The environment in which an organism lives affects its relative reproductive success. As far as we know, all natural populations have natural variation in characters, and some component of that variation is heritable. Individuals that are relatively well adapted to their current environmental conditions are likely to pass on relatively more of their genes to the next generation than those in the population that are not so well adapted. Thus certain parts of the gene pool of the population are increased relative to others: the gene pool therefore changes, which is another way of saying that evolution takes place. The process by which the interaction of individual organisms with their environment is translated into changes in the underlying genetic structure of the population is called natural selection. In terms of the individual organisms, those that have their genes perpetuated or even increased in frequency in succeeding generations are judged to be fitter than others, and the fitness of an individual (measured as gene frequency in absolute terms, or relative terms in comparison with the whole population) is a direct measure of its selective or evolutionary success.

The physical, biochemical, neural, or physiological features that make up the morphology and behavior of the organism are presumed to be formed under direct or indirect genetic influence, and it is these characters that are partly molded by the environment, and interact directly with it, that are acted upon by selection.

In the classic Darwinian view of evolution, natural selection on average acts to favor those individuals with the packages of characters that would normally increase fitness. (Random disasters that strike individuals irrespective of the characters they possess do not affect this argument - thus the question "Bad luck or bad genes?" is not important).

At least in a sexually reproducing population, different combinations of the various characters coded by the gene pool are continually being produced by genetic recombination, and novelties in the individual characters are continually being produced by mutation. So there is natural variation in the development of individual characters, and in the frequency of the combinations of characters that make up individuals. The average state of any character is continually tested by natural selection, based on its effectiveness in increasing fitness. Its effectiveness will be some combination of its own intrinsic value to the organism, and its value as packaged with many other characters, perhaps in an organ system. In any normally large population, favorable combinations of characters increase in frequency, and the average state of any one character is expected to change over time, always in the direction that will give increased fitness to the organism as a whole. In an unchanging environment, characters should gradually be improved until they cannot be improved any more (for reasons that could be genetic or environmental, or associated with the way they meld with others in the organism). In this way, natural selection promotes favorable novelties that are introduced by mutation and juggled by recombination, and since it also rejects unsuitable combinations of characters, it stabilizes morphology toward improved combinations.


The concept of an adaptive landscape is useful to illustrate the point. The best possible set of characters for the environment can be drawn as a set of peaks separated by valleys. The actual set of characters in the population is represented as a point somewhere on the landscape. Selection would drive the characters of the population always uphill, until the population had gone as high as it could: this might not be the highest peak on the landscape, however: that might be occupied by another species.

As the environment fluctuates, the optimal set of characters to deal with it may change: thus the adaptive landscape is dynamic. Peaks may become relatively higher or lower, or even fall below adjacent points. Natural selection will act then to drive the characters of the real populations to new peaks.


In this classical view, selection at the level of the individual drives the process. But individual fit-ness, which is measured by the representation of the individual's genes in succeeding generations, can be increased by the success of related individuals. Since, for example, I share half my genes with my sister, I also share one-fourth of my genes with any of her children. Thus my fitness is not measured simply by my own children (each of whom carries approximately half my genes). My inclusive fitness is measured by the sum of my genes carried by all my relatives: my own children, my nieces and nephews, my cousins' children, and so on. Note that we are still talking about (my) individual success in having my genes represented in the next generation. My genes could be transmitted into the next generation even if I have no children of my own: I could act to help along the rearing of my nephews and nieces, for example, so that they will be more likely to pass along their (and some of my) genes. Although my family is being discussed, there is no selection on the family as such, merely on the individuals in it. Just as inclusive fitness is a more accurate measure of my reproductive sucess, so kin selection is a more accurate way of explaining how some animals act as they do. For example, worker bees are sterile, and so have zero fitness themselves: but they are daughters of the queen bee, sisters and brothers of her fertile offspring, and so their inclusive fitness is likely to be improved if they work toward the successful rearing of those individuals. Bee biology can only be understood by the concept of kin selection, an expanded version of natural selection.


Obviously, there are differences in absolute and relative fitness in any population, and there is some sort of distribution of relative fitness among the individuals. If a new selective pressure begins to act on a population, it prunes off the relatively less fit so severely that the average fitness increases for a while, before reaching a maximum level. This phenomenon is familiar to all artificial breeders of plants or animals. It would suggest, other things being equal, that natural selection would always act to drive fitness to a peak. But it simplifies reality in the sense that it takes no account of the changes in fitness that might result from changes in ecological parameters such as population size. There are alternative methods for judging whether a population has reached some sort of peak of adaptive fitness.

For example, on some small islands it is easy to imagine that food supply would depend on grazing intensity. So, for example, if the population size is low, the fitness of all reproductive members of the population might be high: as numbers increase accordingly, absolute levels of fit-ness would decline. There is likely to be an optimum population level, and theoretical calculations show that in these circumstances, the body size should evolve to result in the maximum carrying capacity of the habitat. For example, island mammal and bird populations characteristically have body sizes that are often significantly large or smaller than their mainland equivalents.

The concept of the ESS, or evolutionarily stable strategy, is valuable. One simply asks whether there are alternative characters, of behavior as well as anatomy, that could possibly be competitively superior to those possessed by the current population. In the island population, for example, larger or smaller body size might result in decreased fitness, often for complex reasons. A behavior pattern that seems intuitively poor may turn out to be the best available. For example, some tropical American songbirds (icterids) routinely have cowbirds lay eggs in their nests. The growing cowbird chick usually destroys all but one of the resident chicks, while the foster-parents feed the nestlings indiscriminately. It would seem that a behavior that would exclude cowbird eggs would be superior, increasing the fitness of the icterids. However, icterids have fly parasites, which the cowbird chick eats: if there is no cowbird in the nest, the icterid family suffers so much from the parasites that they lose, on average, more chicks.

