http://tih.sagepub.com/Toxicology and Industrial Health

http://tih.sagepub.com/content/17/5-10/236The online version of this article can be found at:

 DOI: 10.1191/0748233701th121oa

2001 17: 236Toxicol Ind HealthLawrence A Kapustka, Hector Galbraith, Matthew Luxon and Joan Yocum

decision-makingUsing landscape ecology to focus ecological risk assessment and guide risk management

  

Published by:

http://www.sagepublications.com

can be found at:Toxicology and Industrial HealthAdditional services and information for    

  http://tih.sagepub.com/cgi/alertsEmail Alerts:

 

http://tih.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://tih.sagepub.com/content/17/5-10/236.refs.htmlCitations:  

What is This? 

- Jun 1, 2001Version of Record >>

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

Using landscape ecology to focus ecological risk assessment and guide

risk management decision-making

LAWRENCE A KAPUSTKA,a HECTOR GALBRAITH,b MATTHEW LUXONc AND JOAN YOCUMa

aecological planning and toxicology, Inc., 5010 SW Hout Street, Corvallis, Oregon 97333-9540, USAbGalbraith Environmental Sciences, LLC, 289 Wiswall Hill Road, Newlane, Vermont 05345, USA

cWindward Environmental, LLC, 200 West Mercer Street, Suite 401, Seattle, Washington 98119, USA

Ecological risk assessment (EcoRA) generally suffers from limited application of ecological knowledge in the definition and characterization of real-

world sites. Not surprisingly, most remediation decisions, which follow, have little or no relationship to the valued ecological resources of the site or the

broader region. The practice has evolved to favor engineering-based mitigation strategies, which eliminate excess chemical concentrations at sites, or

otherwise break exposure pathways, but which may not be ecologically beneficial. The heavy emphasis of EcoRA on toxicity threshold levels tends to

focus dollars on clean up of small areas or volumes with high concentrations. Moreover, intrusive remediation technologies often render an area

uninhabitable to the very species that were to be protected. Infusion of ecological knowledge into EcoRA has been difficult. Most professional

ecologists choose not to venture into the messy applied fields, leaving their impressive knowledge untapped. Moreover, narrowly defined responsibilities

within government circles can limit cooperation and coordination. The realization that land use activities often have greater adverse consequences to

wildlife than do chemicals provides an opportunity to change attitudes and practices. We are developing procedures that incorporate landscape features

into the environmental management process. Specifically, we are using an iterative approach to:

a) identify scenarios where habitat value is important in EcoRAs;

b) guide selection of appropriate assessment species,

i) keyed to wildlife distribution ranges;

ii) keyed to a database of habitat suitability models;

iii) cross-linked with the EPA exposure handbook species;

iv) referenced to wildlife distributions (e.g., breeding bird survey);

c) define data collection needs for reconnaissance-, screening-, and definitive-level characterization of habitat quality for potential assessment

species;

d) generate spatially explicit descriptions of habitat quality for various assessment species; and

e) allocate exposure estimates using both habitat quality and spatial variations in chemical concentration.

These refinements in the EcoRA process are expected to improve risk estimates and provide valuable information to be used in structuring risk

management options. The approach can guide the planning process so that an assessment considers the most relevant species of the area and

defines the relevant parameters to be measured. In risk characterization, these data are used to calculate more realistic exposure assessments. In

guiding remediation, the approach logically considers a wider range of land management options than are considered at most sites today. For

example, habitat enhancement can be used to draw animals away from contaminated zones. Contaminated localities that also have poor-quality

habitat may be allowed to go through a slower, less costly bioremediation process until the risk level is lowered to acceptable levels. And direct

comparisons of lost resources stemming from destructive remediation options can be assessed instead of merely focusing on the lowering of

contaminant concentrations. This paper presents the conceptual foundation for incorporating landscape ecology into the risk assessment

process. Toxicology and Industrial Health 2001: 17, 236± 246.

Key words: polygon characterization; remediation options; vegetation mapping; wildlife habitat quality

Introduction

Ecological risk assessments (EcoRAs) estimate the like-

lihood of adverse effects occurring to some valued resource

due to exposure to biological, chemical, or physical agents.

In terrestrial settings, the valued resources (assessment

species) are typically one or more wildlife populations,

which may be large mammals, raptors, songbirds, or other

easily recognized species. Most efforts have focused on

evaluating the toxicity of chemicals released to the envi-

ronment. Relatively little focus has been directed at the

exposure component of the risk equation. Even less atten-

tion has been given to biological or physical conditions.

1. Address all correspondence to: Lawrence A. Kapustka, ecological

planning and toxicology, Inc., 5010 SW Hout Street, Corvallis, OR 97333-

9540, USA

E-mail: kapustka@ep-and-t.com

Toxicology and Industrial Health 2001; 17: 236 ± 246

www.tihjournal.com

© Arnold 2001 10.1191/0748233701th121oa at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

Consequently, ecological risk assessments miss major

ecological factors that influence the status of wildlife

populations. Also, observed ecological effects often are

attributed incorrectly to contaminants.

