Content uploaded by Wayne R Munns Jr
Author content
All content in this area was uploaded by Wayne R Munns Jr on Apr 09, 2018
Content may be subject to copyright.
H Kenneth Hudnell (ed.): Proceedings of the Interagency, International
Symposium on Cyanobacterial Harmful Algal Blooms
Advances in Experimental Medicine & Biology, 855-871 (2007)
Chapter 38: Integrating human and ecological
risk assessment: application to the
cyanobacterial harmful algal bloom problem
Jennifer Orme-Zavaleta, Wayne R Munns Jr
USEPA National Health and Environmental Effects Research Laboratory,
Office of Research and Development
Abstract
Environmental and public health policy continues to evolve in response to
new and complex social, economic and environmental drivers. Globaliza-
tion and centralization of commerce, evolving patterns of land use (e.g.,
urbanization, deforestation), and technological advances in such areas as
manufacturing and development of genetically modified foods have cre-
ated new and complex classes of stressors and risks (e.g., climate change,
emergent and opportunist disease, sprawl, genomic change). In recognition
of these changes, environmental risk assessment and its use are changing
from stressor-endpoint specific assessments used in command and control
types of decisions to an integrated approach for application in community-
based decisions. As a result, the process of risk assessment and supporting
risk analyses are evolving to characterize the human-environment relation-
ship. Integrating risk paradigms combine the process of risk estimation for
humans, biota, and natural resources into one assessment to improve the
information used in environmental decisions (Suter et al. 2003b). A bene-
fit to this approach includes a broader, system-wide evaluation that consid-
ers the interacting effects of stressors on humans and the environment, as
well the interactions between these entities. To improve our understanding
of the linkages within complex systems, risk assessors will need to rely on
a suite of techniques for conducting rigorous analyses characterizing the
exposure and effects relationships between stressors and biological recep-
tors. Many of the analytical techniques routinely employed are narrowly
856 Jennifer Orme-Zavaleta, Wayne R Munns Jr
focused and unable to address the complexities of an integrated assess-
ment. In this paper, we describe an approach to integrated risk assessment,
and discuss qualitative community modeling and Probabilistic Relational
Modeling techniques that address these limitations and evaluate their po-
tential for use in an integrated risk assessment of cyanobacteria.
Introduction
Cyanobacterial blooms occur in both fresh water and marine environ-
ments, producing a variety of toxins, and posing risks to humans and ani-
mals through recreational and drinking water use as well as consumption
of contaminated fish and shellfish (Codd et al. 2005). As a result of their
complex ecology, involving multiple endpoints we propose an integrative
approach in assessing the risks posed by cyanobacterial blooms.
The environmental risk assessment paradigm is shifting from independ-
ent analyses of human health or ecological effects to a more integrative, or
unified, approach. The idea of integrating risk assessment has been the
topic of extensive discussion over the past decade (e.g., Harvey et al. 1995;
WHO 2000). Integration ideally combines the process of risk estimation
for humans, biota, and natural resources into one assessment to improve
the information used in environmental decisions, resulting in more effec-
tive protection of both humans and the environment (Suter et al. 2003b).
A benefit to this approach is a broader, system-level evaluation that con-
siders the interactions of the effects of stressors on humans and the envi-
ronment, as well the interactions between these entities. In addition, stress-
ors other than chemicals need to be considered. The basis for such an
integrated approach would be the perspective that ecosystems serve as part
of the foundation defining human well-being and vice versa.
Risk assessments are important tools for informing public health and
environmental protection decisions. They constitute the scientific reason-
ing for estimating the likelihood of an adverse human or ecological effect
resulting from exposure to a stressor. Although the human health and eco-
logical risk assessment paradigms were developed independently, they are
related (Suter et al. 2003a). In both paradigms, risk characterization is a
key step providing a description of the evidence concerning the hazard, po-
tential exposures, and the uncertainties, variability, and assumptions used
in the assessment. Thus, the integration of risk assessment approaches is
encapsulated in the analytical processes it entails.
The shift in risk assessment to an integrated approach is consistent with
changes in the scientific approach to complex problems. In many in-
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 857
stances, a multidisciplinary approach is a necessity to evaluate cause and
effects relationships fully. Wilson (Wilson 1998) noted that science is no
longer a specialized activity, but involves the synthesis of causal explana-
tions. Thus, scientific research is shifting towards understanding linkages
within highly complex systems (Vitousek et al. 1997; Wilson, 1998; NAS,
2000; Forget and Lebel, 2001).