Thus the tolerance of at least 50% nestling mortality from cowbirds is actually an evolutionary stable strategy, in the sense that it's the best the birds can do: it's better than the higher mortality values caused by the parasites. In some parts of their range, however, icterids have taken to nesting near colonies of wasps or bees, which keep the flies away: those icterids do not tolerate cowbirds, or their eggs. A new behavior has allowed them to change the ESS for a greater payoff in fitness.

Sometimes there are two equally valid ESS and so the population is dimorphic in some character or other. Among some deer, and some birds, male mating behavior consists of close escort of a group of females: there are costs in defending the females and/or their territory, however, and only the largest, strongest males can participate in mating under these circumstances. An alternative strategy for weaker or smaller males is to stage sneak raids on females: fitness may be low, but it is greater than zero, and the energy cost is low. The fact that two radically different mating styles exist suggests that they form ESS of equivalent value.

The concept of an ESS provides a rationale for the stability of a species over time: unless the considerations that led to the establishment of the ESS change, there will be selection for stasis. On the other hand, if those considerations change, either because mutation or recombination introduce new characters into the population, or because perturbations in the environment alter the ESS, then selection will act to induce evolutionary change. Don't forget that selection is happening all the time, whether or not the characters of the population or the species are changing.


The classical view of natural selection (and its expanded form, kin selection) as the driving force of evolution has been diluted somewhat in recent years. Some aspects of the evolutionary process may have a non-selective component: but controversy rages over the relative importance of such non-selective agents.

If, for example, a population is small, especially if it is genetically isolated from the rest of its species, characters coded in its gene pool may be only a small, perhaps biased sample of the characters in the species as a whole. An example in nature might be a small flock of birds blown off course in a storm, arriving on a new island (the founder effect). While the population remains small, the characters may vary widely because of chance events: recombinations in the small population, births and deaths that accentuate or lose significant characters (genetic drift). Such effects are known among small island populations of humans, or among small tribes or religious groups which have separated themselves from other humans by mating solely within the group. It's even possible that the characters of an isolated population will plot far enough away from the parent species on the adaptive landscape that selection will drive them toward a different peak. In this way, chance events that isolated the population may act to begin the process of divergent selection that eventually results in the formation of a new species.

New species could certainly form this way: it cannot be a coincidence that so many of the adaptive radiations we know about have happened among small populations scattered among islands, for example, among the honey creepers and the fruitflies of Hawaii, and the finches and the tortoises of the Galapagos. The controversy deals with the question whether this process, depending largely on chance events of geography or history within a species, is the dominant mode of forming new species. If it is, and if natural selection is not the driving force behind the process of speciation, then we have to think of the formation of new species as analogous to the formation of mutations within the genome of an individual, and we may have to think of some sort of species selection on those variants. However, the question need not be framed in such an "either-or" style: the real question is whether natural selection acts as well as chance events. If it does, then the analogy of speciation events with mutations fails, and we don't need to invoke any higher level selection.

There are four components to the view that the process of speciation is based primarily on the chance events that separate off isolated populations. First of all, it is argued, it's only in small isolated populations that the processes of the founder effect and of genetic drift make it possible for changes to take place on the adaptive landscape. Otherwise, the conservative, stabilizing effects of natural selection make it very difficult for a species to move away from its position on a height in the adaptive landscape, to cross an "adaptive valley" to come to a new equilibrium as a new species on another adaptive peak. Speciation, then, simply does not take place in the normal geographic and ecological range of an existing species. There is no evidence that this argument is correct. We know practically nothing about rates of gene flow in natural populations, or about the constraints that gene flow may apply to genetic divergence (Slatkin 1986).

Second, certain species may have aspects of their biology that makes it easy for them to split off isolated populations would be more likely to form new species. Those species tend to evolve into large clades, not because they are necessarily any better adapted than other species, but simply because they are more likely to speciate. There's no question that some kinds of organisms are more speciose than others: fruit flies and beetles are more speciose groups than dragonflies, and bats are more speciose than bears. Thus species selection is an evolutionary process acting at a higher level than individual selection, and independent of it. The counter-argument is "So what if some organisms form large clades? What does that have to do with the way selection works? How can one say from that that natural selection is not still a dominant process?" We're looking at an effect, not a process in itself.

Third, if all this is correct, then the process of forming a new species is comparatively rare and comparatively rapid, and it occurs among a small population in a very restricted area. It is thus extremely unlikely that we shall see speciation events occurring in the fossil record. Instead, the fossil record is much more likely to contain individuals that lived and died during the much longer time in which the species lay on its adaptive peak, with a stabilized morphology and a limited range of variability. In the fossil record, the morphology of the species would be expected to show stability or stasis, compared with the amount of morphological change to be expected during a speciation event. The fossil record of a set of related species is therefore likely to show punctuated changes in morphology between the species, the changes representing the gaps in the record when reorganization of the founder population of a new species was occurring somewhere else. Furthermore, the timing of speciation events is likely to be unpredictable, as long as the environmental effects that promote speciation do not run in predictable cycles.

Fourth, if there is general conservatism and stability of characters during the normal existence of a species, and if speciation events are the only times when significant reorganization takes place, there are drastic consequences for our view of natural selection. Individual selection is then almost entirely a conservative process: speciation is the process that introduces genuine novelty in evolution. As we have seen, this view depends on very dubious assertions that have very little basis in real observations. Therefore this view is not valid, at least as a general theory.

Return to Evolution topics

Return to Geology Department home page