In the study of the effects of a toxic substance in a single species, we seek to understand how the toxic substance affects the physiological function of individuals and how this toxin affects the demography (birth and death rates) of the population. The effect of the toxic substance may also vary among populations of a species as well as among species. At the community level, we seek to understand how the presence of the toxin alters the relative and absolute abundances of the species present. For example, a disproportionate affect on predators might allow the increase of prey species.

Simple Measures of Toxicity

Some simple measures of toxicity use bioassays to measure death rates in order to quantify the effect of the toxin. Such measures are commonly known as LD50 and LC50. The LD50 is defined as the lethal dose at which 50% of the population if killed in a given period of time; an LC50 is the lethal concentration required to kill 50% of the population. The LC50 is a measure, e.g. in mg/l, of the concentration of the toxin whereas a dose is a more general term (need not be a concentration but may be a specific temperature, etc.). These bioassays involve subjecting several replicate groups of individuals to a range of concentrations (or doses) of a toxic compound and measuring the mortality after a defined time interval, e.g. 24 hours, 1 week, 1 month, etc. The data are then plotted and the LC50 is interpolated from the graph.

Bioassays such as LC50's have been criticized on a number of grounds. Such bioassays are not representative for the species as a whole as they are most often performed on a small subset of one population and then extrapolated to the species as a whole. There can be a wide range of tolerance to toxic agents among different populations of a species which should be taken into account. An often ignored component in such studies is the evolution of resistance to the toxic substance. A sudden release of a pollutant at a concentration that is toxic to most organisms is likely to have an immediate and drastic effect on the density and species distribution within an ecosystem. A slow build-up of pollutants in an ecosystem will more likely allow for an increase in resistance over time; changes in taxonomic composition in this case should be less severe.

Mortality, Fecundity, and Population Growth

Even though species differ substantially in life forms and stages of development, certain basic population processes are common to all; these four fundamental demographic parameters are birth, death, immigration, and emigration. These parameters can be combined in a simple algebraic equation describing change in population size between two points in time:

Nt+1 = Nt+B-D+I-E

where Nt is the present population size (measured as numbers of individuals), Nt+1 is the population size measured on time-period later, B, D, I, and E are the number of individuals which are born, die, immigrate into, or emigrate from the population between time interval t and t+1. Many models of population growth assume either no immigration or emigration, or that these two parameters cancel each other. These simplifications may not be valid and should be investigated in each species. The following account of population growth will utilize this convenient assumption.


Two sets of information are necessary for a demographic model of population growth: survival probabilities and fecundities for each age class in the population. The survival probabilities can be represented in two main ways: as lx values, probabilities of survival from birth to age x, or as px values, probabilities of survival from age x to age x+1. Fecundities, usually represented as mx values, are average numbers of female offspring per female of age x. Life tables and Leslie matrices are two approaches to modelling density-independent growth in age-structured populations. Age structure in a population implies that individuals of different ages will contribute to population growth differentially. A third approach is to use Euler's equation,

1 = e^(-rxlxmx)

and by iteration solve for r (for more details, see Begon et al., 1986 or Krebs 1978). All these approaches to predicting future population growth assume that the survival probabilities and fecundities are constant over time. We will use a hypothetical data set to illustrate the life table and Leslie matrix approaches to modelling population growth.

Life Table Method

Given an initial age distribution (Table 1.A), fecundities, and survival probabilities, population growth can be projected for any number of time periods. In this example it is projected for 9 generations using a software program called POPULUS. The intrinsic rate of increase of the population, r, may be estimated from the net reproductive rate Ro and the mean generation time G by:

r = ln(Ro)/G.

Ro is defined as the average number of female offspring per female during her entire lifespan:

Ro = (lxmx/lo),

and G is the average age at which those offspring are born,

G = (xlxmx/Ro).

The finite multiplication rate, lambda , is given by:

lambda = Nx+1/Nx

once the stable age distribution has been reached.