To improve our understanding of the linkages of complex systems as
part of an integrated risk assessment, risk assessors must rely on a suite of
techniques for conducting rigorous analyses characterizing exposure and
effects relationships among stressors and biological receptors. Current ana-
lytical techniques have been criticized as inadequate and irrelevant; they
can be misinterpreted due to a lack of understanding of the problem and
the inability to deal with uncertainty (NRC 1996; Peterman and Anderson
1999). Further, many of the commonly used techniques are narrow in fo-
cus and unable to evaluate complex systems adequately. In this paper, we
describe integrated risk assessment and review community-level modeling
techniques that account for current limitations. Lastly, we evaluate their
potential for integrated risk assessment of the cyanobacterial harmful algal
bloom (CHAB) problem.
Integrated Risk Assessment Paradigms
Over the past decade, several frameworks for integrating risk have been
proposed that are based on existing approaches for human health and eco-
logical risk assessments. Some approaches view integration in the context
of chemical exposures, combining acute and chronic risks to organisms
and considering exposures from different sources, pathways and routes
(Gurjar and Mohan 2003; Bridges and Bridges 2004). Harvey and cowork-
ers (Harvey et al. 1995) developed a ‘holistic’ approach that consisted of
concurrent and integrated health and ecological assessments. Their process
followed the steps originally outlined by the NAS (NAS 1983) conducting
human health and ecological assessments in parallel. A series of risk
choices is produced for the risk manager by integrating the results of two
parallel assessments during the risk characterization step. Using mercury
as a case study, they developed a risk characterization consisting of a se-
ries of risk estimates for humans exposed through inhalation or ingestion
that address neurological and reproductive effects, and for wildlife ex-
posed through the aquatic food chain addressing reproductive success and
decreased species distribution. The authors suggested that the series of
risk estimates would provide options for risk managers to choose from in
858 Jennifer Orme-Zavaleta, Wayne R Munns Jr
making a decision (Harvey et al. 1995). Although cast as a holistic process,
the Harvey et al. (Harvey et al. 1995) approach is not really integrative, but
rather a comparison of different risk values generated for different expo-
sure scenarios and toxicity endpoints; protective of different species. Thus,
this approach may be too generic and unresponsive to a particular problem
or management decision.
A special forum of the World Health Organization’s International Pro-
gramme on Chemical Safety (IPCS) developed another approach (Munns
et al. 2003). They outlined an integrated process combining elements of
both human health and ecological processes (WHO 2001; Suter et al.
2003b). This paradigm (Figure 1) is more closely aligned with the con-
cepts of the Guidelines for Ecological Risk Assessment (USEPA 1998).
Here, hazard identification becomes an element of problem formulation,
and dose response assessment occurs as part of the effects characterization.
Most importantly, this approach considers the interactions among stressors
and receptors such as wildlife or humans, and the abiotic environment.
One distinct difference of the IPCS integrated approach from the Harvey
et al. (Harvey et al. 1995), NAS (NAS 1983) and ecological risk paradigms
(USEPA 1998) is the involvement of stakeholders and risk managers in the
process. The human health and ecological risk paradigms were designed to
be independent from risk management so that their outcomes reflect scien-
tific analyses that are not influenced by socio-political bias. In the IPCS
approach, stakeholder and risk management involvement throughout the
process is viewed as essential to ensure buy-in and responsiveness of the
assessment to the specific problem, considering both human and ecological
risks where applicable (Suter et al. 2003b). While this, in and of itself,
does not ensure integration, it increases the potential depending on how the
problem is defined at the onset of the risk assessment.
The IPCS approach combines the process of risk estimation for humans,
biota, and natural resources into one assessment for the purpose of improv-
ing the information used in environmental decisions, resulting in more ef-
fective protection of resources valued by society (Miranda et al. 2002;
Suter et al. 2003b). Integration is achieved through all phases of the risk
assessment process (Suter et al. 2003b). Under problem formulation, inte-
gration entails the development of stressor-driven assessment questions
common to both health and environmental problems that focus on potential
susceptible human and ecological endpoints. Exposure and effects charac-
terizations are integrated through an evaluation of all the possible sources
of exposure and an understanding of common modes of toxic action in
humans and other organisms. Similar to the holistic approach (Harvey et
al. 1995), the IPCS risk characterization includes multiple estimates of risk
from which a best estimate of human and ecological risk is selected using a
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 859
common and consistent approach (Suter et al. 2003b). The authors go on to
indicate that evidence for health and ecological risks would be integrated
when appropriate but do not describe how this would be achieved.
