The Implications of Hormesis to Ecotoxicology and
Ecological Risk Assessment (ERA)

Peter M. Chapman, Ph.D.

EVS Environment Consultants

195 Pemberton Avenue

North Vancouver, B.C.

Canada V7P 2R4

Tel: 604-904-4005

Fax: 604-662-8548



Changes required for the explicit recognition of hormesis are outlined for both ecotoxicology and ecological risk assessment (ERA). A major research need is the extension of hormesis beyond chemical stressors to abiotic (e.g., habitat) and biotic stressors (e.g., species introductions, organism interactions). An overreaching research need is to determine for all stressors with model organisms, populations, and communities, whether hormesis has positive, neutral or adverse effects. The latter are the least likely, however neutral effects cannot be ruled out. Based on our present state of knowledge, hormesis is likely to have more of an impact on ecotoxicology than on ERA. In the case of the latter, it is most likely to make a difference only in a detailed level ecological risk assessment (DELRA), the most complex form of ERA. Further, for hormesis to be accepted fully into ecotoxicology or ERA will require a paradigm shift. Three on-going paradigm shifts to which hormesis could be linked are: recognition of the low utility of no-observed effects concentrations (NOECs); recognition of the need for special treatment of essential element dose/concentration-responses, which are similar to hormesis; and, the replacement of environmental toxicology with ecological toxicology (ecotoxicology).


Hormesis has had a long but not always distinguished history (Calabrese and Baldwin, 2000). It has too often been associated with homeopathy, marginalized, and/or ignored. Yet it is a very real phenomenon. Hormetic responses have been reported in hundreds of studies for a broad range of species (protozoa, bacteria, fungi, plants, invertebrates, and vertebrates including humans), biological endpoints (e.g., survival, growth, reproduction), and both inorganic and organic chemicals (Calabrese and Baldwin, 1997). However, perhaps because of its not always distinguished history, its potential roles in both ecological toxicology (ecotoxicology) and ecological risk assessment (ERA) remain to be determined. The purpose of this paper is to build on Chapman (1998, 1999) and explore potential future roles that hormesis could legitimately and usefully play in both ecotoxicology and ERA.


Hormesis - can be generally defined as a stimulatory effect that occurs when a substance which at high doses results in negative effects (e.g., growth or fecundity inhibition), produces positive effects at low doses (e.g., growth or fecundity stimulation). Typically, hormetic effects extend 30 to 60% above control levels, and over a 10-fold range below the no-observed-effects-concentration (NOEC). Hormesis has been hypothesized to be overcompensation to an alteration in homeostasis (Stebbing, 1998).

Toxicology - involves exposing an organism or biological system to a stressor and determining either a response (e.g., toxicity) and/or, in the case of a chemical, uptake of that chemical into biological tissues (bioaccumulation). Ecotoxicology is best defined in functional terms; it is distinguished from environmental toxicology by Chapman (2001a,b). Ecotoxicology involves laboratory or field testing that focuses on organisms important in the food chain/community structure and function. It is not restricted to individual species tests, and is not solely concerned with chemical effects but may also include consideration of potentially more important stressors such as habitat loss and species introductions. Testing is conducted by toxicologists but with appropriate input from ecologists and, in some cases, other specialized disciplines. For instance, bacterial toxicity testing, if truly ecotoxicological, would not be conducted without appropriate input from microbiologists.

Ecology - is a scientific discipline, focusing on: interactions between organisms; distributions and abundances of organisms; the functioning of biological populations and communities; and the processes that affect all of these parameters (Andrewartha and Birch, 1954). Initially, ecological studies were concerned with elucidating natural relationships; however, as anthropogenic influences on natural communities have increased, so have ecological studies expanded to include both natural and anthropogenic changes to populations and communities. Typically, ecology has been primarily concerned with low dose effects, while toxicology has been primarily concerned with high dose effects.

Ecological Risk Assessment (ERA) - is an iterative process, a framework providing a basis for eventual risk management. It typically involves three tiers: problem formulation or hazard assessment (initial planning and information gathering); effects and exposure assessment (data gathering and analysis); risk characterization (assimilation and integration). ERA differs from human health risk assessment (HHRA) based on the fact that it must consider a very large number of species rather than a single species (human beings). Further, whereas an HHRA aims to protect individuals, an ERA is not concerned with individuals or even with species, with the possible exception of endangered species, but rather with populations and with ecosystem process and function. An ERA can be performed to two levels of detail: a screening level ERA (SLERA) or a detailed level ERA (DLERA) (Hill et al., 2000, Table 1).