The function of an EcoRA is to predict the potential

effects of stressors (to date, typically chemical) on

ecological resources. One of the essential components

of the EcoRA process is the identification of suitable

assessment endpoint species (AES). In selecting AES,

efforts are usually made to identify species likely to have

the greatest exposure, may be sensitive to the stressors of

concern, and are valued by stakeholders. Nevertheless,

EcoRAs are often criticized 1) for focusing on the same

few species and arbitrarily omitting too many others; 2)

for focusing on species unlikely to occur on the site

because of habitat or home range limitations; and 3) for

ignoring spatial patterns in habitat quality. The latter two

points (2 and 3) are particularly important in the expo-

sure assessment portion of an EcoRA as they largely

determine a species’ use of the site and, therefore, its

degree of exposure.

Biological factors that influence exposure, though

recognized qualitatively as being important, have not

been studied nor used effectively in characterizing risk

to wildlife posed by chemicals. Most of the assessment

species have been described reasonably well in terms

of life history patterns and habitat preferences. Natural-

ists and wildlife managers have understood, at least in

qualitative terms, the importance of critical habitat for

various stages of the life history (e.g., nesting sites,

winter range, and so forth). Animals key on physical

structure and food availability and avoid areas of lower

quality. The term habitat, though often used loosely as

an indication of environmental quality, refers to the

combination of physical and biological features for a

particular species. What is great habitat for prairie

chicken is unacceptable for barred owls. Different

habitat preferences reflect evolution and adaptation of

species separating from each other in `n-dimensional

niche space’ (Whittaker, 1975). There are differential

use rates by different species. Animals cue on partic-

ular features of the landscape for foraging, loafing,

nesting/birthing, and so forth. Some species are

attracted to disturbance zones and edges, but others

avoid such areas. As contaminants are distributed

differentially across a landscape, behavioral preferences

among wildlife species for different portions of the

landscape can result in minimal exposure (avoidance)

to excessive exposure (attraction). Thus, conventional

approaches in ecological risk assessment, which do not

factor in species use patterns, may greatly overestimate

risk for some species, and greatly underestimate risk

for others. The largest uncertainty in risk assessments

comes from the generally poor characterization of habitat

requirements and landscape level use patterns of the

assessment species.

Characterization of habitat for certain species was for-

malized by the U.S. Fish and Wildlife Service in the 1980s.

The resulting Habitat Evaluation Procedures (HEP) gene-

rated a flurry of activity, which resulted in Habitat

Suitability Index (HSI) models for many species of

interest. Guidelines for developing community-level hab-

itat evaluation models were published to standardize the

process (Schroeder and Haire, 1993). Both EcoRA and

HSI procedures can range from generalized efforts, which

yield gross qualitative answers, to detailed treatments

having considerable quantitative rigor. The challenge is

to meld these two flexible processes into an integrated,

seamless method that improves the utility of both pro-

cesses for modern environmental management.

Exposure to hazardous chemicals at contaminated sites

is affected by many biological and physical factors.

Traditionally, external factors that influence bioavailabil-

ity and internal factors that alter dose (e.g., assimilation,

metabolism, depuration) are given careful consideration

in estimating effective exposure levels. Important eco-

logical factors have received much less attention. To a

limited extent, potential temporal residence or home

range is used to adjust exposure estimates. Simplistically,

if home range is much larger than the contaminated area,

exposure might be adjusted substantially downward.

However, if a contaminated area attracts wildlife, species

with larger home ranges may be at greater risk than

species with smaller home ranges. Current practices in

exposure assessment have not formally addressed issues

of habitat quality.

Applications of the HSI models include environmental

impact assessment, where determination of the amount of

wildlife habitat lost or created by a particular set of

actions requires prediction and post hoc documentation.

Rand and Newman (1998) described the applicability of

HSI models for ecological risk assessment in general

terms, but provided no examples of its use and did not

give specific details to integrate habitat information with

exposure assessment or risk characterization. Freshman

and Menzie (1996) described two approaches to take into

account spatial differences in contaminant concentrations

with respect to foraging activities and the proportion of a

local population likely to be exposed to the contaminants.

Their approach does not incorporate HSI models formally,

but does demonstrate the utility of doing so. Hope (2000)

developed a probabilistic model to generate spatially

explicit population exposure estimates and showed that

habitat quality can play a major role in levels of exposure

to local populations (Hope, 2002). Similarly, Linkov et al.

(2002) and Linkov (2002) demonstrated the importance of

spatial and temporal patterns of exposure that are affected

by habitat conditions.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

237

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

The most significant application of HSI technology is

the implementation of Section 404 of the Clean Water Act,

which mandates `no net loss’ of wetlands. In addition to

defining acreage as `wetland’ according to strict rules de-

veloped by the Army Corps of Engineers, an assessment

of the number of wildlife habitat units lost or gained also

is required. This aids in mitigation banking where the

preservation of habitat units in one area may be traded for

the loss of equal numbers of habitat units in another.

Wildlife habitat improvement projects also rely on HSI

models to define habitat attributes and document that

improvement has occurred. Habitat features have linked

geo-referenced databases of physical and biological fea-

tures into spatially displayed ranges of habitat quality for

various species. In Washington, for example, statewide

maps of habitat quality (both theoretical and verified

through ground observations) have been produced for

256 bird species. Though the scale of such maps would

not be suitable for most site assessments, they should aid

regional risk assessments. And, the fundamental approach

should be applicable for smaller scale geo-referenced dis-

plays. Despite the widespread use of HSI models in land

management applications, there is little testing and doc-

umentation of their applications in the peer-reviewed

scientific literature or in other areas of environmental

management.