Problem Formulation
With Hazard Identification
Analysis
Risk Characterization
INTEGRATED RISK ASSESSMENT
Characterization
of
Effects
Characterization
of
Exposure
Exposure
Assessment
Dose-
Response
Assessment
RISK MANAGEMENT
STAKEHOLDER PARTICIPATION
Fig. 1 Integrated Risk Assessment Paradigm. Adapted from WHO 2001.
The IPCS integrated approach was applied to several complex environ-
mental problems (Table 1). The case studies developed using the inte-
grated approach identify aspects of where integration can or should occur
with respect to exposure and effects characterization, but they do not actu-
ally conduct integrated assessments. Rather, they illustrate how such as-
sessments might be conducted. The risk characterization section in each of
the case studies largely reflects parallel risk comparisons. Two studies
(Ross and Birnbaum 2003; Vermeire et al. 2003) propose a common quan-
titative approach, a Toxic Equivalency Factor (TEF) approach as a means
of integrating risks. It is not clear, however, that having a common quanti-
tative approach to estimate risks for different species is actually integra-
tive, but rather reflects the commonalities in the toxic endpoints and
mechanisms of toxicity for the exposures and species of interest. Thus, the
IPCS approach goes beyond Harvey et al.’s (Harvey et al. 1995) holistic
approach in describing levels of integration throughout the risk assessment
process. However, the information included in the risk characterization
step largely presents parallel risk estimates for human and ecological end-
points under different exposure and effect scenarios. The responsiveness of
860 Jennifer Orme-Zavaleta, Wayne R Munns Jr
the assessment to a particular problem is likely to be greater under the
IPCS approach given the interaction with risk managers and stakeholders
throughout the process.
Other approaches to integrative assessments have been proposed that
focus on human and environmental linkages including socioeconomic and
political factors (e.g., Bruins and Heberling 2005; Stahl et al. in press), or
have focused more broadly on human health-ecological integrity reflecting
dimensions of both the natural and social systems (Miranda et al. 2002).
Epstein (Epstein 1994) developed an integrated assessment framework of
climate change and ecosystem vulnerability. His generalized framework
depicted overlapping and interacting climate and social systems with eco-
systems whose intersection directly or indirectly produced various out-
comes ranging from changes in health, crop yields, and demography to
economic productivity. Epstein noted that integration was dependent on
the use of specific biological, social or geochemical indicators depicting
the functions of complex systems. Referring to the complex relationship
between disease emergence and changes in climate and ecosystems, Ep-
stein (Epstein 1994) proposed a number of principles for modeling and
monitoring complex ecosystems. He emphasized the need to account not
only for direct impacts to the different systems but also those indirect ef-
fects resulting from the interactions among factors within the three over-
lapping systems. He noted that those diseases transmitted directly from
person to person reflect changes in population density with little interac-
tion among the three systems, while vector-borne diseases reflect envi-
ronmental changes involving all three systems in his integrated model. In-
tegration in Epstein’s approach also occurs through scientific and political
collaborations. He did not present an overall assessment of risk but sug-
gested guidelines for identifying system vulnerabilities affecting overall
stability and resilience; key elements in his view for mitigating disease
emergence.
Chapter 38: Integrating human and ecological risk assessment: application to the cyanobacterial harmful algal bloom
problem 861
Table 1 Summary of IPCS integrated risk assessment case studies.
Environmental
Problem Assessment
Endpoints Areas of
Integration Proposed Risk Characterization Reference
“Dioxin-like” Persistent
Organic Pollutants
Humans and upper
trophic level wildlife
٠Route of exposure
٠Mode of action
٠Toxicity
Apply Toxic Equivalency Ap-
proach (TEF) to both humans and
wildlife
Ross and Birnbaum
2003.
Tributyl- and triphenyltins Humans and piscivo-
rous wildlife
٠Route of exposure
٠Mode of action
٠Toxicity
Species and exposure-specific hu-
man and ecological risk estimates
Sekizawa et al. 2003.
UV-Radiation Amphibians, coral,
humans, and oceanic
primary productivity
٠Exposure pathways
٠Mechanistic pathways
Parallel characterization of risk
across assessment endpoints.
Hansen et al. 2003.