Table 1. Some Generic Differences: Screening Level Ecological Risk Assessment (SLERA) and Detailed Level Ecological Risk Assessment (DLERA)*



Relative level of effort

Low High

Level of conservation

High (over-protective) Decreased (reasonably protective)

Level of uncertainty

High Decreased

Hazard quotients (HQs)

Generic Site-specific


Broad Limited

*Adapted from Hill et al. (2000).


Hormesis is still not fully understood by far too many (eco)toxicologists. Most ignore it. Some who do not, nevertheless misunderstand it and ascribe anomalous concentration-response curves to hormesis when this is not the case. For example, it is well known that increasing water hardness (mg/L calcium carbonate) ameliorates the bioavailability and toxicity of most metals (CCME, 1999). Toxicity testing involving elevated metals concentrations in hard water can result in concentration-response curves that appear to be hormetic. There may be no toxicity at the highest test concentrations in very hard water. However, toxicity can appear at intermediate concentrations due to dilution effects. Dilution of the highest test concentrations will decrease not only metals concentrations but also water hardness. If water hardness is decreased such that it no longer ameliorates metals bioavailability, toxicity may occur. At lower concentrations, metals may be diluted sufficiently that, despite even lower water hardness, no toxicity will result. Hence, the appearance of a hormetic concentration-response curve can be provided by extrinsic factors (e.g., water hardness modifying toxicity), rather than by intrinsic factors (i.e., attempts to maintain homeostasis).

But there are also too many instances of actual hormesis that researchers are not seeing because they are not conducting tests with a sufficiently low range of concentrations. However, convincing researchers and toxicity testing laboratories to extend their testing range downward is difficult for two reasons. First, there is a long tradition in toxicology of focusing only on higher concentrations. Second, and much more important, adding additional test concentrations has economic (cost) consequences, which are currently difficult to justify. Only if hormesis is shown to have real significance that needs to be taken into account in risk and other assessments as mandated by regulatory agencies, will low dose testing become a standard feature of ecotoxicology.

To some extent, there is a circular logic involved here. For hormesis to become accepted, it needs to be demonstrated in tests by individual researchers and laboratories, not merely reported in literature summaries. However, for such demonstrations to become common, publication of hormetic curves in ecotoxicology papers must become more common. Yet hormetic curves are typically not published, and papers debating the importance and relevance of hormesis are not common in the ecological toxicity literature. One way to break this circle of inaction might be to develop a well-advertised web site where hormetic data can be entered and evaluated, and examples of different dose-response curves and advice on their interpretation, particularly in a regulatory context, can be obtained.

Biostatistical tools for dealing with hormesis also need to be developed, and existing tools need to be more widely used. It is not widely appreciated that Van Ewijk and Hoeskatra (1993) have published a method for determining EC50 values at low doses that incorporates hormesis. Hormesis has been shown to occur and has been evaluated by some authors using this equation (e.g., Muyssen and Janssen, 2001). Other methodology has been proposed by Bailer and Oris (1997). Unfortunately, most researchers and toxicity testing laboratories do not use either methodology, and too many are probably not aware it exists. The tendency is to use "off the shelf" methodology, approved by the regulatory agencies that mandate the vast majority of toxicity test data. Again, until this changes, hormesis will not be recognized in most toxicity test results, even if low enough concentrations are tested. Table 2 outlines the differences required to move from present ecotoxicology testing which does not explicitly recognize hormesis to future testing which does.

Table 2. Implications of Hormesis for Ecotoxicology.

No Allowance For Hormesis

Allowance Made For Hormesis

(Present Testing)

(Future Testing?)

High toxicity test exposures

Both high and low toxicity test exposures

Few exposure concentrations, regimented spacing

More exposure concentrations, appropriate statistical power and spacing

Pooling of controls and lower test concentrations possible

No pooling of controls or lower test concentrations possible

Range finders to pick test concentration ranges

Use of environmental fate models to pick test concentrations

Assumption of monotonic or linear data pattern; statistics used to force this pattern

No preconceived assumptions or confining statistics statistics used to force this pattern (e.g., application of quadratic model to generalized linear model [Kerr and Meador, 1996]; use of regression-type models allowing for hormesis [Van Ewijk and Hoekstra, 1993; Bailer and Oris, 1997])


Hormesis is basically simply a response to stress. The form of stress may make no difference to the hormetic response, at least at the individual level. For example, hormetic responses by humans are well documented in response to low levels of non-chemical stress (e.g., deadlines, presentations, etc.).

