e.g., o n the category o f
different ecosystems. Ecosystem 1 is close to its desired threat axis, removal would get a h i g h e r n u m b e r than
future condition; however, its d i s a p p e a r a n c e d u e to eco- alteration). Thus, a value would be p l a c e d on each o f
system removal (category o f threat) is i m m i n e n t (class the categories, classes, a n d distances o f their respective
4 level o f threat). In Figure 5, ecosystem 2 is shown to axes. O n c e the values for an ecosystem have b e e n a d d e d
be t h r e a t e n e d by d e g r a d a t i o n a n d far from its desired together, the sum can be c o m p a r e d to o t h e r r a t e d eco-
future condition. However, without intervention, the systems a n d be priority ranked.
ecosystem will have d e c l i n e d within five years (class 2 Decisions on two o f the three dimensions, category
level of threat), b u t does n o t face i m m i n e n t disappear- o f threat a n d level o f threat, could be m a d e by a scien-
ance as in the case of ecosystem 1. In c o m p a r i n g the tific p a n e l c o m p o s e d o f representatives from EPA, con-
two ecosystems, ecosystem 1, which faces i m m i n e n t re- servation organizations, s t a k e h o l d e r groups, a n d o t h e r
moval, is further away from the origin of the axes than a p p r o p r i a t e agencies (federal, state, o r local). M t h o u g h
ecosystem 2. Thus, ecosystem 1 is a h i g h e r priority for category o f threat a n d level o f threat are mainly qualita-
action than ecosystem 2. T h e further away a d o t is from tive d e t e r m i n a t i o n s by the panel, they would be based
the origin o f the t h r e e axes, the h i g h e r priority the on the p a n e l ' s review o f quantitative information. These
ecosystem it represents is for intervention. d e t e r m i n a t i o n s would r e p r e s e n t a n e g o t i a t e d consensus
In d e v e l o p i n g this r a n k i n g system, the distances of of e x p e r t j u d g m e n t . T h e p a n e l should emphasize the
tic marks o n the axes can be m o d i f i e d to increase their use o f site-specific data. These data are often f o u n d
relative weights in the rank. F o r example, if it was de- in studies c o n d u c t e d by local universities, conservation
c i d e d that a class 4 level o f threat should be a c c o r d e d organizations, a n d state a n d county agencies. However,
m o r e i m p o r t a n c e , the class 4 tic can be m o v e d f u r t h e r such i n f o r m a t i o n is often n o t available on a national
o u t on the level of t h r e a t axis. An ecosystem with a class basis. Thus, a place-driven ecosystem a p p r o a c h seeks
604 o.J. Gonzalez
and uses as much information as possible linked to the about how ecosystems are defined, and how problems
ecosystem (i.e., location) of interest. and solutions are framed. In summary:
Judgment regarding distance from desired future
condition, the third dimension, should be the purview • Ecosystems are places, large and small, nested in a
of an expanded panel with heavy stakeholder involve- spatial, temporal, and functional hierarchy.
ment. To reach consensus, the desires of the community • Ecosystem delineations must be scientifically defen-
regarding the standard an ecosystem acheives or main- sible and administratively practical.
tains must be recognized. The distance (i.e., time and • Boundaries for ecosystems are climatic factors and
effort required) from that desired state, however, is landscape features.
more a scientific question and would probably be han- • Ecosystem delineations should emphasize func-
dled best by the scientific panel. tionality.
Each of the three axes, category of threat, level of • Ecosystem scale has implications for monitoring
threat, and distance from desired future condition, is methods.
a continuum. The farther out a point is on the level of • Category of threat, level of threat, and "distance"
threat axis, the higher the threat class. Similarly, the from desired future condition can be combined to
farther out on the distance from the desired future rank ecosystems at risk.
condition axis, the greater the effort and time needed • Ranks should be based on a review of quantitative
for ecosystem recovery. Category of threat can be consid- information by a scientific panel with stakeholder
ered a continuum of reversibility, with ecosystem re- participation.
moval being the least reversible effect. Conceptually, • Ranks are determined using negotiated consensus
the further away from the origin the x-y-z coordinate and summing values from the three ranking di-
point is, the higher priority that ecosystem is for mensions.
agency response. • Ranks can be used to plan and prioritize EPA action
for ecosystems at risk.
