With many sorts of habitats and some entire ecosystems dwindling in extent, the extinctions of many species are imminent. Attempts at saving some of these species as ecologically functioning members of more or less natural communities, rather than zoo populations, involve the establishment of reserves whose extent is very modest in comparison to the original range of the species, and which, for that reason, can only maintain comparatively small relict populations. We wish, therefore, to estimate the viabilities of these small populations, and to learn what management measures and reserve-design features will enhance their viabilities. Furthermore, since there is inevitably pressure, in a crowded world, to encroach upon reserves, we should like to estimate the minimum extent of a reserve that will suffice to confer upon a population an expected time of extinction that is, by some criterion, acceptably remote. We shall scale this measure of reserve extent in units of population size. This is the minimum viable population problem.

At the most elementary level, the minimum viable population problem can be framed in demographic terms; but the magnitudes of the demographic variables will depend on a variety of factors, such as habitat quality, environmental variation, and genetic composition of the population (Shaffer, 1981; Soulé, 1980).

A fundamental fact of population genetics is that in closed populations (i.e., without immigration) the presence of only a small number of individuals, sustained over several generations, will lead to the depletion of genetic variation. Thus, the number of individuals is a crucial parameter in determining the amount of genetic variability that can be maintained in a population. This, in turn, influences the probability of long-term survival of a population because genetic variation is requisite for evolutionary adaptation to a changing environment. Thus, maintaining population numbers and genetic variation must be a central theme of plans for long-term population management.

In the last decade there have been several discussions of the role of population genetics in the management and conservation of threatened species (e.g., Soulé and Wilcox, 1980; Frankel and Soulé, 1981; Schonewald-Cox et al., 1983); such references provide a useful background for persons interested in this topic. Here we extend this treatment in four ways. First, we suggest criteria for the management of populations from a genetic perspective. We show how the effective size of a population, the pattern of natural selection, and rates of mutation interact to determine the amount and kinds of genetic variation maintained. Time scales associated with the different processes are also discussed.

The saving of critically endangered species is costly, and it is likely to conflict with other societal objectives. Methods are needed for clarifying and resolving such conflicts. In this chapter we will discuss an analytical tool called decision analysis (Raiffa, 1968). Decision analysis provides an explicit framework for identifying species in immediate danger of extinction, defining cases that may require intervention, evaluating the risks and benefits of alternate management strategies, and assessing whether or not the management efforts required to prevent a species' extinction can be justified in terms of their costs to society.

Why is an explicit framework needed? Conservation biology is essentially a crisis discipline (Soulé, 1985); neither time nor abundant economic resources are on its side. Difficult choices often must be made, usually in the absence of adequate data. When the outcomes of alternate actions are uncertain, it is hard to anticipate intuitively which one will be best. Furthermore, there are often several criteria for evaluating outcomes, such as minimizing costs versus maximizing protection; one action may seem to be best under the former criterion but a second far more desirable under the latter. Decision analysis provides a means of evaluating alternatives in a logical and repeatable manner; it is also a useful tool for communicating alternate management plans to others so that they can be persuaded to endorse one or more of them.

The bottom line

Administrators, policy makers, and managers have a right to ask for the bottom line — in this context, the bottom line is the MVP for a ‘typical’ vertebrate. And biologists have the right and sometimes the obligation not to give an oversimplified, misleading answer to such a question (Soulé, 1986). Nevertheless, I think that scientists owe it to the rest of society to provide rules of thumb, even when they know that sometimes the rules will be misunderstood and misused.

Let's rephrase the question: When taking into account all of the relevant factors mentioned by the authors of this book, what is the lowest MVP that one might expect for a vertebrate? Here, I am assuming a 95% expectation of persistence, without loss of fitness, for several centuries. My guess is that it would be in the low thousands. The bases for this order of magnitude number are theory and observation (empirical biogeography). Regarding observation, there isn't a lot of data, but it appears that populations with carrying capacities much smaller than this don't persist for very long, except, perhaps, in very constant environments, and even then will lose most of their variation.

Newmark (1986) has shown that the most consistent predictor of persistence of mammalian species in western US national parks is estimated population size at the time of establishment of the parks (averaging about 75 years ago).

Definition: How much is enough?

Given biblical precedence, it is not surprising that for millennia, a pair (male and female) has been deemed sufficient to initiate, if not perpetuate, a population. In fact, there is more than scriptural authority behind this myth. With luck, two can indeed be a sufficient number of founders.