The focus of our work has been to develop procedures

to integrate habitat quality as a major component EcoRA.

Particular emphasis is directed at the exposure assessment

portion. We have outlined a process to:

f ) identify scenarios where habitat value is important in

EcoRAs;

g) guide selection of appropriate assessment species,

keyed to wildlife distribution ranges;

h) define data collection needs for reconnaissance-,

screening-, and definitive-level characterization of

habitat quality for potential assessment species;

i) generate spatially explicit descriptions of habitat

quality for various assessment species; and

j) allocate exposure estimates using both habitat quality

and spatial variations in chemical concentration.

Approach

Our effort is structured around two processes in modern

environmental management that until now have developed

separately, namely EcoRAs and HEPs. Ecological risk

assessment (U.S. EPA, 1992) has been used widely to

gauge the likelihood of adverse consequences resulting

from chemical exposures. The Superfund program has

developed guidelines for performing scoping, screening,

and definitive stages of EcoRA for use at contaminated

sites (U.S. EPA, 1998). Habitat evaluation procedures (HSI

models) have been developed for wildlife management

activities (Schroeder and Haire, 1993).

We have located 62 HSI models for bird species, 17 for

mammals, and 6 for reptiles/amphibians.1 Information from

these publications has been encoded into an ACCESS data-

base. Database fields include species distribution by EPA

region, state, and specific area for which the model was

produced (e.g., the range of Black Bear Model is the Great

Lakes, whereas the species has a much larger distribution);

parameters required to compute the HSI; and prioritized

methods that can be used to obtain data to parameterize the

models.

Entries on all North American terrestrial and wetland

wildlife species (avian, herpetofauna, or mammalian) are

included. Notations are provided as to similarity of habitat

requirements to HSI model species and to dietary pre-

ferences for U.S. EPA exposure species [i.e., those listed

in the U.S. EPA Wildlife Exposure Factors Handbook

(U.S. EPA, 1993)]. Built-in queries permit searches on

any species or list of species to generate a compiled report

of all potential species in a project area and the level of

overlapping information for each taxon in terms of habitat

and dietary preferences. For the bird HSI model species, we

have identified an additional 107 `overlap species’. These

are species for which no HSI models exist but for which, be-

cause of close similarities in their habitat requirements,

existing individual HSI models may be appropriate (per-

haps after modification). In total, published HSI models

exist for 169 primary and overlap bird species.

Scenarios

The distribution and abundance of organisms varies across

landscapes. Indeed, the degree of heterogeneity of vegeta-

tion and physical features generally relates to species

diversity. Habitat for a given species is dictated by partic-

ular landscape features. The quality of habitat for a given

species is judged according to the occurrence of various

landscape feature requirements or preferences. Some wild-

life species thrive at the boundaries of discontinuities (i.e.,

edges, ecotones). Generalized home ranges for particular

species may expand or contract in relation to patchiness.

Certainly, the time spent by an organism in different

portions of a landscape reflects preferences for particular

landscape features.

Contaminant concentrations may appear in patches (`hot

spots’) or gradually grade across an area. Both the source of

contaminant and dispersion forces (e.g., wind, water flow)

affect the spatial distribution of chemicals at a site.

1The list of HSI models currently in the ACCESS database may be

viewed at www.ep-and-t.com; most of the published models may be down-

loaded from the USGS website at http://www.nwrc.gov.wbd/pub/hsi/

hsiindex.htm

Landscape ecology and ecological risk assessment

LA Kapustka et al.

238

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

The size of a species home range and the distributions of

suitable habitat patches across a landscape as well as

contaminant concentrations can be important determinants

of exposure. An organism that inhabits a home range that is

much larger than the contaminated area is likely to have a

lower exposure than one whose home range is entirely

circumscribed by the area of contamination. That is unless

the site provides habitat that is of better quality than is

available in the rest of the wide-ranging organism’s home

range, in which case the site may function as an attractive

nuisance and result in higher exposures. Conversely, if the

contaminated area provides little good-quality habitat com-

pared to adjacent areas, for a wide-ranging individual, it is

unlikely to spend much time there, and its exposure may

be low.

If the contaminated site is larger than the individual’s

home range and more than one individual may inhabit it,

the exposure of individuals may be different depending

on the spatial distributions of contaminant concentrations

and habitat quality. If, for example, the contaminants

occur in `hotspots’, the individuals inhabiting those par-

ticular areas may have higher exposure than individuals

may have elsewhere on the site. Moreover, if the hotspots

coincide with areas of relatively high quality habitat, a

large proportion of the site’s population may be highly

exposed. If, on the other hand, the `hotspots’ coincide

with areas of low-quality habitat, only a small segment of

the population may be highly exposed. Thus, the spatial

distributions of contaminants and habitat may interact

and the degree of exposure to an individual or a pop-

ulation may not be accurately described by the customary

statistical metrics of exposure (e.g., mean site contami-

nation level, upper 95% percentile).

The interplay between contaminant distribution and

wildlife use patterns can be viewed in different scenarios.

