temporal scales in ecological risk assessment

Arch. Environ. Contain. Toxicol. 23,484-488 (1992)

E A r c h i v e s o f

n v i r o n m e n t a l c o n t a m i n a t i o n

q l m a n d

| ox ico logy © 1992 Springer-Verlag New York Inc.

Temporal Scales in Ecological Risk Assessment

Joanna Burger,*'** and Michael Gochfeld**'***

*Department of Biological Sciences, Rutgers University, Piscataway, New Jersey 08855, USA; **Environmental and Occupational Health Sciences Institute, Piscataway, New Jersey 08855, USA, and ***Environmental and Occupational Medicine, UMDNJ-Robert Wood Johnson Medical School, Piscataway, New Jersey 08854, USA

Abstract. The process of human health risk assessment (HRA) was formalized in 1983 by the National Research Council to include hazard identification, dose-response analyses, exposure assessment and risk characterization. Risk assessment for ecologic endpoints is emerging as a new discipline. Al- though environmental impact statements have been conducted for many years, ecologists, managers and policy makers are beginning to formalize the process in terms of risk, and are adapting the HRA paradigm to ecological risk assessment (ERA). In this paper it is suggested that the temporal scales of the two processes differ, and that these differences should be incorporated in ecological risk assessment. Even when HRA techniques are applied to a single non-human species there are temporal variations including: (1) different and often variable life spans, (2) unpredictable lengths of lifestages and different metamorphic stages, and (3) indeterminate growth in some species. Wben these differences are considered for multispecies assemblages, the impact on the food web will result in exposures of differing magnitudes affecting different species. The challenges for ERA include developing general principles for estimating or predicting exposure to critical life stages of the dominant species in an ecosystem, and establishing the appropriate temporal scales for predicting impacts or evaluating outcomes.

In the last decade, Environmental or Health Risk Assess- ment has played an important role in decisions about potential adverse effects of exposure to chemical and physical hazards in the environment. The Health Risk Assessment Paradigm (hereafter called HRA) was formalized by the Na- tional Research Council publication (1983) which listed four phases: Hazard identification, Dose-response analysis, Expo- sure assessment and Risk characterization. More recently, the applicability of the HRA paradigm to Ecological Risk Assess- ment (hereafter referred to as ERA) has been the subject of much discussion (Barnhouse 1992). HRA evaluates the risks to humans of potential exposure to environmental hazards, whereas ERA evaluates the risks to ecosystems with their component parts (O'Neill et al. 1982; National Research Council 1986). Thus, ERA is a broader assessment with inherently

more complexity, and may include humans as an integral part of a system.

In this paper, the HRA paradigm and its applicability to ERA is briefly reviewed, and one of the major differences that affects how these two types of risk assessment are viewed is discussed. Variations in temporal scales of exposure and effect increase the difficulties of ERA because of the differential rates of movement of pollutants through parts of the ecosystem are due to very different lifespans of vector organisms. This results in variations in exposure at each trophic level. Ecological Risk Management decisions should be based on different temporal criteria as well.

The Health Risk Assessment Paradigm

The HRA paradigm was formalized to provide some structure and consistency to health risk assessment (National Research Council 1983). Although many of the steps were performed previously, often not all steps were performed, or they were not given appropriate weight.

Hazard identification is the phase wherein the specific chemical or physical agents and the corresponding biological endpoints are identified. Both target or receptor human populations and specific disease outcomes are defined. The Dose-response phase involves using information from toxicological and epidemiological studies to determine the relationships between dose and the endpoint(s) of interest. Endpoints are specific diseases or demographic consequences, and excess cancer mortality is frequently used. A basic assumption of dose-response analysis is the validity of extrapolating from one animal species (often rodents) to another, humans (Rall 1969; Okker- man et al. 1991). A similar assumption must be made for ecological systems given the large number of plant and animal species in very diverse groups. Exposure assessment esti- mates the magnitude and the temporal pattern of how individuals are exposed to different media (air, food, water via inges- tion, inhalation and dermal absorption). Finally, in the risk characterization phase a particular dose is associated with a particular probability of an adverse outcome (i.e. cancer risk), either through the application of mathematical models or "safety factors."