The intrinsic rate of population increase can also be computed as:

r = ln(lambda)

Leslie Matrix Approach

Given an initial age distribution (Table 1.B), a Leslie matrix can be built for the fecundities and survival probabilities, and population growth can be projected for any number of time periods by pre-multiplying the age distribution at each time period by the Leslie matrix to get the new age distribution for the next time period.

Table 1.A. POPULUS: Age-Structured Population Growth Projection

(input x, N, lx, and mx values and POPULUS gives the following projections)


 x N lx mx N(1) N(2) N(3) N(4) N(5) N(6) N(7) N(8) N(9)
 0  1600  1  0  1880  2600  3260  4273  5531  7197  9345  12146  15780
 1  100  0.1  5  160  188  256  326  427  553  720  935  1215
 2  60  0.06  10  60  96  113  154  196  256  332  432  561
 3  20  0.018  15  18  18  29  34  46  59  77  100  130
 4  0  0  0  0  0  0  0  0  0  0  0  0
 SUMS        2118  2862  3658  4786  6200  8066  10474  13612  17705

Ro = 1.58, est. r = lnRo/T=.2545, lambda = 1.3006

T=G=1.7975, r(max)= 0.2619 (solved by Euler's (Lotka's) equation)

Differences in Mortality and Fecundity Between Species and Populations

Environmental pollutants may have a variety of effects on the organisms found there. Another way to assertain the effects of a toxin is to measure differences in population growth rate (i.e. differences in survivorship and/or fecundity) in the species exposed to the toxin. Experiments comparing the toxicity of metal-rich sediments to population growth in the oligochaete, Limnodrilus hoffmeisteri (Klerks 1987, Klerks & Levinton 1992), were performed in glass beakers (10 cm diameter). Each beaker contained a 1 cm layer of sediment and 9 cm overlying water. At the start of the experiments, 10 adult worms were added. There were three replicates per group. Worms from South Cove and Foundry Cove were each grown in their own sediment and the opposite cove's sediment. Foundry cove sediment containing 55,300 µg/g Cd was used in all experiments; South Cove sediment contained approximately 19 µg Cd per g dry sediment. Ground fish food (0.05g Tetramin) was added to each beaker at the start of the experiment and after 14 days. The survival of worms and the number of newborn juveniles were recorded after 28 days. Additional sediment-toxicity experiments were run as described above, but using a range of Cd levels from Foundry Cove (540 - 55,300 µg/g).

Foundry and South Cove worms had similar survival when reared in South Cove sediment, whereas South Cove worms reared in Foundry Cove sediment (55,300 µg/g Cd) all died (ANOVA, Klerks 1987, Klerks & Levinton 1992). Foundry Cove worms reared in Foundry Cove sediments had high survival (90%). In the experiments comparing a range of Cd concentrations in Foundry Cove, worms from Foundry Cove and South Cove do not differ in survival or fecundity at the lowest Cd levels (<540 µg/g), but at high Cd levels (>22,400 µg/g) the two populations differ in both survival and fecundity (Klerks 1987). Although individual worms from both populations produced few or no offspring in sediments with high Cd concentrations, Foundry Cove worms produced more offspring than South Cove individuals at these elevated metal levels (p=0.05, borderline significance).

A series of similar experiments were performed on the dominant chironomid species, Tanypus neopunctipennis (Klerks 1987, Klerks & Levinton 1992). In experiment I, the survival of first instar larvae (for each sediment-cove combination) was monitored for 14 days. In experiment II, eggmasses were added to sediment from either South Cove or Foundry Cove and adult emergence was monitored for 6 weeks. In experiment III, third instar larvae were added to sediments and adult emergence was monitored for 21 days. Fish food was added to beakers as in the oligochaete sediment exposure experiments. No statistically significant differences in mortality of chironomids were detected between Foundry and South Cove populations in any of the experiments. Individuals of Tanypus neopunctipennis from Foundry Cove do not have a greater resistance to Cd than conspecifics from the control area. It is important to ask: why a difference in resistance in worms, but not in chironomids?