The utility of habitat evaluation as a modifier of expo-

sure estimates varies among different combinations of

these attributes. We have recognized 12 contingencies

based on three spatial categories relating home range to

site area and four patterns of habitat heterogeneity for

contaminant distribution heterogeneity (Table 1).

Cases where habitat strongly influences exposure estimates

Heterogeneous landscapes coupled with heterogeneousdistribution of contaminants introduce great uncertaintyin exposure estimates for any species (see cases no. 2, 6,and 10 of Table 1). In such situations, the relative sizeof the site to the home range of the species does notmatter.

Two other cases occur in which habitat modificationsof exposure estimates would reduce uncertainty. One iscase 7 (Table 1), in which contaminant distribution is het-erogeneous and home range is small relative to the area ofthe site. The other is case no. 9 (Table 1), in which habitat

is heterogeneous and the home range is very large relativeto the contaminated area.

Cases where habitat would not influence exposure

estimates The combination of homogeneous habitat andhomogeneous contaminant distribution precludes usinghabitat conditions as a modifier of exposure. This occursin cases no. 4, 8, and 12 (Table 1) regardless of the homerange to site area relationship. Homogeneous contaminantdistribution also makes habitat conditions moot for specieswith home ranges equal to or less than the site (cases no. 1and 5, Table 1). Finally, with homogeneous habitat con-ditions, exposure estimates for species having home rangesequal to or larger than the contaminated area (cases no. 3and 11, Table 1) would not be improved.

EcoRA framework

The U.S. EPA (1998) described an overall process for the

performance of EcoRA. Specifically it provides a frame-

work for incorporating HSIs into exposure and effects

calculations. The U.S. EPA (1998) acknowledged the

importance of considering habitat when evaluating risk;

however, it proposed no specific methods for doing so.

The approach we propose is integrated with, and builds on

the three phases (problem formulation, analysis, and

characterization) of the U.S. EPA (1998) EcoRA process

(Figure 1).

Problem formulation phase The key goals of problem

formulation are to outline the potential problem and todevelop a plan for analyzing and characterizing risk. Threeinterdependent products of problem formulation are theidentification of assessment species, identification ofassessment endpoints, and construction of the conceptualmodel of the site (U.S. EPA, 1998). The EPA frameworkcan be expanded to incorporate species-specific habitatquality into the iterative steps used in problem formulation.Selection of contaminants of concern and developing ananalysis plan are not affected by the inclusion of habitatconsiderations in the process.

Landscape mapping Early in problem formulation, land-scape cover types should be identified and mapped. Also,as other features critical to particular wildlife use patternssuch as areas subject to human activities that can affectecological resources should be noted. Mapping creates apicture of the site that is critical to determining theinteraction of the stressors with the receptors, and suchrelationships are used in computing habitat suitability formany species. Even very coarse resolution landscape mapscan be informative as to whether a risk assessment is evennecessary. For example, if no suitable habitat exists atthe site (e.g., if the site is entirely paved), an EcoRA maynot be warranted. Mapping unit resolution should be

Landscape ecology and ecological risk assessment

LA Kapustka et al.

239

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

Table

1.

Contingen

cyta

ble

illu

stra

ting

rela

tionsh

ips

of

hom

era

nge

(cir

cle)

rela

tive

tosi

tesi

ze(s

quar

e)Ð

case

sw

her

eH

SI

may

be

use

fulin

reduci

ng

unce

rtai

nty

of

exposu

rees

tim

ates

and

case

sw

her

ehab

itat

consi

der

atio

ns

may

be

moot.

Spat

ial

rela

tionsh

ipH

abitat

het

erogen

eous

conta

min

atio

nhom

ogen

eous

Hab

itat

het

erogen

eous

conta

min

atio

nhet

erogen

eous

Hab

itat

hom

ogen

eous

conta

min

atio

nhet

erogen

eous

Hab

itat

hom

ogen

eous

conta

min

atio

nhom

ogen

eous

1)

Exposu

reto

org

anis

ms

isa

funct

ion

of

site

mea

nco

nta

min

atio

nle

vel

.

2)

Exposu

reto

org

anis

mis

not

afu

nct

ion

of

site

mea

nco

nta

min

atio

nle

vel

.

3)

Exposu

reto

org

anis

mis

afu

nct

ion

of

site

mea

nco

nta

min

atio

nle

vel

.

4)

Exposu

reto

org

anis

ms

isa

funct

ion

of

site

mea

nco

nta

min

atio

nle

vel

.H

SI

wei

ghting

isnot

requir

ed.

HSI

wei

ghting

isnec

essa

ry.

HS

Iw

eighting

isnot

requir

ed.

HS

Iw

eighting

isnot

requir

ed.

5)

All

indiv

idual

seq

ual

lyex

pose

d.

HSI

wei

ghting

isnot

requir

ed.

6)

All

indiv

idual

snot

equal

lyex

pose

d.

7)

All

indiv

idual

snot

equal

lyex

pose

d.

HS

Iw

eighting

requir

edto

estim

ate

8)

All

indiv

idual

seq

ual

lyex

pose

d.

HS

Iw

eighting

isnot

requir

ed.