Organophospherous pesti-
cides
Humans and wildlife ٠Exposure pathways
٠Toxicity
Species-specific TEFs Vermeire et al. 2003.
Nonylphenol Humans (occupational
and environmental
exposure), wildlife,
and aquatic organisms
٠Exposure pathways
٠Mechansim of action
٠Toxicity
Species and exposure-specific hu-
man and excological risk estimates
Bontje et al. 2004
862 Jennifer Orme-Zavaleta, Wayne R Munns Jr
Vanleewen et al. (Vanleewen et al. 1999) presented a conceptual ‘but-
terfly’ model that focuses on human health in an ecosystem context. Hu-
man health is determined from the intersection of biophysical socioeco-
nomic environments. The boundaries of the butterfly could be at the
community, watershed, or population level and include the interactions be-
tween humans and the nonhuman environment. Their model is not an ap-
proach for assessing risk per se but can be viewed as a mechanism for de-
termining risk factors influencing human health. As the authors noted, this
model focuses only on human health and does not determine health for
other species in the ecosystem, limiting its utility for comprehensive as-
sessment of risk.
Integrative Analytical Approaches to Risk Assessment
The integrated paradigms described above provide frameworks for consid-
ering human and environmental interactions but fall short of demonstrating
specific analytical techniques for conducting an integrated risk analysis.
The examples include a mix of conceptual, integrated approaches that are
either descriptive or consist of parallel risk assessments. Considering the
models presented by Epstein (Epstein 1994) and VanLeeuwen et al.
(VanLeeuwen et al. 1999), it is clear that an evaluation of interactions
among human populations, their environment, and other important eco-
logical factors are needed in conducting an integrated analysis. This type
of evaluation is similar to that encompassed by an ecoepidemiological ap-
proach. Similar to human epidemiology, ecoepidemiology has been used to
study the ecological effects that are prevalent in certain areas among popu-
lation groups, communities and ecosystems and their potential causes
(Bro-Rasmussen and Løkke 1984; Martens 1998). This approach focuses
on a description of the effects, identification of causes, and understanding
their linkages. Humans are considered as part of the environment in these
analyses.
An ecoepidemiological approach is similar to community and systems-
level ecological risk assessment with respect to understanding relationships
between biotic and abiotic factors. Levins (Levins 1973) noted that ad-
dressing more complex systems required breaking down disciplinary
boundaries to create an integrated process that addresses management
goals in which community structure and other mechanistic factors could be
examined as a whole. A system in this context is defined as a habitat, geo-
graphic area, human community or network of communities (Levins
1998). As complexity increases, the ability to gather quantitative informa-
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 863
tion is complicated by the impracticality of the number of parameters to
measure and the loss of realism (Levins 1966; Puccia and Levins 1991).
Qualitative models can simplify complex systems without sacrificing
realism (MacArthur and Levins 1965; Levins 1966) and enable an inte-
grated analysis of a system. Qualitative modeling in the form of signed di-
graphs, ‘loop analysis,’ and matrix analysis facilitates the understanding of
a system where there is incomplete information. Because qualitative mod-
els involve only the signs of the interactions among variables (positive,
negative, or no change), variables representing poorly quantified aspects of
the system can be included in the analysis (Puccia and Levins 1991). Such
variables represent not only different species, but also resources, climate,
or socioeconomic variables that influence community structure and func-
tion. When constructing models, qualitative modeling methods can help
determine which variables should be included, what should be measured,
and how system dynamics might be affected under different perturbation
(stresses that result in a permanent change in a growth parameter) scenar-
ios (Levins 1998).
Loop analysis and the corresponding community matrix is a useful ana-
lytical tool for exploring and understanding the effects of natural and an-
thropogenic stress on a system. Dambacher et al. (Dambacher et al. 1999)
used this modeling procedure to characterize a predator-prey system in-
volving snowshoe hare and arctic fox. This technique also proved useful in
predicting the impact of species introductions into a community (Li et al.
1999; Castillo et al. 2000) and explaining complex transitions in commu-
nity composition over time (Bodini 1998; Ortiz and Wolff 2002). Loiselle
et al. (2000; 2002)) used loop analysis to examine different economically-
based management scenarios in a wetland ecosystem to identify manage-
ment options and guide monitoring programs.