What is luck? Without going into theories of randomness and probability, luck implies a fortunate or unusual circumstance leading to a good result. The result of interest in this book is the survival of a population in a state that maintains its vigor and its potential for evolutionary adaptation. Such a population is a viable population. Legend does not question Noah's success with each of his multitudinous experiments. He must have been very lucky indeed. He also had some advantages over us, not the least of which was a fresh, well-watered planet.

The problem that we address in this book is ‘How much is enough?’. Put more concretely, it is: What are the minimum conditions for the long-term persistence and adaptation of a species or population in a given place? This is one of the most difficult and challenging intellectual problems in conservation biology.

A population, regardless of its size, faces some real probability of becoming extinct by a ‘random walk’ through a series of inopportune events; this idea is not a new one (Goodman, Chapter 2). A number of mathematical models have been constructed to depict this probability of extinction over time. Most of the models have been based upon the intrinsic variation in the reproductive output and deaths of individuals in a species population (MacArthur and Wilson, 1967; MacArthur, 1972; Richter-Dyn and Goel, 1972; Hubbell, 1979; Feller, 1939). The intrinsic variation referred to in the models arises from random changes in birth and death rates and differences in these values that are inherent to individuals (genotypes) composing the population. Solutions to these models demonstrate small probabilities of population extinction for all but very small populations. More recently, it has been demonstrated that populations are much more susceptible to extinction if the environment (Leigh, 1975; 1981; Wright and Hubbell, 1983; Goodman, Chapter 2; Ginzburg et al., 1982; Roughgarden, 1979), including competitors (MacArthur, 1972), contributes to variations in birth and/or death rates.

Another type of extinction model does not rely on variations in birth and death rates. Rather, such models employ a reduction (‘catastrophe’) in population size that occurs at random intervals and perhaps intensities (Strebel, 1985; Hanson and Tuckwell, 1978; Ewens et al., Chapter 4).

There are two broad concepts of a minimum viable population (MVP) size. The first is a genetic concept, based on the rate at which genetic variation in a population is lost, and hence fitness decreased, through random genetic drift. The second is a demographic concept and is concerned with the probability of complete extinction of a population through random demographic forces. Although at an overall level these concepts are related, since inbreeding decreases fecundity and increases the death rate, present theory treats these as distinct concepts, since normal practice has been to assume the population size constant in defining and calculating the genetic MVP. For convenience, we also preserve this distinction in this chapter, and note that until a generalized theory covering both concepts is attempted, confusion may arise by the loose transfer of a numerical value of the MVP from the genetic to the demographic case, particularly the genetically derived values offered by Franklin (1980) and Soulé (1980).

For each of these two MVP concepts, the numerical value for the MVP eventually reached will depend on two assumptions. The first is the criterion chosen to define an MVP; for example, using the demographic concept, the size of the population which guarantees 95% probability of survival for y years clearly depends on the value chosen for y.

As the biosphere retreats in the face of physically superior forces, some conservation biologists are tempted to employ emotive rhetoric in the defense of ecological and species diversity, believing that such utterances will inspire others to join in their cause. Such a tactic may not be appropriate in a volume directed at managers and scholars, so let me only say that the subject of this book is central to conservation and conservation biology. It is also distinguished by its intellectual challenge.

The ‘viable population problem’ is very young. As documented in Chapter 1, it is only in the last decade or so that its importance has become recognized and its complexity appreciated. Herein we describe the significant advances that have already occurred. Our purpose is to spur increased interest in this aspect of conservation biology.

The logic of this volume is accretionary. Chapter 1 describes the viable population issue, examines its history, and warns of its complexity. In Chapter 2, Daniel Goodman provides the first of several major elements — a theory of persistence based on population dynamics, especially the interaction of environmental variability and the rate of population growth. He shows how it is possible to directly estimate the likelihood of persistence.

Chance extinctions: a growing concern

The Global 2000 Report to the President (1980) estimates that between 500 000 and 2 000 000 species — 15–20% of all species on Earth — could become extinct by the year 2000. Though direct exploitation and pollution will certainly be factors, the principal cause of this projected wave of extinctions is the continuing loss of wild habitats. Thus, habitat conservation and management will be the key elements in any program to minimize or reduce this expected diminution of the world's biotic diversity.

Given an expanding human population with rising economic expectations, competition for use of the world's remaining wildlands will be intense. Conservationists will often face the problem of determining just how little habitat a species can have and yet survive. At the same time, biologists are increasingly coming to recognize that extinction may often be the result of chance events and that the likelihood of extinction may increase dramatically as population size diminishes.