HSI

wei

ghting

requir

edto

estim

ate

exposu

refr

equen

cies

inpopula

tion.

freq

uen

cies

of

exposu

ream

ong

popula

tion.

9)

Exposu

reto

org

anis

ms

isa

funct

ion

of

site

conta

min

atio

nan

dre

lative

hab

itat

qual

ity.

10)

Exposu

reto

org

anis

ms

isa

funct

ion

of

site

conta

min

atio

nan

dre

lative

hab

itat

qual

ity.

11)

Exposu

reto

org

anis

ms

isa

funct

ion

of

conta

min

atio

n.

HSI

wei

ghting

isnot

requir

ed.

12)

Exposu

reis

afu

nct

ion

of

conta

min

atio

n.

HSI

wei

ghting

isnot

requir

ed.

HSI

wei

ghting

nec

essa

ryto

estim

ate

exposu

refr

equen

cyto

indiv

idual

.H

SI

wei

ghting

nec

essa

ryto

estim

ate

exposu

refr

equen

cyto

indiv

idual

.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

240

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

considered carefully, as it has consequences for subsequentdecisions in the EcoRA process. Apparent homogeneity ofthe landscape, spatial extent of stressors of interest, poten-tial home range/foraging range of wildlife species, and costare important factors in defining the appropriate mappingunit resolution.

Ecological context of the site Readily available in-formation about the site and surroundings’ ecologicalcharacteristics should be compiled. This may include infor-mation on geomorphology, potential natural vegetation,fauna, climate, surface water characteristics, disturbance

regime, and land uses. There are two goals for this infor-mation. The first, in conjunction with the habitat map, isto identify a suite of species from which appropriate AESmay be selected. The second is to aid in developing thesite conceptual model by identifying the likely spatialand temporal patterns of use of the site by organisms, andidentify and evaluate potential exposure pathways. Specieslists from publications, reports, or interviews with localexperts should be compiled.

Information from the mapping and data compilationsteps may be compiled in an environmental checklist suchas that presented in Representative Sampling Guidance

Document, Volume 3: Ecological (U.S. EPA, 1997; seeAppendix B). Key questions that may be addressed by thechecklist include:

Y What is the environmental setting, including natural

areas (e.g., upland forest, on-site stream, nearby

wildlife refuge) as well as disturbed/man-made

areas (e.g., waste lagoons)?

Y What are the on- and off-site land uses (e.g.,

industrial, residential, or undeveloped; current and

future)? What type of facility existed or exists at the

site?

Y What are the suspected contaminants at the site?

Y Which habitats present on site are potentially

contaminated or otherwise disturbed?

Y Has contamination migrated from source areas and

resulted in `off-site’ impacts or the threat of

impacts in addition to on-site threats or impacts?

Y What species that are likely or known to be present

on the site might comprise suitable assessment

species?

Selection of assessment endpoint species The suite ofspecies used for the EcoRA ultimately must be assess-able; that is data must be available to calculate or toinfer exposure to and effects of stressors. Frequently,surrogate species are used to represent broadly definedgroups of potentially important species defined primarilyby trophic levels (i.e., a mammalian herbivore, an avianinsectivore, or a first-order carnivore). Ecological rele-vance of the species used in the assessment is frequentlya contentious point of debate. But monetary constraintstypically limit the opportunities to perform the assess-ment anew using more relevant species. If a broadersuite of species were considered as assessment species atthe onset, such arguments could be avoided. To do sorequires a structured approach that considers all potentialspecies at a site and documents in the administrativerecord the rationale for narrowing the list to a manage-able number.

The central premise of our approach is that the high-est quality EcoRAs are those that focus on assessment

Reconnaissance Visit -- Habitat Checklist

Wildlife Habitat Present

Determine Agents of Concern

Select Assessment Species

Compile Habitat

Parameters for

Assessment Species

Stop

Problem Formulation

Analysis

Delineate habitat areas (qualitatively by cover types, terrain, etc.)

Acquire HSI input data

Orientation Regional Ecology Context

Calculate HSIs

Estimate Exposure [i.e., wildlife exposure factors --

separately for each zone; each species]

Estimate Population (N)

by zone; species

N=Area x HSI x CC

Modify Exposure Estimates

Risk Characterization

Determine Magnitude and Extent of Affected

Populations

Uncertainty Analyses

Sensitivity Analyses

Risk Communication

Risk Management

yes

no

Figure 1. Integration of habitat considerations in the ecological riskassessment process.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

241

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

species for which both wildlife habitat requirements/preferences and exposure parameters (dietary preferences,feeding rates, metabolic rates, and so forth) are known.To achieve a high-quality EcoRA using species forwhich such information is missing requires considerablecommitment of time and money in order to obtain therequisite data. High profile sites (i.e., ones of greatinterest to the public or with large consequences) maywarrant the expenditures to gather the information. How-ever, for most sites, the collection of basic biological orecological data is beyond consideration. In such situa-tions, a transparent process could facilitate communica-tion among stakeholders and improve acceptance of therisk assessment.