In the context of integrated risk, Levins (Levins 1998) extended qualita-
tive modeling to the problem of vector-borne disease. In his system, he
identified the invasiveness of vectors and disease reservoirs as core vari-
ables that would be important in an epidemic, adding vector habitat re-
quirements, vector and host behaviour, host health status, and economic
variables as other factors to be considered. With an increasing ‘web of
causation,’ Levins (Levins 1998) argued that internal processes critical to
community function could be examined. On further analysis of this prob-
lem, Orme Zavaleta and Rossignol (Zavaleta and Rossignol 2004) devel-
oped a procedure to predict disease risk that combines recent developments
in qualitative community modeling with biomathematical theory of vector-
borne disease transmission. This procedure predicts the change in risk of
vector-borne disease following perturbations such as increases in vector
864 Jennifer Orme-Zavaleta, Wayne R Munns Jr
abundance, animal control measures, habitat alteration, or global warming.
Like Levin’s postulated epidemic-disease community, this procedure al-
lows the consideration of a complex community structure linking ecologi-
cal factors to human disease. This procedure results in a rigorous predic-
tion of an ecological community response to a perturbation with minimal
to no quantitative parameterization. It generates focused hypotheses to
guide data collection and control management strategies as interventions.
Bayesian analyses in the form of Bayesian networks are another tool
that can be useful in an integrated risk analysis. A Bayesian approach is
based on probability theory and is a useful decision-making or inferential
technique when there is incomplete information or it is not possible to
gather enough information to reduce uncertainties (Reckhow 2003). A
Bayesian network is used to model a system containing uncertainty. It of-
fers both qualitative and quantitative information in the form of conditional
probabilities and can be applied to multivariate problems involving com-
plex relationships among variables (Reckhow 2003). A Bayesian network
consists of a directed acyclic graph and a probability distribution. The net-
work characterizes variable relationships through interrelated nodes and
arcs. The nodes represent variables and the arcs represent conditional de-
pendencies between the nodes. Bayesian networks are used to identify
those key variables influencing relationships within a system, and thus are
an integrative analytical tool.
The use of Bayesian networks is increasing in scientific analyses of
complex problems. Crome et al. (Crome et al. 1996) applied a Bayesian
approach to evaluate the impact of logging on bird and mammal species in
rain forests. The investigators had too few data to detect potential impacts
using traditional statistical analysis. However, results of a Bayesian analy-
sis suggested a correlation between canopy cover and impacted bird spe-
cies that was not previously apparent. Further, of the 76 species of birds in
question, only four species were identified as having a high probability of
being adversely impacted by logging.
Bayesian networks have also been used to guide such diverse analyses
as land management decisions (Marcot et al. 2001), fish stock assessment
(Varis et al. 1993; Hammond and Ellis 2002), and potential risk factors as-
sociated with heart disease (Buntine 1991). Each of these cases started
with a hypothesized model that could be updated as additional information
became available, and involved large uncertainties, the pooling of informa-
tion from different datasets, and expert judgment in the analysis.
When a specific model is not known, a data discovery technique,
Probablistic Relational Modeling (PRM), can conduct a heuristic search of
independent data sets to generate data-derived models (Jorgensen 2003).
This technique involves machine learning guided by expert judgment to
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 865
develop a probabilistic model. The PRM extends Bayesian networks to the
relational level, modeling uncertainty related to variables, their properties,
and relationships among them (Getoor et al. 2001). The probabilistic rela-
tionship between variables is such that a change in any one variable affects
all the others. Thus, PRMs are well suited for application to complex sys-
tems.
There are a few examples of where PRM has been used to evaluate
complex problems. Getoor et al. (Getoor et al. 2001) described a PRM
analysis to determine possible probabilistic relationships between patients
from a tuberculosis clinic, certain risk factors, and specific strains of tuber-
culosis. In a second example, Jorgensen et al. (Jorgensen et al. 2003) used
a PRM approach to explore the long-term changes in the clarity of Crater
Lake using information summarized in multiple databases. The PRM
analysis enabled the investigators to construct multiple, complex hypothe-
ses concerning the entire lake ecosystem given data obtained from the
long-term studies of the lake.
Probablistic Relational Modeling was also used by Orme Zavaleta et al.
(Zavaleta in review) to identify probabilistic relationships associated with
the transmission of West Nile virus in Maryland. Similar to the Crater
Lake study (Jorgensen et al. 2003), the RBM approach was used to explore
relationships among multiple, independent databases. Multiple hypotheses
were generated suggesting spatial and temporal relationships between key
vector, host and habitat variables related to disease transmission.