However, although chemical hormesis has been demonstrated, there is less evidence that physical and/or biological stressors can also cause hormesis. Such evidence is important as, arguably, these ecological (rather than toxicological) stressors are more important overall to the health and persistence of populations and communities than are chemical stressors (with the obvious exception of chemical "hot-spots") (Chapman, 1995).

There is some evidence that hormesis may not be just a generalized toxicological phenomenon, but also a generalized ecological phenomenon, for instance the intermediate disturbance hypothesis in ecology (Chapman, 1998, 1999) and other analogues (Gentile, 2000). The intermediate disturbance hypothesis proposed historically by Tansley and Adamson (1925) and more recently by Paine and Vadas (1969) describes both natural biotic (e.g., predation, competition) and physical stressors (e.g., fires, extreme weather events) as necessary to maintain ecosystems at optimum levels over large spatial and temporal scales. Specifically, species richness is predicted to be greatest following moderate levels of disturbance (O'Neill et al., 1986; Collins et al., 1995). However, the intermediate disturbance hypothesis does not appear to apply to all situations, possibly due to differential life-history features and responses to disturbance (Underwood, 2000).

Anthropogenic stress may also induce hormesis in natural populations. A good example comprises soil microbial communities exposed to heavy metal stress. Traditionally, it was assumed that a linear relationship existed: diversity decreased as stress increased. Giller et al. (1998) instead hypothesize, from the available literature, that low levels of stress result in the predominance of a few competitive species with consequent lack of diversity. As stress increases, these species lose their competitive advantage and diversity increases. At higher stress, progressive species extinctions occur, and diversity decreases.

Bartell (2000) addressed the question: "Are ecosystems hormetic?" He noted that ecosystem-level hormesis could involve key ecological processes such as nutrient cycling, usage of energy inputs, and stability. He argued that hormesis may well be occurring at the ecosystem level and offered examples including the following: a prediction for increased probability of increased phytoplankton production as a function of copper exposure for coastal marine ecosystems; increased species diversity resulting from habitat changes; increased fish species related to eutrophication. He believes there is enough evidence to more fully explore the possibility of ecosystem-level hormesis and its possible implications.

However, Gentile and van der Schalie (2000) point out the difficulties inherent in evaluating beneficial as opposed to harmful outcomes from hormesis for ecological systems. Specifically, they note that an increase in nutrients in a nutrient-poor lake may alter species composition (a negative) and also increase populations of sport fish (a positive).

Further, stimulation of a single life history trait, such as growth or fecundity, can at times be explained by a trade-off between traits (Stearns, 1989, 1992). A positive effect on one measured life history trait may in fact be at the expense of another life history trait, resulting in no net gain for the organism. It is not uncommon with daphnids or other animals for exposure to a toxicant concentration to increase clutch size (hormesis), but reproduction may be delayed or egg size reduced (Stearns, 1992; Ebert, 1993, 1994).

To incorporate hormesis into ecology and fully understand its significance (i.e., is it a significant positive attribute, neutral [irrelevant/trivial], or a significant negative attribute) requires knowledge of:

- Individual-level consequences of suborganism hormesis - including any trade-offs between life history traits

- Population-level consequences of individual hormesis -- including any trade-offs between organisms

- Processes (abiotic, biotic) regulating population size and health - which may be affected (positively, negatively or not at all) by hormesis

- Minimum viable population size (and generic constraints) - which must be maintained and which can be affected (again, positively, negatively or not at all) by hormesis.

Clearly, it is not possible to fully understand all of the above, in all cases, for all organisms. Thus actual data must be supplemented by modelling (e.g., Caswell, 2000).


Risk assessments are conducted in a tiered manner, with initial assessments (problem formulation, SLERA) tending to be most conservative and thus to have the highest level of uncertainty. Uncertainty is reduced in successive tiers. Gentile and van der Schalie (2000) note that "it is unlikely that hormesis is a critical factor in most ecological risk assessments, given the magnitude of other uncertainties inherent in the process." They further argue that default adjustments for hormetic effects on chemical exposure are unlikely to be practical.