We have developed a straightforward process to gen-erate a prioritized list of potential AES relevant to aparticular site (Figure 2). First, a comprehensive wildlifespecies list is compiled from prior knowledge of a site(box 1). A query of HSI models is run to produce a listof HSI models that are applicable for the geographiclocation of the site (box 2). The comprehensive list isused as input and the applicability list is used as a filterin a query of the `correspondence matrix’ table imbeddedin the HSI database. The resulting report (box 3) is a listof all species identifying taxa for which both habitat andexposure information are available (group 1), only habitatinformation (group 2), only exposure information (group3), and neither habitat nor exposure information areavailable (group 4). State or federal sensitive, threatened,or endangered species can be highlighted to ensureproper consideration is afforded these species duringthe assessment. Expert judgment may be used to expandor reduce the candidate list of species under consider-ation. And the rationale for setting priorities for thedifferent groups or for particular species can easily bedocumented in the administrative record.

Using the AES habitat requirements to devise a sampling

plan Once the candidate assessment species have been

identified, the list of HSI models to be used can be used todevise the sampling plan. The list of HSI species is usedto query the database; the resulting database report providesa compiled list of all variables needed to calculate HSImodels for all the selected species.

The majority of the variables used in terrestrial andwetland HSI models are routine measurements of vegeta-tion or other landscape-level features. These includeparameters such as percentage canopy cover, distancebetween cover types (e.g., forest edge to water; forestpatch to forest patch), or other features that can beacquired from aerial imagery. Other variables requireon-site determination, such as height of shrub canopy,percentage herbaceous cover under a forest canopy, size-class distribution of trees, and the like. Still others requiredetailed quantification of plant community structure, withvariables such as the number of nesting cavities in largetrees.

Examination of particular HSI equations also revealsdifferences in sensitivities of the variables. Qualitativesensitivity features of the models have been coded in a com-ment field in the database. For scoping- or screening-levelassessments, estimates of variables that are particularlydifficult to parameterize can provide a rapid, preliminaryindication of the importance of gathering particular data.By examining the list of variables, the preferred andalternative methods that may be used to generate therequired data, and reviewing the sensitivity of the varia-bles, one can rapidly identify those variables that can besatisfied using aerial images, routine on-site survey meth-ods, and specialized or detailed on-site survey procedures.From this, it is a relatively straightforward process todevise a progressive sampling plan from reconnaissance-level through definitive-level EcoRAs that maximizes thenumber of models satisfied with different levels of sam-pling effort. Such a plan can also be used to produce afinancial risk assessment for a project (i.e., what are thebenefits of obtaining all the information at once versusdeferring certain data collection procedures into laterstages of risk assessment).

Analysis phase

The overall objectives of this phase of the EcoRA process

are to estimate exposure and characterize potential stressor

effects to the AES. Here we describe modifications of the

EcoRA process (U.S. EPA, 1998) pertaining to habitat

characterization that can be used to modify the exposure

estimates.

There are structural differences among HSI models that

are dictated by species’ foraging preferences and ranges.

Some species key on several cover types (e.g., requiring a

mixture of forest, shrubland, grasslands, or agricultural

fields) to provide shelter and food; other species tend to

use the interior portions of individual patches or cover

Compile comprehensive, site-specific species list

1

Query Database using "HSI species geographic applicability" field

2

Species List Sorted by Levels of Overlap

3

Figure 2. Process flow diagram to guide the selection of assessmentendpoint species.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

242

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

types. Calculation of the HSI for some species can be done

on each polygon delineated over a site, whereas calcula-

tions of HSIs for other species may circumscribe several

polygons, or even the entire site. For purposes of illustra-

tion, we show a simplified system in which an HSI is

calculated for each polygon (Figure 3); in which the panel

in the upper left shows HSIs for each polygon; a stylized

chemical plume is shown as an overlay in the upper right

panel; and the delineation of habitat-chemical polygons

with labels (a1, a2, and so forth) in the lower right panel

indicate subdivisions of polygons for which localized risk

estimates would be calculated. These steps would be

repeated for each assessment species and the cumulative

risk values for each polygon would reflect the localized

levels of risk; an area-weighted site-wide risk value could

also be obtained.

Situation a Ð for species with relatively small home ranges:

estimating the numbers of animals in each habitat

subdivision The number of animals likely to use an areais a function of their social organization, their territory orhome range sizes, and the quality of the habitat. For anyspecies, the density of individuals will vary across a sitedepending on the spatial variation in habitat quality, withhigher densities (with smaller home ranges) inhabitingareas of higher habitat quality. The number of a givenspecies that is likely to inhabit any habitat subdivision canbe approximated using the area of the subdivision, the HSIscore for that subdivision, and information on either therange of the animals’ density or home range sizes, asillustrated in the following equations.

Equation (1): for use of home range data.

Ns ˆ As

HRs

…1†

Equation (2): for use of density data.

Ns ˆ As£CCs …2†

whereNs the number of individuals likely to inhabit the

subdivision

As the area of the subdivision

HRs the approximate home range size of the animals

within the subdivision

CCs the approximate carrying capacity of the subdivision

HRs in this equation is a function of the HSI score for thesubdivision and information from the literature on theapproximate range of sizes of home ranges used bythe species. We assume that animals that inhabit areas ofmedium-quality habitat (scoring about 0.5 on the HSI scale)will have home ranges and densities that approximate thecentral tendency. Conversely, areas that score closer to thetwo extremes (0 and 1) will have home ranges and densitiesthat are closer to the minimum and maximum, respectively.