To see this, consider the life of an individual living thing. To fulfill life's potential, the individual must be conceived, go through an intricate developmental period, be born, further develop, mate, reproduce, and all the while survive. Anywhere along this continuum of events the individual may die.

Population extinction can result from many factors. Even though most cases are hidden from observers, its processes can be viewed in different ways. The other chapters of this book focus on single features of species biology that contribute to species extinction. Superficially, this chapter treats yet another factor: the extension of the spatial stage on which the extinction drama is played. Nonetheless, space is something different. It affects and is, in turn, affected by the other aspects of species biology that contribute to extinction.

The spatial extension of ecological systems, which considers the actual locations of organisms in the landscape, is not routinely incorporated into theoretical formulations of population genetics, demography, population dynamics, and community ecology. Our theories typically present variables such as N's and p's that summarize, with a single number, the ecological and genetical states of a system over some conceptually delimited region of physical space. That is, N is a count of all the animals in this space; it does not tell where they are, or how they are clumped or otherwise associated. Similarly, p represents a gene frequency in a ‘population’ of organisms, but the region over which this estimate is valid is not normally specified.

This reluctance to address questions of spatial extension results from at least two important considerations.

INTRODUCTION

In comparing the reactions of different aquatic organisms to toxic chemicals (whether they are inorganic in the form of heavy metals, ammonia, cyanides etc. or more complex organic pesticides), it has long been the aim of the toxicologists involved to attain some standardisation or uniformity in the experimental techniques they devise. This approach to unification of methods has been particularly marked in the case of freshwater fish, for which group there are now standard methods acceptable in most countries.

These standard methods lay down guidelines on the exact conditions of the tests vis-à-vis water quality, temperature, number and size of test organisms, duration of exposure to the chemical dilution and, finally, the criteria to be adopted when measuring the effect in terms of immobility or mortality.

The need for the same degree of uniformity in testing other forms of freshwater life, invertebrates in particular, though slow to develop has gained increasing momentum in the last 10 years or so. Much of this incentive has come from the US Environmental Protection Agency (EPA) which now plays an international role in the screening and clearance of all pesticides and other toxic chemicals likely to have a harmful effect on the environment. One of the first fruits of this increased interest in aquatic invertebrates appeared in 1972 in the form of an exhaustive summary – compiled by the EPA – of all published information on their reactions in the laboratory to a wide range of pesticides, mainly insecticides, and herbicides (NTIS, 1972).

INTRODUCTION

The use of herbicides to control undesirable plant growth first developed on a large scale shortly after World War II and has been extending rapidly ever since that time. With their increasing use in agriculture, forestry and water-way clearance particularly in developing countries these chemicals now rank alongside insecticides as major environmental contaminants (Balk & Koeman, 1984). The continuous monitoring programme of streams flowing into the Great Lakes over the last 10 years for example, has shown that herbicide use in agricultural land has now increased to such an extent that they now constitute more than half the total volume of pesticides used in agriculture (Frank et al., 1982). Even in the UK where there are unusually stringent regulations controlling pesticides in the environment – particularly with regard to natural water bodies – many of the long-established herbicides such as 2, 4-D, dalapon, dichlobenil and diquat have been cleared under the Pesticides Safety Precautions Scheme, 1973, for use as aquatic herbicides for control of submerged and emergent aquatic weeds, and for the control of vegetation along the banks of rivers and drainage channels (Ministry of Agriculture, Fisheries & Food, 1985).

Early recognition of possible effects on fish life of herbicides applied directly to water or contaminating water by run-off from agricultural land, led to very thorough laboratory investigations in the UK on fish toxicity, and established the relative lethal levels of about 20 common herbicides based on 24-h LC50 values (Alabaster, 1969).

At the time when I last carried out a review of the subject of pesticides and freshwater fauna (Muirhead-Thomson, 1971) it was still possible for a single author to do justice, within one book, to the information then available regarding all forms of freshwater animal life and all types of freshwater body. In the 15 years since that book was published, there has not only been an enormous proliferation of knowledge about this subject but also noteworthy changes in emphasis and priorities. There has been increasing specialisation within this general subject as well, making it increasingly difficult for a single author to encompass all aspects of this problem. For all these reasons, the scope of the present review is restricted to running waters, rivers and streams, and to the macroinvertebrate fauna of such water bodies. The restriction to macroinvertebrate fauna is dictated in part by the fact that a great deal of the voluminous literature in the last 15 years deals with studies on the reactions of freshwater fish, to such an extent that a competent review of that aspect, including all the physiological work on uptake and retention of pesticides by different organs, would require a separate volume. However, one aspect of those fish studies cannot be omitted from any review devoted to aquatic macroinvertebrates, that is the effect of pesticides and allied toxic chemicals on feeding habits of fish in so far as these are influenced by drastic changes in the availability of different invertebrate fish food organisms, as measured by changes in the composition of the stomach contents.