If Gentile and van der Schalie (2000) are correct, hormesis is best incorporated into an ERA where realism will be highest, and where conservatism and uncertainty will be lowest. On this basis, hormesis most properly fits in the effects assessment stage of a DLERA. This limited role, at least at present, appears to fit given the present state of the science relative to hormesis. We know it exists, we know it is extensive, but we do not truly know its overall significance either generically or in specific situations. The "advantages" of hormesis are significant but not enormous. A 30 to 60% increase over control levels can be lost within the factor of two difference that is not unreasonable between toxicity tests on the same substance conducted by different laboratories (Chapman, 1995). And, toxicity test results showing hormetic responses may be trivial/irrelevant or not beneficial in the real world (Suter, 1993), or not similarly applicable to all species. The latter possibility requires identifying key organisms and life-stages and ensuring that hormetic responses related to stressors of potential concern have been correctly identified and also correctly interpreted.

Further, the possibility of countervailing energetic or other costs that could conceivably negate the "advantages" of hormesis cannot be totally dismissed even though it is likely that such costs, if present, may not be overwhelming. Hormesis appears to be an inherent response similar to adaptation, rather than a short-term stress response similar to acclimation. Adaptation is genetic, extends beyond the lifespan of the individual, and tends to occur without appreciable metabolic cost. In contrast, acclimation is a physiological/structural mechanism of gaining increased tolerance within the lifespan of the individual, and may well have appreciable metabolic cost.

The exact use of hormesis within the risk assessment paradigm will likely be similar to that for essential elements (Table 3). The hormesis dose-response curve has many similarities to that for essential elements - at low concentrations of chemicals there are clear benefits, at high concentrations there is toxicity (Figure 1). Gentile and van der Schalie (2000) note that "it may be difficult to represent hormesis by the single, quantitative numeric with associated uncertainty typical of regulatory standards, criteria, and benchmarks." However, in fact, hormesis effectively replaces uncertainty factors, as is the case for essentiality (Table 3). The lower effects level (e.g., NOEC/L or ECx) should not be lower than the hormetic range and should not enter the deficiency range.

Table 3. Implications of Hormesis for ERA. Similar Implications Apply to the Incorporation of Essentiality (e.g., essential elements) into ERA.


Present Focus Future Focus

Problem Formulation/Hazard Assessment

Few and generally high toxicity test exposures More exposures; both high and low exposure concentrations

Exposure Assessment

Upper exposure bounds Distribution of actual doses/concentrations from exposure (both upper and lower bounds possible)

Effects Assessment

Few and generally high toxicity test exposures More exposures; both high and low exposure concentrations

Assumption of monotonic/linear data pattern No preconceived assumptions linear data pattern (or confining statistics)

Risk Characterization

Application of safety factors No safety factors (if hormetic thresholds can be determined)

Effect of a stressor on a defined endpoint Net effect of a stressor on health

Upper uncertainty bounds Upper and lower uncertainty bounds

Likelihood of adverse effects Likelihood of adverse and beneficial effects

Linear, no threshold dose- response default assumptions No linear, no threshold dose-response default assumptions response default assumptions

Point estimates of risk possible Risk always characterized as a range

Risk Management

Removing all chemicals Determining appropriate amounts of some chemicals

Figure 1.

Dose-response Relationships for Synthetic Organic Chemicals (A), Essential Metals and Metalloids (B) and Nonessential Metals and Metalloids (C). From Chapman and Wang (2000).

As noted by Gentile and van der Schalie (2000), a major issue regarding the use of hormesis within ERA is decoupling toxicity from bioaccumulation for those substances with a major difference between modes of exposure for acute and chronic toxicity. For most substances and many organisms, chronic and acute toxicity both result primarily from water column exposures. However, for some substances such as the organic forms of mercury and selenium, PCBs, DDT, chronic exposure results primarily if not entirely from dietary uptake. And for some organisms (e.g., those burrowing into contaminated sediments), dietary uptake can be an extremely important uptake route not only for these, but also for other substances. This issue must be addressed in ERA, specifically by identifying stressor-response pathways that are most likely to be significantly affected by hormesis and those (e.g., dietary uptake) for which hormetic responses are inappropriate. As Gentile and van der Schalie (2000) point out, it would be inappropriate to apply a toxicologically derived lower effects level that allowed bioaccumulation to levels of concern.