Comparison with Other Ranking Systems
Acknowledgments
Other ranking systemsfor ecological risk consider a
variety of threats (for examples, see EPA 1990b, TNC This paper was written while I was a 1994 American
1994, EPA Region III no date). Some common disadvan- Association for the Advancement of Science (AAAS)
tages of a number of other ranking systems is that they: Environmental Science and Engineering Fellow in the
(1) may not be able to separate problems occurring at US Environmental Protection Agency's Office of Policy,
different scales, (2) are not place-specific, (3) do not Planning and Evaluation, Environmental Results
consider consequences of action or inaction within a Branch. I thank AAAS for its generous support. I also
certain time period (e.g., without intervention ecosys- thank Stephen Nelson, Claudia Sturges, Celia McEna-
tem will have declined five years from now), and (4) hey, Kristin Raab, and Chris McPhaul at AAAS, and
do not ascertain and incorporate a desired future condi- Karen Morehouse at EPA, for their fine organization
tion. An advantage of the proposed three-dimensional and stewardship of the fellowship program. At EPA
ranking system is that it does include these points. How- headquarters, Kim Devonald provided many useful in-
ever, it does not explicitly consider the rarity of an sights as my fellowship mentor. Also at EPA Headquar-
ecosystem or its resilience, which are included in some ters, I thank Wayne Davis, Sidney Draggan, Bill Painter,
other ecological ranking schemes. Peter Truitt, Nathan Wilkes, Margaret Saxton, and oth-
The proposed ranking system provides a framework ers for their suggestions and comments during the de-
for involving EPA, partner agencies, and stakeholders velopment of this paper. A special thanks to Tom De-
in determining threats to ecosystems at risk and de- Moss, Greene Jones, and Randy Pomponio, of EPA
termining priorities for agency response. The details of Region III, for their strong interest in the ideas ex-
both ecosystem delineation and the three-dimensional pressed herein. ! also thank Professor Burton V. Barnes,
ranking system must be developed and refined through University of Michigan, for his clarity and enthusiasm
testing in the field. in presenting the landscape ecosystem concept. This
paper benefited from the very constuctive comments of
R. G. Bailey, D. L. DeAngelis, and J. S. Rowe, whose
Conclusions
efforts I appreciate. The views expressed herein are
entirely the author's and do not represent official policy
An ecosystem approach to environmental protection
by the EPA or other agencies will require new thinking of either the EPA or AAAS.
605
Formulating an Ecosystem Approach
19-46 in S. Woodley,J. Kay, and G. Francis (eds.), Ecological
Literature Cited
integrity and the management of ecosystems. St. Lucie Press,
220 pp.
Albert D. A., S. R. Denton, and B. V. Barnes. 1986. Regional
landscape ecosystems of Michigan. School of Natural Re- Kiijn, F., and H. A. Udo de Haes. 1994. A hierarchical approach
sources, University of Michigan, Ann Arbor, 32 pp. to ecosystems and its implications for ecological land classifi-
cation. Landscape Ecology 9:89-104.
Allen T. F. H., and T. B. Starr. 1982. Hierarchy: Perspectives for
MacKenzie, S. 1994. Great Lakes intergovernmental coopera-
ecological complexity. University of Chicago Press, Chicago,
tion: A framework for endangered species conservation. En-
310 pp.
dangered Species Update 10 (3&4) :48-51.
Bailey, R. G. 1983. Delineation of ecosystem regions. Environ-
Odum, E. P. 1971. Fundamentals of ecology. W. B. Saunders,
mental Management 7:365-373.
Philadelphia, 574 pp.
Bailey, R. G. 1985. The factor of scale in ecosystem mapping.
Omernik, J. M. 1987. Ecoregions of the coterminous United
Environmental Management 9:271-276.
States. Annals of the Association of American Geographers 77:
Bailey, R. G. 1987. Suggested hierarchy of criteria for multiscale
118-125.
ecosystem mapping. Landscape and Urban Planning 14:
O'Neill, R. V., D. L. DeAngelis, T. F. H. Allen, andJ. B. Waide.
313-319.