Thus, the PRM technique appears to be an effective means of conduct-
ing an integrated risk analysis through the qualitative and quantitative
evaluation of complex community interactions. The hypotheses generated
by the PRM analysis can be used to guide further quantitative testing of
specific relationships between probabilistically linked variables.
Integrated Risk of CHABs
The concepts of integrated risk assessment can be applied to the problem
of CHABs to help define the specific information, tools and research
needed for effective decision-making and action. Although it would be
presumptuous to attempt a fully developed assessment in this paper, com-
munication of an initial conceptual model can facilitate the discussion and
additional analyses required to advance such an assessment. The concep-
tual model in Figure 2 reflects existing knowledge about the factors con-
tributing to blooms, the health, ecological and socioeconomic effects of
blooms, and the linkages among important components of this multifaceted
866 Jennifer Orme-Zavaleta, Wayne R Munns Jr
system. Different pathways of exposure and effect are shown for humans,
wildlife, and aquatic plants and animals, as the processes influencing these
groups of receptors vary. However, the pathways intersect to illustrate
system linkages, or when biological processes are common to multiple re-
ceptor groups. Along the bottom row of the model are loose expressions
of candidate assessment endpoints for an integrated risk assessment, re-
flecting some of the values whose protection may underlie the need for
management action to control or mitigate CHABs.
Additional refinement of the conceptual model is required to advance an
integrated assessment of the risks of cyanobacterial blooms. Are the im-
portant environmental processes and factors controlling blooms captured?
Is the array of assessment endpoints important to the CHABs problem ar-
ticulated fully? Are the key system components and their linkages de-
scribed adequately? With agreement on the adequacy of the conceptual
model, planning discussions can address the availability of tools and in-
formation required to evaluate critical risk hypotheses represented in the
model, potentially leading to identification of additional data and research
needed to complete the assessment. Though challenging, performance of a
thoroughly-planned integrated risk assessment would support comprehen-
sive decisions for managing the risks of CHABs.
Chapter 38: Integrating human and ecological risk assessment: application to the cyanobacterial harmful algal bloom
problem 867
Fig. 2. Conceptual model for cyanobacterial harmful algal blooms.
868 Jennifer Orme-Zavaleta, Wayne R Munns Jr
Discussion
To conduct an integrated risk assessment of CHABs, a suite of tools is
needed that integrates human and environmental health in the problem
formulation (for hypothesis generation) and analysis phases of the assess-
ment, not simply during the risk characterization phase. Such tools should
consider the interacting system as a whole, from the environmental proc-
esses that influence CHAB formation to the changes caused by those
blooms in the combined human and ecological system. Although this adds
complexity in the analysis, models and other decision support methods are
available that can simplify and reduce complexity.
The ‘integrative’ models reviewed in this paper may not be robust
enough to integrate multiple stressors or multiple endpoints in their use of
either parallel assessments or deductive reasoning to remove stressor inter-
actions from consideration. The analytical techniques employed in these
models to characterize risk are applied to either human health or ecological
assessments. Qualitative modeling and Probablistic Relational Models
provide an integrated risk analysis framework that identifies relationships
important in the system and thus, guide the application of quantitative
models or provide sufficient information for management decisions. Both
techniques rely on community structure for generating hypotheses and test-
ing predictions. Experimental comparison of various community theories
suggests that loop analysis may be the theoretical approach best suited for
predicting the behaviour of complex community structures following a
perturbation (Hulot et al. 2000). Used in conjunction with mechanistic
models, the integrated analytical techniques provide a balanced, iterative
approach for not only assessing risk, but evaluating possible consequences
of different CHABs management scenarios.
References
Bodini A (1998) Representing ecosystem structure through signed digraphs;
model reconstruction, qualitative predictions and management: The case of
freshwater ecosystem. Oikos 83:93-106
Bontje D, Hermens J, Vermeire T, Damstra T (2004) Integrated Risk Assessment:
Nonylphenol Case Study. International Programme on Chemical Safety
(IPCS). WHO/IPCS/IRA/12/04 Geneva, Switzerland
Bridges JW, Bridges O (2004) Integrated risks assessment and endocrine disrupt-
ers. Toxicol 205:11-15
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 869
Bro-Rasmussen F, Løkke H (1984) Ecoepidemiology – A casuistic discipline de-
scribing ecological disturbances and damages in relation to their specific
causes: Exemplified by chlorinated phenols and chlorophenoxy acids. Regula-
tory Tox and Pharm 4:391-399
Bruins RJF, Heberling MT (2005) Economics and Ecological Risk Assessment.