Ecological Risk Assessment 485

Ecological Risk Assessment

The process of discussing the impact of industrial, other human activities, or development on the environment has been formalized and legalized in the environmental impact assessment (EIS) process (National Research Council 1981, 1986). By this method, new or modified human activities are examined in light of their potential effects on the environment, including specific plants and animals as well as on the total system (see Sheehan et al. 1984). The process often involves determining whether there are endangered or threatened species on the site, and how such species will be impacted by the proposed activities.

Although the EIS process has proven useful to managers and regulators, there are shortcomings inherent in its usual imple- mentation including: (1) there is no formalized procedure that insures uniformity of methodology or assessment, (2) there is often no attempt to assess the methods or routes of exposure, (3) there is sometimes no real assessment of the risks in an attempt to predict future adverse outcomes to a functioning ecosystem, (4) approaches are usually reductive, focusing on each species as a separate entity, and, finally, (5) the actual impacts are usually not evaluated, so there is no feedback to the process. This latter problem is severe, because evaluation of whether particular hazards caused specific effects or are likely to cause these effects in the future is often impossible. In very few cases is there dose-response information for environmental hazards.

The complexities of ERA often encourage reliance on indica- tor species to evaluate ecosystem degradation (O'Conner and Dewling 1986; Karr 1991). Individual species are often evaluated by ecotoxicologists using the same kinds of toxicity tests used to predict human effects. Traditional ecotoxicological tests (Butler 1978) no doubt will work well for single species evaluations (Miller 1984; Sheehan 1984a; Dawson and Wilke 1991), but their predictive value for the overall effects of hazards on populations, communities and ecosystem functions are often not examined. The examination of higher-order ecological functions has gained attention mainly since the mid-1980s (Sheehan 1984b; National Research Council 1986; Cairns 1990; Webber et al. 1992). Examining ecosystem effects will not only involve direct measurements, but will include model- ing (Gilpin and Hanski 1991; Johnson et al. 1991).

Recently, there has been interest in applying the formalized, four-part HRA to Ecological Risk Assessment. This has the advantage of ensuring that all four steps are invoked in all cases. It further ensures that future outcomes are carefully considered in view of the component parts of the ecosystem. Assessing ecological risk on an ecosystem or a regional scale is in its early stages (Hunsaker et al. 1990). However, Graham et al. (1991) designed an ecological risk assessment on a regional scale using land cover, edge habitat, landscape indices, and lake water quality.

Applying the HRA paradigm to ERA has several problems, largely because instead of considering the outcome for only one species (humans), the risk assessor must consider not only status, but interactions of many populations, many species, and several communities in a complex ecosystem. Developed to its fullest, ERA will focus on the ecosystem functions such as food web structure, energy transfer, population interactions, and biomass production. Ecologists have studied these functions for decades, but applying them to ERA is in its infancy.

There are not enough resources to examine every species, so difficult choices must be made. Nonetheless, ERA is essential to evaluate the risks to ecosystems rather than to consider risks for only one or two species. Although the public is interested in the welfare of species that are appealing, endangered, or of economic interest, other species may have more important or keystone roles in a particular ecosystem (Paine 1966). More- over, public concerns seldom revolve around habitats and ecosystems (Fischer et al. 1991) and these are often neglected endpoints in environmental impact statements.

Temporal Similarities in HRA and ERA

In both HRA and ERA the difference between acute and chronic exposures, between single and recurrent exposures, and between predictable and non-predictable exposures must be accounted for. In both cases there is a variable, and often unknown, latency between exposure and consequence. Acute exposures may have immediate and short-term consequences or long-delayed, chronic effects. Chronic exposure may be cumulative, producing an impact only after some threshold is ex- ceeded. These considerations apply to both HRA and ERA. However, it is the differences in temporal characteristics that concern us here.