Situation b Ð for species with relatively large home ranges;

estimating the proportion of time animals would spend in

each area of contamination The proportion of its timethat a wide-ranging organism is likely to spend on a sitewill be a function of the size of the site relative to its home

subunits defined by habitat x conc.Within each:�bootstrap concentration�combine with HSI�sum resulting exposure estimate

overlay

habitat unitsCoC distribution

Figure 3. Stylized delineation of polygons and hypothetical HSI values associated with each polygon.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

243

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

range requirements, the quality of the habitat on the siterelative to its surroundings, and the rate at which habitatquality may change through time.

If the contaminated site’s habitat quality is approxi-mately equal to that of the site surroundings, the proportionof its time that an animal will spend on the site can beestimated using a simple proportion calculation (if the siteis roughly 25% of the animals home range, then it willspend approximately 25% of its time there and get 25% ofits diet (or accidental ingestion) there.

If the habitat on the site is of lower or higher qualitythan the surroundings, the animal is likely to spend propor-tionally less or more of its time there. However, therelationship between habitat quality and use in a wide-ranging organism is unlikely to be linear. No matter howgood the habitat on the site is, it is unlikely that theorganism will be able to obtain all its life history require-ments there — it will be predisposed to spend some of itstime in the surroundings. Also, an efficient organism livingin an area of temporally varying habitat will probably wantto track shifts in habitat quality, and thus visit all parts of itshome range, at least intermittently.

Once again, this time allocation may be estimated usingHSI scores, except that on this occasion we require an HSIscore that represents the habitat quality in the areas that theorganism may use off-site. By comparing the size ofthe site relative to the animals’home range and the habitatquality on- and off-site or within any subdivision of thesite, we can approximate time allocation, as illustrated inEquation (3).

Ps ˆ As=HRs

Pn

sˆ1

…As=HRs†…3†

wherePs proportion of time spent foraging in subarea s

As area of subarea s

HRs home range size associated with habitat quality in

subarea s

As in Equation (1), HRs in this equation is a functionof the HSI score for the subdivision and information fromthe literature on the approximate range of sizes of homeranges used by the species. We assume that animals thatinhabit areas of medium quality habitat (scoring about 0.5on the HSI scale will have home ranges that approximatethe central tendency of home range sizes reported in theliterature. Conversely, areas that score closer to the twoextremes (0 and 1) will have home ranges that are closer tothe minimum and maximum, respectively.

Risk characterization phase

As in the analysis phase, risk characterization calculations

also differ according the relative size of home ranges of the

assessment species and contaminated areas. The two alter-

native approaches follow.

Determining risk for species relatively small home

ranges Once the animal density for each habitat subdivi-sion is determined, the proportion of the site populationexposed at contaminant concentrations higher than accept-able levels can be easily determined. The appropriatecontaminant concentration in each habitat subdivision isinput into individual-based wildlife exposure models (cita-tion) to characterize exposure in each habitat subdivision.The proportion of the population at risk is then determinedby summing the number of individuals in subareas whereexposure is above an acceptable threshold then dividing thisnumber by the total number of individuals in all subareas.

Estimating exposure to organisms with relatively large

home ranges The U.S. Fish and Wildlife Service (1981)defined an HSI as `a numerical index that represents thecapacity of a given habitat to support a selected fish orwildlife species’. An index, as defined by Inhaber (1976) isthe ratio of a value of interest divided by a standard ofcomparison. For habitat evaluation, the value of interest isan estimate of measure of habitat condition in the studyarea and the standard of comparison is the optimum habitatconditions for the same evaluation species. Therefore,HSI=(study area habitat conditions)/(optimum habitat con-ditions). The HSI has a minimum value of 0.0, whichrepresents unsuitable habitat, and a maximum value of1.0, which represents optimum habitat. An HSI modelproduces an index value between 0.0 and 1.0, with theassumption that there is a relationship between the HSIvalue and carrying capacity. HSI models can be used incases where the required output is either a measure of theprobability of use of an area by individuals or by apopulation. In applying HSIs to exposure assessment foranimals with exclusive or nonexclusive home ranges largerthan the contaminated site, HSI output should be viewed asa measure of the probability of an individual using a givensubsection of its home range. For this type of application,output from the HSI model can be used to estimate theproportion of its time that an individual will spend exploit-ing a given area within its home range. For organisms withhome ranges smaller than the contaminated area, the HSI ofa section of the site may be treated as a surrogate forcarrying capacity or population density. For organisms withlarge home ranges the proportion of time that the organismspends on the site is incorporated as an area use factor(AUF) in the dietary exposure equation as follows:

ADDpot ˆXm

sˆ1

Ps

Xn

jˆ1

…Cjs £FRjs £NIRj†‡…Ds £FS£FIRtotal†" #

…4†

Landscape ecology and ecological risk assessment

LA Kapustka et al.

244

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

whereADDpot potential average daily dose

Ps AUF; the proportion of time spent foraging in

subarea s (Equation 2)

Cjs average concentration of contaminant in food

type j in subarea s

FRjs fraction of food type j contaminated in subarea s

NIRj normalized ingestion rate of food type j

Ds average contaminant concentration in soils in

subarea s

NIRtotal normalized ingestion rate summed over all foods

FS fraction of soil in diet

The AUF is incorporated as illustrated in Equation (5).