Because of the issues discussed above, Gentile and van der Schalie (2000) conclude that "hormesis will likely not be a very important factor in most ecological risk assessments." As is the case with most generalizations, this will not always be true, and there will be instances where hormetic effects result in a shift in "bright lines" that more appropriately reflect the balance between over- and under-protection. The true need is to focus ERA efforts on those instances where this is truly the case and not waste time or resources on trivial cases, which will solely denigrate hormesis as a general and acceptable scientific principle that is useful for both ecotoxicology and ERA.

An alternative, general DLERA scheme for hormetic substances could involve:

· Estimating the exposure range of selected organisms to stressors of potential concern (SOPCs) in appropriate habitats/communities.

· Establishing hormetic dose/concentration-response curves.

· Determining the extent and range of hormesis to the selected organisms.

· Determining whether hormesis has a positive, neutral or negative effect on the overall health of both individual organisms and populations/communities.

· Factoring this information into the risk characterization.


Incorporation of hormesis into ecotoxicology and ERA requires a paradigm shift. Given that the "advantages" of hormesis are not large, a paradigm shift based on hormesis alone appears unlikely. However, linking hormesis to other phenomena in a "combined paradigm shift" offers good possibilities for moving forward. There are at least three different such phenomena to which hormesis could be linked.

First, a philosophical change in toxicology endpoints is underway. The utility of no-observed-effect-concentrations/levels (NOEC/Ls) is in question in scientific and regulatory arenas, for reasons explained by Chapman et al. (1996). There is strong pressure to move philosophically from the idea of no effect, to the idea of a limited level of effect (e.g., a set effects concentration, or ECx, or a no-effects-concentration [NEC] that does not need to equal zero - Kooijmann [1996]). Thus the International Standards Organization Working Group ISO TC147/SC5 - Water Quality, Biological Methods recently agreed that the NOEC will be phased out of standards (September 18-24, 2000 meeting, Antalya, Turkey). This philosophical shift mirrors that engendered by hormesis in that total elimination of chemicals is not sought, but rather the focus is on their decrease to levels where they are not an issue.

Second, increasing recognition that essential elements cannot be treated similarly to non-essential elements is also resulting in a philosophical change. It is increasingly being recognized that, for some chemicals, toxicity occurs not only with excess but also with deficiency. The regulatory implication is that lower effects bounds, be they NOEC or ECx values, cannot be set below deficiency levels. Similarly, the lower bounds for hormetic substances should not be set below hormetic levels. This lower bounding of effects values has an important consequence: if the lower bounds are known, there is no need for uncertainty (i.e., safety) factors, instead the boundaries can be set much more precisely and without concern for over- or under-conservatism (Chapman et al., 1998).

Third, there is a continuing shift in the field of toxicology from reductionist to more holistic approaches. As noted by Chapman (2001a,b), environmental toxicology is giving way to ecological toxicology. Rather than toxicologists being primarily concerned with getting toxicity tests done according to preset formulae, the emphasis is shifting to ensuring that the tests are as ecologically relevant as possible. Hormesis is a demonstrable, inherent phenomenon common to a very wide variety of organisms and endpoints. Whereas it can be ignored by environmental toxicology, it cannot be ignored by ecotoxicology.


The implications of hormesis for both ecotoxicology and environmental risk assessment are potentially profound: present techniques used to estimate the no-effect-concentration (NEC) may not be beneficial to biota or the ecosystem. Effectively, some level of stress is good and the current emphasis on zero discharge of any chemicals may well be less than beneficial to at least some if not many organisms. However, the profundity of these implications requires non-trivial changes in both our philosophical and methodological approaches to both ecotoxicology and ecological risk assessment. This will not be an easy process. We have only begun on a path that may or may not lead where we think it does, but which in any case cannot help be both interesting and stimulating in our continuing search, as scientists, for "truth."


I thank Ed Calabrese for encouraging me to write this paper. Final word processing was done by Jackie Gelling (EVS Environment Consultants).


Andrewartha HG, Birch LC. The Distribution and Abundance of Animals. University of Chicago Press, Chicago, IL, 1954.

Bailer AJ, Oris JT. Estimating inhibition concentrations for different response scales using generalized linear models. Environmental Toxicology and Chemistry 1997; 16: 1554-1559.

Bartell SM. Are ecosystems hormetic? Human and Ecological Risk Assessment 2000; 6:237-243.