Temporal Differences in Risk Assessment

One aspect that varies markedly between Human Risk Assess- ment and Ecologic Risk Assessment is the temporal scales. Exposure for humans is assigned to particular parts of a life cycle, so many hours per week over so many years. A worst case is full-time exposure for a 70 year life span. Similarly, the time course of the adverse outcome can vary from immediate to long-delayed (e.g., cancer), and from acute to chronic. For organisms in an ecosystem there are also critical periods of exposure, but lifetimes may range from minutes (bacteria) to hundreds of years (trees). Functions are also highly seasonal and often vary unpredictably. Species interactions may vary through epicycles lasting months to decades. Such considerations have profound effects on Ecological Risk Assessment.

There are three characteristics of humans that constitute a difference between HRA and ERA with respect to temporal variables: (1) the human lifespan can be taken as uniform (bar- ring disease and accidents) with discrete life stages, (2) growth (and thus size) ceases for humans following adolescence, and (3), the stages of the human species are readily definable and do not involve metamorphosis (Table 1).

When the complexities of multispecies assemblages are considered, as they should be for ERA, then these individual species differences translate into a complex web of exposure scenarios. Because of these three differences in life history and development patterns, the movement of chemicals through an ecosystem, and the respective vulnerabilities and dependen- cies, will interact to produce repeated exposures of differing magnitudes.

In the case of humans, risk assessors must evaluate the hazards for chemicals or other toxics during fetal development, in infancy and childhood, and in adults. Each of these developmental periods in humans is circumscribed both in duration and

486 J. Burger and M. Gochfeld

Table 1. Comparison of temporal differences in human health risk assessment and ecological health risk assessment

Endpoint HRA ~ ERA b

Population dynamics Effective population size Temporal variables

Lifespan Lifeform

Developmental stages

Adult size

Humans Small to large

70 + years One form Fetal, Infancy, childhood, adult.

Predictable in length Determinate

All plant and animal species in the ecosystem (or only dominants) Very small (endangered species) to large (e.g., algal blooms)

Minutes to several hundred years Variable from one form to several metamorphic stages Immature and adult may be predetermined or variables depending

on biotic factors such as food, climate, temperature. Determinate to indeterminate

aHRA = Health risk assessment hERA = Ecological risk assessment

timing, and relative size progresses in a recognizable developmental pattern. Although this is true for many other vertebrates, it is not for all vertebrates (e.g., many fish continue to grow, albeit slowly, throughout life), and is generally not true for

Acute many invertebrate animals and plants. Subacute

The life cycle can also be reduced to three stages of evolu- Subchronic tionary consequence: Pre-reproductive, reproductive and post- Chronic reproductive. Most human populations value their post-repro- Lifetime ductive members, whereas in natural populations this stage is seldom reached.

Even for some vertebrates, such as snakes, the prereproduc- tive period in a given species can vary depending on food lo0- availability and temperature regimes (which influence the ac- ,- " tive period for cold-blooded vertebrates). Reptile species living .o 80-

. m

farther north have shorter growth periods each year, and thus ~- - O may take years longer to mature (Tinkle 1961; Gibbons and u 60-

Semlitsch 1982; Gibbons 1987). Fish may require different *~ . e -

time periods to mature as well. ~ 40- That growth (i.e., size) generally ceases for humans in the

late teens simplifies exposure assessments for any hazard. For " species that continue growing throughout life, the complexities ~ zo- of developing growth curves are increased, and the relationship g between exposure and body weight (often called "dose") con- 0 tinually changes.

For species that undergo metamorphosis, such as frogs, toads, and many insects, separate dose-response curves need to be developed for each life phase. The addition of different life stages only compounds the problems associated with different lifespans and developmental periods.