Acclimation and Resistance

Organisms may exhibit resistance to a pollutant by 2 means: physiological acclimation and genetic adaptation. Individuals may acquire tolerance by physiological acclimation during exposure to sublethal concentrations of the toxin at some prior time in the life-cycle; acclimation, however, does not confer tolerance in the offspring. Conversely, populations may evolve genetically based resistance through the action of natural selection on individually based variation in resistance. Offspring from such individuals will be toxin resistant even if born and raised in a clean environment.

To assess long-term changes as a result of pollution, it is important to distinguish if acclimation rather than adaptation is the cause of the increased resistance. The best way to distinguish between the two phenomena is to obtain offspring of resistant individuals which were born and raised in a clean environment. It is even better to use second generation offspring ("Grandchild Test") to diminish the influence of maternal effects.

Heritability: A Genetic Means of Understanding Resistance

Heritability refers to the degree of transmission of a trait from parent to offspring or the resemblance among relatives with regard to a given trait because relatives share, to varying degrees, their genes; they also share individual genetic deviations. Parents and offspring share exactly half their genes, full siblings share on average half their genes, and half siblings on average share one quarter of their genes. We can thus use the phenotypic resemblance between relatives to estimate heritabilities for traits. If appreciable heritabilities can be detected, the genetic basis of the trait is demonstrated. Heritability estimates for various traits in the field are often much lower than those estimated in laboratory cultures because environmental variation also affects the phenotypic expression of a trait; environmental variation is not controlled in field estimates.

Parent vs. Single or Multiple Offspring Comparisons

To estimate heritability in resistance to a toxin, we begin by raising offspring from many pairs of resistant parents in a clean environment and let them complete 2 generations. The following methods can be used to estimate heritabilities for any trait of interest, be it resistance to a toxin, tail length, beak width, body color, etc. Variation among relatives in a trait may be due to genetic and non-genetic components. The genetic variation that can be ascribed to differences among alleles is the so-called additive genetic variance, or VA. If we have a sample of parents and single or multiple offspring from each pair, we can compute the covariance among pairs of parent and offspring (COVP-O). This is directly related to the additive genetic variance (VA) by:

COVP-O = 1/2 VA

In practice, we estimate heritability directly as the slope of the linear regression of single offspring (Y) on single parent's (X) value. Heritability, symbolized as h2 is,

h2 = 2(slopeP-O) = 1/2 VA

Midparent vs. Single or Multiple Offspring

The midparent value is the mean of both parent's trait values, and the covariance is related to the genetic variance by:

COVMP-O = 1/2 VA,


but the regression is different because the variance of the midparent is actually the variance of a mean, since VMP = VP/2. Thus the heritability,

h2 = slopeMP-O

in this case. Therefore one can estimate heritability from data on one or both parents. One might want to do estimates on mother and father separately; differences between the two might say something about maternal effects or the sex linkage on loci affecting the trait. There are also other estimates of heritability using half sibs and full sibs (see Falconer 1981 for more details).

Heritability of Resistance of Limnodrilus hoffmeisteri

To determine if the increased Cd resistance of Foundry Cove worms had a genetic basis, resistance in laboratory-reared offspring was investigated. Worms from Foundry Cove were reared in pairs in clean sediment for two generations; offspring were regularly separated from the parents and reared in separate containers. Resistance to Cd was then compared among offspring and parents by determining survival times of worms in water spiked with 8.9µM Cd. The log of survival times of parent and offspring pairs were plotted and the slope of the regression line was used as an estimate of heritability (Klerks 1987, Klerks & Levinton 1989, Klerks & Levinton 1992). Two heritability estimates were obtained using parent-offspring (h2= 1.07±0.10) and midparent-offspring (h2= 0.93±0.12) regressions. Such high heritability estimates indicate that most of the variation for resistance to Cd consists of additive genetic variation.

Conclusions That Can Be Drawn

Changes in the taxonomic composition of a polluted area relative to a control (similar but unpolluted) area can never definitively be attributed to the pollution factor. Many environmental factors, as well as random events, influence the composition of the community. Demonstrated resistance to a specific toxic substance is indirect evidence for a negative influence of the substance on the ecosystem. If the resistance has a genetic basis, this implies that the toxin has killed the more sensitive individuals or reduced their reproductive output.