CRC Press, New York
Buntine W (1991) Theory refinement on Bayesian networks. IN: Proceedings of
the Seventh Conference of Uncertainty in Artificial Intelligence. Morgan
Kaufmann (eds). San Fransisco pp 52-60
Castillo GC, Li HW, Rossignol PA (2000) Absence of overall feedback in a ben-
thic estuarine community: A system potentially buffered from impacts of bio-
logical invasions. Estuaries 23:275-291
Codd GA, Lindsay J, Yound FM, Morrison LF, Metcalf JS (2005) Chapter 1.
Harmful Cyanobacteria. From mass mortalities to management measures. In:
Huisman J, Matthijs HCP, Visser PM (eds) Harmful Cyanobacteria. Springer,
Netherlands, pp 1-23
Crome FHJ, Thomas MR, Moore LA (1996) A novel Bayesian approach to assess
the impacts of rain forest logging. Ecological Applications 6:1104-1123
Dambacher JM, Li HW, Wolff JO, Rossignol PA (1999) Parsimonious interpreta-
tion of the impact of vegetation, food, and predation on the snowshoe hare.
Oikos 84: 530-532
Epstein PR (1994) Framework for an integrated assessment of climate change and
ecosystem vulnerability. In: Wilson ME, Levins R, Spielman A (eds) Disease
in Evolution: Global Changes and Emergence of Infectious Diseases. New
York Academy of Science, New York, pp 423-435
Forget G, Lebel J (2001) An ecosystem approach to human health. International J
of Occup and Env Health Supplement to 7:3-36
Gurjar BR, Mohan M (2003) Integrated risk analysis for acute and chronic expo-
sure to toxic chemicals. J Haz Mat A102:25-40
Hammond TR, Ellis JR (2002) A meta-assessment for elasmobranches based on
dietary data and Bayesian networks. Ecological Indicators 1:197-211
Hansen L, Hedtke SF, Munns WR Jr (2003) Integrated human and ecological risk
assessment: A case study of ultraviolet radiation effects on amphibians, coral,
humans, and oceanic primary productivity. Human and Ecol Risk Assessment
9: 359-377
Harvey T, Mahaffey KR, Velazquez S, Dourson M (1995) Holistic risk assess-
ment: An emerging process for environmental decisions. Reg Tox and Pharm
22:110-117
Hulot FD, Lacroix G, Lescher-Moutoue F, Loreau M (2000) Functional diversity
governs ecosystem response to nutrient enrichment. Nature 405:340-344
Jorgensen J, D’Ambrosio B, Rossignol PA (2003) Data-driven construction of
community models of Crater Lake. NSF Biocomplexity Workshop -The verti-
cal organization of energy, carbon, and nutrient cycles in an ultraoligotrophic
ecosystem: A workshop on Crater Lake, Oregon February 16 -18
870 Jennifer Orme-Zavaleta, Wayne R Munns Jr
Lackey R (1997) If ecological risk assessment is the answer, what is the question?
Human and Ecol Risk Assessment 3:921-928
Levins R (1966) The strategy of model building in population biology. Am Nat
54:421-431
Levins R (1973) Fundamental and applied research in agriculture. Science
181:523-524
Levins R (1998) Chapter 11 Qualitative mathematics for understanding, predic-
tion, and intervention in complex ecosystems. IN: Approaches to Assessing
Ecosystem Health. Rapport D, Costanza R, Epstein P, Gaudet C, Levins R
(eds). Blackwell MA, pp 178-204
Li HW, Rossignol PA, Castillo G (1999) Chapter 30: Risk analysis of species in-
troductions: Insights from qualitative modeling. In Nonindigenous Freshwater
Organisms. Vectors, Biology, and Impacts. Claudi R, Leach JH (eds). Lewsi
Publ, Boca Raton pp 431-447
Loiselle S, Carpenato GM, Hull V, Waller T, Rossi C (2000) Feedback analysis in
reserve management: studying local myths using qualitative models. Ecol
Modelling 129:25-37
Loiselle S, Hull V, Permengeat E, Falcucci M, Rossi C (2002) Qualitative models
to predict impacts of human interventions in a wetland ecosystem. Web Ecol-
ogy 3:56-69
MacArthur RH, Levins R (1965) Competition, habitat selection, and character
displacement in a patchy environment. Proc National Academy Science
51:1207-1210
Marcot BG, Holthausen RS, Raphael MG, Rowland MM, Wisdom MJ (2001) Us-
ing Bayesian belief networks to evaluate fish and wildlife population viability
under land management alternatives from an environmental impact statement.