1986. A hierarchical concept of ecosystems. Monographs in
Barnes, B. V., K. S. Pregitzer, T. A. Spies, and V. H. Spooner. population biology. Princeton University Press, Princeton,
1982. Ecological forest site classification. Journal of Forestry NJ, 272 pp.
80:493-498.
O'Neill, R. V., A. R.Johnson, and A. W. King. 1989. A hierarchi-
Begon, M. E.,J. L. Harper, and C. R. Townsend. 1990. Ecology: cal framework for the analysis of scale. Landscape Ecology 3:
Individuals, populations and communities. Blackwell Scien- 193-205,
tific Publications, Boston, 945 pp.
Platt,J. 1969. Theorems on boundaries in hierarchical systems.
EPA (US EnvironmentalProtection Agency). 1987. Unfinished Pages 201-214 in L. L. Whyte, A. G. Wilson, and D. Wilson
business: A comparative assessment of environmental prob- (eds.), Hierarchical structures. American Elsevier, New
lems. 230/2-87-025A. February 1987. Office of Policy, Plan- York, 322 pp.
ning and Evaluation. Washington, DC, 100 pp.
Ricklefs, R. E. 1983. The economy of nature. Chiron Press,
EPA (US EnvironmentalProtection Agency). 1990a. Reducing New York, 510 pp.
risk: Setting priorities and strategies for environmental pro- Rowe,J. S. 1961. The level-of-integrationconcept and ecology.
tection. SAB-EC-90-021. September 1990. Science Advisory Ecology 42:420-427.
Board, Washington, DC.
Rowe, J. S., and B. V. Barnes 1994. Geo-ecosystems and bio-
EPA (US Environmental Protection Agency). 1990b. The re- ecosystems. Bulletin of the Ecological Society ofAmerica 75:36-38.
port of the ecology and welfare subcommittee; reducing
Rowe, J. S., andJ. W. Sheard. 1981. Ecological land classifica-
risk, appendix A. EPA SAB-EC-90-021A. September 1990.
tion: A survey approach. Environmental Management 5:
Science Advisory Board. Washington, DC, 77 pp.
451-464.
EPA (US EnvironmentalProtection Agency). 1992. Framework Simon, H. A. 1962. The architecture of complexity. Proceedings
for ecological risk assessment. EPA/630/R-92/001. February of the American Philosophical Society 106:467-482.
1992. Risk assessment forum. Washington, DC, 41 pp.
Spurt, S. H., and B.V. Barnes, 1980. Forest ecology. John
EPA (US Environmental Protection Agency). 1993. A guide- Wiley & Sons, New York, 687 pp.
book to comparing risks and setting environmental priori-
Surer, G. W. 1993. Ecological risk assessment. Lewis Publishers,
ties. EPA 230-B-93-003. September 1993. Office of Policy,
Chelsea, Michigan, 538 pp.
Planning and Evaluation, Washington, DC, 199 pp.
Suter, G. W., and S. Bartell. 1993. Ecosystem-level effects. Pages
EPA (US Environmental Protection Agency). 1994. The new 275-308 in G. W. Suter (ed.), Ecological risk assessment.
generation of environmentalprotection: EPA's five year stra- Lewis Publishers, Chelsea, Michigan, 538 pp.
tegic plan. EPA 200-B-94-002.July 1994. Office of the Admin-
Swanson, F.J., T. K. Kratz, N. Caine, and R. G. Woodmansee.
istrator, Washington, DC, 167 pp.
1988. Landform effects on ecosystem patterns and processes.
EPA (US Environmental Protection Agency). Region III. No BioScience 38:92-98.
date. Comparative risk project: A risk-based assessment of
TNC (The Nature Conservancy) 1994. The conservation of
environmentalproblems. Region III, US EnvironmentalPro-
biological diversity in the Great Lakes ecosystem: issues and
tection Agency, Philadelphia, 46 pp.
opportunities. The Nature Conservancy Great Lakes Pro-
Grumbine, R. E. 1994. What is ecosystem management? Conser- gram. The Nature Conservancy, Chicago, 118 pp.
vation Biology 8:27-38. Wiens, J. A. 1989. Spatial scaling in ecology. Functional Ecol-
King, A. W. 1993. Considerations of scale and hierarchy. Pages ogy 3:385-397.