Ecological Risk Assessment: Complexity of Exposures

One key feature of ecosystems is the movement of energy and nutrients (i.e., food) through a complex food web (Odum 1957; National Research Council 1986). It is this complexity that distinguishes Ecological Risk Assessment, because the movement rates of food (and thus of toxics) will influence exposure at all levels. Recently Schoenly and Cohen (1991) found large temporal variations in the 16 food webs they examined. This variation translates into different exposure scenarios for the component parts of the food webs.

The toxicologic data base on which HRA is based, is divided into experiments employing acute, sub-acute, sub-chronic, chronic, and lifetime exposure categories by a general consensus (Table 2). In a natural ecosystem, exposure is dependent on the duration of utilization of a particular prey species which

Table 2. Temporal categories of exposure for toxicologic studies (Klaassen 1991)

Rats

1 day 14 days 90 days 6-24 months 2 years

D B

C

A

I I I ' I A

0 ~ 20 40 60 80

/ Time (years)

Fig. 1. Recovery trajectories compared to baseline condition for a hypothetical ecosystem structural or functional measure reduced to 25% of baseline by an acute impact: (A) no recovery over 70 years; (B) slow but complete recovery; (C) rapid but incomplete recovery; (D) the resilient ecosystem with rapid and complete recovery.

may be sporadic (for a carnivore capturing a pesticide-contam- inated bird), subacute (for birds feeding on erupting caterpillar population), or subchronic (for zooplankton feeding on a sum- mer algal bloom). The scale of exposure varies not only as a function of duration, but as a percent of lifespan. For organisms with short life spans, cumulative effects may be negligible, yet if toxicity accelerates mortality, it may shift exposure from a predator (now deprived of viable prey) to decomposers, now faced with an unexpected food source (Figure 1).

This variation in temporal patterns of food webs means that even one exposure to a hazard causes reverberations through the system for several years as the different components of the system feed on the differentially exposed organisms. With a

Ecological Risk Assessment 487

chronic or repeated exposure, the pattern of exposure becomes even more complicated.

Thus, the methods to assess ecological risks to the structure and function of ecosystems will be more complicated than for HRA (O'Neill et al. 1982; Sheehan et al. 1984; Sheehan 1984a, 1984b, 1984c). New methodologies must be developed for combining the methods and indices ecologists use (National Research Council 1986; Sheehan et al. 1984) to create a risk analysis process that can be applied by ecologists, ecosystem and risk managers and policy-makers to provide on-going evaluation of hazards, and biomonitoring of the system to detect changes.

Hypothetical Case

A pond ecosystem embodies the bottom sediment, the air-water interface and the water. The biotic components include the primary producers (phytoplankton and green plants), and con- sumers such as zooplankton, small invertebrates, small fish, and large terminal predators such as fish, frogs, and turtles. In reality, the pond ecosystem also exchanges energy, nutrients, and organisms with the surrounding land and air. If a toxic is introduced at a particular season, effects will vary depending on the life stages present (eggs, young, larval stages or adults), size of the organisms, and activity (active or dormant). What- ever the initial level of direct exposure to the various organisms in that pond, the movement of the toxic through the system will vary because of the different temporal considerations such as life stages, lifespan and growth which determines who is eating what at any point in time. Since these three factors vary mark- edlyfor the different organisms, .rate of movement, bioaccumu- lation, bioamplifiction will vary in the different organisms (Laskowski 1991).

This will also lead to continued exposure of new organisms entering the system through birth or immigration. In temperate climates, contamination during winter when many species are dormant may have much less impact than during mid-spring when many populations are undergoing rapid expansion.

At one extreme there may be subtle effects on one or two species that either self-correct or quietly disappear, their functions assumed by other species. At the other extreme, disrup- tion of a critical food source (for example Daphnia in ponds or krill Euphausa in the sub-Antarctic Oceans) may have wide- spread ramifications for many other species as well.