For Ecol and Management 153:29-42
Martens P (1998) Health and Climate Change. Earthscan Publs Ltd, London
Munns WR Jr, Kroes R, Veith G, Suter II GW, Damstra T, Waters M (2003) Ap-
proaches for integrated risk assessment. Human and Ecol Risk Assessment 9:
267-272
National Research Council (NRC), National Academy of Sciences (1983_ Risk
Assessment in the Federal Government. National Academy Press, Washington
DC
National Research Council (NRC), National Academy of Sciences (1996) Under-
standing Risk: Informing Decisions in a Democratic Society. National Acad-
emy Press, Washington DC pp 249
National Research Council (NRC) (2000) Grand Challenges in Environmental
Sciences. National Academy Press, Washington DC
Orme Zavaleta J, Rossignol PA (2004) Community-Level Analysis of Risk of
Vector-Borne Disease. Trans Royal Soc Trop Med & Hyg 98(10): 610-618
Orme Zavaleta J, Jorgensen J, D’Ambrosio B, Altendorf E, Rossignol PA (in
press) Discovering Spatio-Temporal Models of the Spread of West Nile Virus.
Risk Analysis
Chapter 38: Integrating human and ecological risk assessment: application to the
cyanobacterial harmful algal bloom problem 871
Ortiz M, Wolff M (2002) Application of loop analysis to benthic systems in north-
ern Chile for the elaboration of sustainable management strategies. Marine
Ecol Progress Series 242:15-27
Peterman RM, Anderson J (1999) Decision analysis: A method for taking uncer-
tainties into account in risk-based decision making. Human and Ecological
Risk Assessment 5:231-244
Puccia CJ, Levins R (1991) Chapter 6. Qualitative modeling in ecology: Loop
analysis, signed digraphs, and time averaging. IN: Qualitative Simulation
Modeling and Analysis. Fishwick PA, Luker PA (eds). Springer-Verlag Publ,
New York pp 119-143
Reckhow KH (2003) Bayesian approaches in ecological analysis and modeling.
IN:The Role of Models in Ecosystem Science. Canham CD, Cole JJ, Lauen-
roth WK (eds). Princeton Univ Press, In press
Ross P, Birnbaum L (2003) Integrated human and ecological risk assessment: A
case study of persistent organic pollutant (POPs) risk to humans and wildlife.
Human and Ecol Risk Assessment 9: 303-324
Sekizawa J, Suter GW, Birnbaum L (2003) Integrated human and ecological risk
assessment: A case study of tributyltin and triphenyltin compounds. Human
and Ecol Risk Assessment 9: 325-342
Stahl RG Jr, Kapustka L, Bruins RJF, Munns, WR Jr (eds) (in press) Valuation of
Ecological Resources: Integration of Ecological Risk Assessment and Socio-
economics to Support Environmental Decisions. SETAC Press, Pensacola, FL
Suter GW II, Norton SB, Barnthouse LW (2003a) The evolution of frameworks
for ecological risk assessment from the Red Book ancestor. Human and Ecol
Risk Assessment 9: 1349-1360
Suter GW II, Vermeire T, Munns W Jr, Sekizawa J (2003b) Framework for the in-
tegration of health and ecological risk assessment. IPCS Workshop summary.
Human and Ecol Risk Assessment 9: 281-301
US Environmental Protection Agency (USEPA) (1995) Guidance for Risk Char-
acterization. Office of Science Policy. February
Varis O, Kuikka S, Kettunen J (1993) Belief networks in fish stock assessment –
The Baltic salmon case. ICES Statutory Meeting Ref M Statistics Committee
Ref Anacat Committee pp 1-18
Vermeire T, MacPhail R, Waters M (2003) Integrated human and ecological risk
assessment: A case study of organophospherous pesticides in the environ-
ment. Human and Ecol Risk Assessment 9: 343-357
Vitousek PM, Mooney HA, Lubchenco J, Mellilo LM (1997) Human Domination
of Earth’s Ecosystems. Science 277:494-499
Wilson EO (1998) Integrated science and the coming century of the environment.
Science 279:2048-2049
World Health Organization (2001) Integrated Risk Assessment. International Pro-
gramme on Chemical Safety (IPCS). WHO/IPCS/IRA/01/12 Geneva, Switzer-
land.