Calabrese EJ, Baldwin LA. The dose determines the stimulation (and poison): Development of a chemical hormesis database. International Journal of Toxicology 1997; 16:545-559

Calabrese EJ, Baldwin LA. History of chemical hormesis. Human and Experimental Toxicology 2000; 19: 2-31.

Caswell H. Matrix Population Models, 2nd Edition. Sinauer Associates, Sunderland, MA, 2000.

CCME. Canadian Environmental Quality Guidelines. Canadian Council of Ministers of the Environment, Winnipeg, MN, 1999.

Chapman, PM. Ecotoxicology and pollution - key issues. Marine Pollution Bulletin 1995; 31:167-177.

Chapman PM. New and emerging issues in ecotoxicology - The shape of testing to come? Australasian Journal of Ecotoxicology 1998; 4:1-7.

Chapman PM. Whole effluent toxicity testing: usefulness, level of protection, and risk assessment. Environmental Toxicology and Chemistry 1999; 19:3-13.

Chapman PM. Putting the ecology back into toxicology. Canadian Fisheries and Marine Sciences Technical Report 2001a (in press).

Chapman PM. Integrating toxicology and benthic ecology: Putting the eco back into ecotoxicology. Marine Pollution Bulletin 2001b (in press).

Chapman PM, Wang F. Issues in ecological risk assessment of metals and metalloids. Human and Ecological Risk Assessment 2000; 6 (965-988).

Chapman PM., Cardwell RS, Chapman PF. A warning: NOECs are inappropriate for regulatory use. Environmental Toxicology and Chemistry 1996: 15:77-79.

Chapman PM, Fairbrother A, Brown D. A critical evaluation of safety (uncertainty) factors for ecological risk assessment. Environmental Toxicology and Chemistry 1998; 17:99-108.

Collins SL, Glenn SM, Gibson DJ. Experimental analysis of intermediate disturbance and initial floristic composition: decoupling cause and effect. Ecology 1995; 76: 486-492.

Ebert D. A maturation size threshold and phenotypic plasticity of age and size at maturity in Daphnia magna. Oikos 1994; 69:309-317.

Gentile JH. Hormesis in ecology and ecological assessments. Human and Ecological Risk Assessment 2000; 6:223-226.

Gentile JH, van der Schalie WH. Hormesis and ecological risk assessment: fact or fantasy? Human and Ecological Risk Assessment 2000; 6:227-236.

Giller KE, Witter E, McGrath SP. Toxicity of heavy metals to microorganisms and microbial processes in agricultural soils: A review. Soil Biology and Biochemistry 1998; 30:1389-1414.

Hill RA, Chapman PM, Mann GS, Lawrence GS. Level of detail in ecological risk assessment. Marine Pollution Bulletin 2000; 40:471-477.

Kerr DR, Meador JP. Modelling dose response using generalized linear models. Environmental Toxicology and Chemistry 1996; 15:395-401.

Kooijmann SALM. An alternative for NOEC exists, but the standard model has to be abandoned first. Oikos 1996; 75:310-316.

Muyssen BTA, Janssen CR. Zinc acclimation and its effect on the zinc tolerance of Raphidocelis subcapitata and Chlorella vulgaris in laboratory experiments. Chemosphere 2001; in press.

O'Neill RV, DeAngelis DL, Waide JB, Allen TFH. A Hierarchical Concept of Ecosystems. Princeton University Press, Princeton, NJ, 1986.

Paine RT, Vadas RL. The effects of grazing by sea urchins, Strongylocentrotus spp. on benthic algal populations. Limnology and Oceanography 1969; 14:710-719.

Stearns SC. Trade-offs in life history evolution. Functional Ecology 1989; 3:259-268.

Stearns SC. The Evolution of Life Histories. Oxford University Press, NY, 1992.

Stebbing ARD. A theory for growth hormesis. Mutation Research 1998; 403:249-258.

Suter GW II. Ecological Risk Assessment. Lewis Publishers, Chelsea, MI, 1993.

Tansley AG, Adamson RS. Studies on the vegetation of the English chalk. III. The chalk grasslands of the Hampshire-Sussex border. Journal of Ecology 1925; 13:177-223.

Underwood AJ. Experimental ecology of rocky intertidal habitats: What are we learning? Journal of Experimental Marine Biology and Ecology 2000; 250:51-76.

Van Ewijk PH, Hoekstra JA. Calculation of the EC50 and its confidence interval when subtoxic stimulus is present. Ecotoxicology and Environmental Safety 1993; 25:25-32.