INTRODUCTION TO BLACKFLY CONTROL

Blackflies (Simuliidae) are biting flies which are widely distributed in both temperate and tropical regions. In some northern countries such as Canada their main economic importance is as biting and bloodsucking pests of humans and domestic stock. In other regions such as tropical Africa and Central America their main importance is in their role of vectors of human diseases such as onchocerciasis caused by a parasitic filarial worm. A feature common to all species of Simulium is their association with running waters which form the larval habitat. According to species and country, these habitats or breeding places may range from quite small trickling streams to very large rivers of Africa such as the Niger, the Zaire, the Volta and the Nile. In many of these rivers and larger streams the highest larval populations tend to be concentrated in the fast-flowing sections of turbulent water such as those associated with rapids and dam spillways.

Shortly after the discovery and rapid developments of DDT in the early 1940s, it was found that Simulium larvae are extremely sensitive to this insecticide and that, in some cases, effective larval control was still achieved many miles downstream from the point of application. This opened up entirely new possibilities for Simulium control on a large scale by effectively reducing or eradicating larval populations with insecticide. In one of these early and successful control operations DDT was applied at high dosages of 5–10 ppm, or even greater on occasions, which produced massive fish kills and drastic effects on other stream fauna.

THE SPRUCE BUDWORM IN CANADA AND THE US

Since the early 1950s the spruce budworm (Choristoneura fumiferana) has posed a serious threat in parts of eastern Canada, particularly New Brunswick, (Eidt, 1975, 1977; Symons, 1977a) and adjacent states of the US (Nash, Peterson & Chansler, 1971). In order to protect the valuable timber trees against defoliation, the method of control originally adopted was aerial spraying with DDT, which was practised from 1952 onwards. Since that time the intensity and extent of the infestation has increased. In New Brunswick for example, between 1952 and 1957 the sprayed area increased from 75 × 103 ha to 2.3 × 106 ha (8876 sq. miles). Many of the areas treated twice a year recorded a total application of 560 g/ha DDT per annum, and it was after such heavy treatment that Atlantic salmon, living in streams and rivers in the sprayed forest area, were found to be severely affected.

DDT began to be phased out in 1968, and by 19 70 was replaced completely by organophosphorus compounds, mainly fenitrothion. By 1976 the sprayed area in New Brunswick had increased to 4.0 × 106 ha (15000 sq. miles). By that year infestation had extended to other provinces, Quebec, Ontario and parts of Newfoundland and Nova Scotia up to a total area of 30 × 106 ha.

Fenitrothion continues to be the insecticide of choice, in Canada, and at the time of writing appears unlikely to be superseded by other insecticides (D. C. Eidt, personal communication).

The term ‘pesticide’ embraces a wide range of toxic chemicals used for controlling or eradicating undesirable forms of life. Compounds specifically designed for the control of insects and other arthropods, i.e. insecticides, make up the bulk of these; another range of pesticides is designed to deal with undesirable fish (both predatory and competitive) while still others were developed for use against the aquatic snails which harbour intermediate stages of human parasites. The term pesticide now conveniently includes herbicides, chemicals specifically designed for control of undesirable plant growth, and this inclusion recognises the fact that the greatly increased use of herbicides in recent years has, in many cases, placed these chemicals equal to or ahead of insecticides as major environmental contaminants (Balk & Koeman, 1984)

CONTAMINATION AS A DIRECT CONSEQUENCE OF PEST CONTROL OPERATIONS

DIRECT APPLICATION OF PESTICIDE TO WATERBODY

Pesticide contamination of running waters can occur in many different ways and from many different sources, and may be only of short duration or it may be prolonged. In view of the emphasis in this review on the problems of evaluation, it would be convenient to consider pesticide contamination under two main categories. In the first category would be listed all those cases where the presence of pesticide in running water is the direct consequence of control operations carried out against undesirable fauna or flora.