The Challenge

Consensus is required, based on direct measurements in model ecosystems, for what constitutes appropriate temporal windows for evaluating both exposure and outcome. One window is for the ecosystem that is chronically exposed (i.e., continuously, every day). Another window is the one time, one day exposure followed by an echo as the agent is distributed through a food web or destroyed. How many windows in between are necessary? We suggest that attention to exposures lasting for one breeding season in animals would be important, but it is not clear what constitutes an appropriate study period for an ecosystem. This may vary from three months in the high Arctic to ten months in the tropics. Many tropical systems have at least two dry months of relative inactivity each year.

What are the appropriate scales for measuring effect? Clearly acute effects, what might be called the "fish-kill" outcome, are manifest within hours or days, but it is not clear how long one must follow an ecosystem to be confident that no adverse effect has occurred.

It seems logical to use some multiple of the life span of a long-lived species in the system. But we foresee this as a fertile field of debate. Maybe the much less tangible life span of a community is the critical window. If Ecological Risk Assessors conclude that we must study ecosystem outcomes over a period longer than a generation (or perhaps even longer than a lifespan), the public and risk managers can be expected to lose interest. But focusing on a short time frame will be nonsensical. A major challenge, therefore, is to identify a spectrum of mean- ingful latencies for various higher order ecological effects.

Temporal Issues for Management

Lastly, there is an important difference in risk management. HRA is established to set policies regarding a particular deci- sion which must be made "now". Regulators are concerned with short-term considerations or with population effects over a hypothetical 70 year life span. However, it is necessary to define a different set of time scales appropriate for ecological risk management. Figure 1 illustrates several alternative trajectories for recovery of ecosystems after some perturbation. In the worst case (curve A), there is no recovery. An ecosystem that self-corrects and regains its structure and function within 70 years (curve B) is likely to be acceptable to some, while others will consider such long disturbance intolerable. On the other hand, ecosystem resiliency, the tendency to self-correct after disturbance, may be apparent much sooner (curves C and D). Yet full recovery to a pre-disturbance state may never recur (curve C).

Final/y, the natural processes of succession may recreate a somewhat different, but recognizable assemblage of species over a scale of 1-5 centuries. This will have little public appeal, and will probably not impact on management decisions. Thus management decisions must define a time scale for which a degree of recovery is specified (e.g., 90% within 5 years). A separate consideration, not dealt with in this paper is the crite- rion for acceptable recovery. How much does the restored ecosystem have to resemble the predisturbance form (contrast curves C and D) to be declared a success? This thorny question is central to restoration ecology, and ultimately to ecological risk analysis.

Acknowledgments. We thank B. G. Goldstein, M. Gallo, and D. Wartenburg for discussions of risk over the years. This research was partially funded by grant ESO 5022 from the National Institute of Environmental Health Sciences.

References

Barnhouse LW (1992) Case studies in ecological risk assessment. Environ Sci Technol 26:230-231

Butler GG (1978) Principles of ecotoxicology, Wiley and Sons, Chich- ester, England

488 J. Burger and M. Gochfeld

Cairns J Jr (1990) The genesis of biomonitoring in aquatic ecosystems. Environ Profess 12:169 176

Dawson DA, Wilke TS (1991) Initial evaluation of developmental malformation as an endpoint in mixture toxicity hazard assessment for aquatic vertebrates. Ecotoxicol Environ Safety 21:215-226

Fischer GW, Morgan MG, Fischoff B, Nair I, Lave LB (1991) What risks are people concerned about. Risk Anal 11:303-314

Gibbons JW (1987) Why do turtles live so long? BioSci 37:262-269 Gibbons JW, Semlitsch R (1982) Survivorship and longevity of a

long-lived vertebrate species: how long do turtles live? J Animal Ecol 51:523-527

Gilpin M, Hanski I (eds) (1991) Metapopulation dynamics: empirical and theoretical investigations. Academic Press, London, England

Graham RL, Hunsaker CT, O'Neill RV, Jackson BL (1991) Ecological risk assessment at the regional scale. Ecol Applications 1:196-206