ADDpot ˆXm

sˆ1

Ps

Xn

jˆ1

…Cjs £FRjs £NIRj†‡…Ds £FS£FIRtotal†" #

…5†

wherePs AUF; the proportion of time spent foraging in

subarea s (Equation 3)

Cjs average concentration of contaminant in food type j

in subarea s

FRjs fraction of food type j contaminated in subarea s

Ds average contaminant concentration in soils in

subarea s

Discussion

The current practice of EcoRA has been criticized as

lacking ecological context. The approach described in this

paper addresses this concern by explicitly considering

landscape features, which are important to wildlife species.

The approach builds on the U.S. EPA Framework and

presents a structured process to consider a wider array of

assessment species, guide development of a comprehensive

work plan, and use habitat quality estimates to modify

exposure assessments. Though the use of habitat features in

EcoRA were forwarded in a few publications, the practice

has not been adopted widely.

Linkov (2002) and Hope (2002) provide additional

foundations for using ecological features in EcoRA,

which could result in wider use of the approach. Hope

(2002) demonstrated the importance of considering the

quality of habitat in characterizing exposure. HSI values

described for discrete polygons could readily be sub-

stituted for the habitat quality designations used by

Hope.

These advances in the EcoRA process are expected to

improve risk estimates and provide valuable information

to be used in structuring risk management options. The

approach can guide the planning process so that an

assessment considers the most relevant species of the

area and defines the relevant parameters to be measured.

In risk characterization, these data are used to calculate

more realistic exposure assessments. In guiding remedia-

tion, the approach logically considers a wider range of

land management options than are considered at most

sites today. For example, habitat enhancement can be

used to draw animals away from contaminated zones.

Contaminated localities that also have poor-quality hab-

itat may be allowed to go through a slower, less costly

bioremediation process until the risk level is lowered to

acceptable levels. And direct comparisons of lost resour-

ces stemming from destructive remediation options can

be assessed instead of merely focusing on the lowering

of contaminant concentrations.

Acknowledgement

This work is funded through the American Chemistry

Council, Long-Range Research Initiative, Project Reference

No. 0236. Project Title: Evaluating habitat use to improve

exposure assessment in ecological risk assessments.

References

Freshman, J.S. and Menzie, C.A. 1996: Two wildlife exposure

models to assess impacts at the individual and population

levels and the efficacy of remedial actions. HERA 2, 481±

98.

Hope, B. 2000: Generating probabilistic spatially-explicit individ-

ual and population exposure estimates for ecological risk

assessments. Risk Analysis 20, 573±89.

Hope, B. 2002: Consideration of bioenergetic factors in spatially-

explicit assessments of ecological receptor exposure to

contaminants. Proceedings: Theories and Practices in Toxi-

cology and Risk Assessment, Cincinnati, OH 15±18 April

2002. Toxicology and Industrial Health (in review).

Inhaber, H. 1976: Environmental Indices. New York: John Wiley

and Sons, 178.

Linkov, I. 2002: Implementation and validation of spatially explicit

exposure models: application to military sites. Proceedings:

Theories and Practices in Toxicology and Risk Assessment,

Cincinnati, OH 15±18 April 2002. Toxicology and Industrial

Health 17, 230±35.

Linkov, I., Burmistrov, D., Cura, J. and Bridges, T.S. 2002: Risk-

based management of contaminated sediments: consideration

of spatial and temporal patterns of exposure modeling.

Environmental Science and Technology 36, 238±46.

Rand, G.M. and Newman, J.R. 1998: The applicability of habitat

evaluation methodologies in ecological risk assessment.

HERA 4, 905±29.

Schroeder, R.L. and Haire, S.L. 1993: Guidelines for the develop-

ment of community-level habitat evaluation models. Bio-

logical Report 8. Washington, DC: U.S. Department of

Interior, Fish and Wildlife Service.

Landscape ecology and ecological risk assessment

LA Kapustka et al.

245

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from

U.S. EPA. 1992: Framework for ecological risk assessment. EPA/

630/R-92-001. Washington, DC: U.S. Environmental Protec-

tion Agency.

U.S. EPA. 1993: Wildlife exposure factors handbook. Volume 1.

EPA/600/R-93/187a. Washington, DC: U.S. Environmental

Protection Agency.

U.S. EPA. 1997: Ecological risk assessment guidance for Super-

fund: process for designing and conducting ecological risk

assessments (interim final), Washington, DC: U.S. Environ-

mental Protection Agency.

U.S. EPA. 1998: Guidelines for ecological risk assessment. EPA/

630/R-95/002F, Washington, DC.

Whittaker, R.H. 1975: Communities and ecosystems. New York:

Macmillan Publishing Company.

USFWS (U.S. Fish and Wildlife Service). 1981: Standards for the

development of habitat suitability index models. ESM103,

Washington, DC.

USGS Website: http://www.nwrc.gov.wbd/pub/hsi/hsiindex.htm

Landscape ecology and ecological risk assessment

LA Kapustka et al.

246

at Universitats-Landesbibliothek on March 16, 2014tih.sagepub.comDownloaded from