Hunsaker CT, Graham TL, Barnhouse LW, Gardner RH, O'Neill RV, Suter II GW (1990) Assessing ecological risk on a regional scale. Enviro Manage 14:325-332

Johnson ML, Huggins DG, DeNoyelles F Jr (191) Ecosystem model- ing with LISREL: a new approach for measuring direct and indi- rect effects. Ecol Applications 1:383-398

Karr JR (1991) Biological integrity: a long-neglected aspect of water resource management. Ecol Applications 1:66-84

Klaassen CD ( 1991) Principles of toxicology. In: Amdur MO, Doull J, Klaassen CD (eds). Casarett and Doull's Toxicology. McMillan, New York, NY

Laskowski R ( 1991) Are the top carnivores endangered by heavy metal biomagnification? Oikos 60:387-390

Miller DR (1984) Distinguishing ecotoxic effects. In: Sheehan PJ, Miller DR, Butler GC, Bordeau P (eds) Effects of pollutants at the ecosystem level. Wiley and Sons, Chichester, England, pp 7-14

National Research Council (1981) Testing for effects of chemicals on ecosystems. National Academy Press, Washington, DC

- - . ( 1 9 8 3 ) Risk assessment in the federal government: Managing the process. National Academy Press, Washington, DC

(1986) Ecological knowledge and environmental problem-solv- ing. National Academy Press, Washington, DC

O'Conner JS, Dewling RT (1986) Indices of marine degradation: Their utility. Environ Manage 10:335-343

Odum ET (1957) Fundamentals of ecology, 2nd ed. WB Saunders, Philadelphia, PA

Okkerman PC, Plarsche EJVD, Sloof W, VanLeenwen C J, Canton JH (1991) Ecotoxicological effects assessment: A comparison of several extrapolation procedures. Ecotoxicol Environ Safety 21:182- 193

O'Neill RV, Gardner RH, Bamhouse LW, Suter GW II, Hildebrand SG, Gehrs GW (1982) Ecosystem risk analysis: A new methodology. Environ Toxicol Chem 1:167-177

Paine RT (1966) Food web complexity and species diversity. Am Natur 100:65-75

Rall DP (1969) Difficulties in extrapolating the results of toxicity studies in laboratory animals to man. Environ Res 2:360-367

Schoenly K, Cohen JE (1991) Temporal variation in food web structure: 16 empirical cases. Ecol Monogr 6:267-298

Sheehan PJ (1984a) Effects of individuals and population. In: Sheehan PJ, Miller DR, Butler GC, Bourdeau P (eds) Effects of pollutants at the ecosystem level. Wiley and Sons, Chichester, England, pp 23-50

- - ( 1 9 8 4 b ) Effects on community and ecosystem structure and dynamics. In: Sheehan PJ, Miller DR, Butler GC, Bourdeau P (eds) Effects of pollutants at the ecosystem level. Wiley and Sons, Chichester, England, pp 51-100

- - ( 1 9 8 4 c ) Functional changes in ecosystems. In: Sheehan PJ, Miller DR, Butler GC, Bourdeau P (eds) Effects of pollutants at the ecosystem level. Wiley and Sons, Chichester, England, pp 101-146

Sheehan P J, Miller DR, Butler DR, Bourdeau P (eds) (1984) Effects of pollutants at the ecosystem level. Wiley and Sons, Chichester, England

Tinkle DW (1961) Geographic variation in reproduction, size, sex ratio, and maturity ofSternothaerus odoratus (Testudinata: Chely- dridae). Ecology 42:68-76.

Webber EC, Deutsch WG, Bayne DR, Seesock WG (1992) Ecosys- tem-level testing of a synthetic-pyrethoid insecticide in aquatic mesocosms. Environ Toxicol Chem 11:87-105

Manuscript received April 9, 1992 and in revised form June 19, 1992.

temporal scales in ecological risk assessment

Documents