Natural selection serves as the basis of adaptive evolution. Fundamentally, it is a biased movement of alleles between generations. Contrasting with genetic drift, or indeed mutation, it also a purely deterministic process, though nonetheless can cause stochastic change in the course of genetic hitchhiking. Those alleles displaying higher fitness are more likely to increase their representation in the next generation than those displaying lower fitness. Indeed, to an extent this is a tautology since we can define "fitness" as a function of that same potential to increase that representation. In this sub-section I provide brief discussions of various concepts of natural section that are relevant especially to an understanding of issues of microbial evolution. Some of these issues are discussed in greater depth in the chapter, "Natural selection".

Adaptive Evolution

Adaptive evolution is simply the accumulation of alleles that display positive fitness effects, that is, which are improvements in some manner on the benefits provided by the alleles they replace. Adaptations themselves can be physiological, anatomical, behavioral, etc., and always will have an underlying molecular basis. For our purposes, however, simply providing a measurable increase in organism fitness qualifies as an adaptation. Even within that simple concept there exist a number of variations, as I discuss.

Positive and Purifying Selection

Positive selection, a.k.a., directional selection, is a deterministic increase in the prevalence of an allele or alleles within a population. The resulting increase, towards or to fixation, can be thought of as moving in a consistent direction and positive selection thus also can be described as directional. The increasing fitness, in turn, is by definition adaptive, at least locally and in the short term, since it is a product of natural selection. Note however that there is a difference between the fitness of individuals and that of populations, where the latter is an average. Thus, positive selection can involve an increase in the overall fitness of a population, as a beneficial alleles increase in frequencies. At the same time, however, this does not necessarily imply that the highest fitness genotypes found within a population will display any higher a fitness over time to the extent that those individuals already contain all relevant beneficial alleles. The proportion of the population made up of those higher-fitness individuals, however, should increase as population average fitness increases.

As beneficial alleles rise towards fixation, detrimental alleles found at the same loci will decline towards extinction. Positive selection consequently also can be described as purifying, that is, increasing the "purity" of alleles found at a given locus. Polymorphisms, that is, can be thought of as "less pure" versus where only a single allele exists within a population at a given locus (fixation). If an allele becomes fixed due to deterministic evolution, i.e., as a consequence of natural selection acting on that allele, then the associated evolution can be described as adaptive.

Parallel and Convergent Evolution

The concepts of parallel and convergent evolution derive from macrobiological consideration particularly of organism morphology, such as in animals. The ideas, though, have at their root adaptive evolution, with organisms in these cases resembling each other not because of similarity due to common descent – though there is some of that as well – but instead because of similar adaptive paths that have been taken. The major difference is that with microorganisms these paths for the most part are seen in nucleic acid sequences rather than in overall organism morphology. As with a number of issues involved in microorganism similarity, however, there is an ongoing concern in making comparisons between organisms that one may be observing the consequences of horizontal gene transfer rather than consequence of vertical descent, or vice versa.

With these caveats in mind, note that parallel evolution means that two organisms started out similar and remained similar despite no longer sharing a gene pool. The implication is that adaptive changes in nucleic acid sequence are observed in both organisms and, to some extent, those changes are similar if not identical. Note, though, that in addition to concerns that horizontal gene transfer can provide the same signal, i.e., by replacing divergent sequence with similar sequence, it is also important to make sure that two sequences are not similar simply because in fact they are identical by descent, that is, unchanged from their sequence as found in the common ancestor. Thus, parallel evolution is interesting in concept but in practice difficult to definitively demonstrate. When one has both ancestor and descendant organisms "in hand", however, then observation of parallel evolution can serve as reasonably strong indication of adaptive evolution, such as may be observed over the course of well controlled experimental evolution studies.

While parallel evolution can be thought of as starting the same and then changing in equivalent ways, including displaying identical nucleotide substitutions, convergent evolution is one of starting out differently and then becoming more similar. In some ways this signal is easier to tease out because at least the possibility of identity due to horizontal gene transfer is much less of an issue. Alternatively, it is possible for two lineages to give the appearance of divergence if they started out the same but then changed in different ways, with the resulting similarities that which is retained rather than what is gained. Interestingly, however, retention of similarities too is a sign of natural selection, though not of positive selection (but instead is an example of stabilizing selection). True convergent evolution, as seen at the nucleotide sequence level, nonetheless can be indicative of positive selection, that is, for specific nucleotides, or amino acids, at equivalent locations within molecules that are not otherwise identical.

Hard versus Soft Selection

Hard and soft selection both involve positive/purifying selection at least potentially, that is, adaptive evolution. As also considered under the heading of positive selection, however, there is a difference between a single individual being fitter and the entire population being fitter. With hard versus soft selection, this situation is again seen, on the one hand with selection either resulting in populations displaying an increased absolute fitness (product of hard selection) or instead resulting in a population displaying no increase in absolute fitness. The difference has to do with whether individuals are adapting to non-conspecific environmental conditions (hard selection) or instead solely in terms of their competitiveness with other members of the same population (soft selection).

Hard selection, strictly, is an environmental condition that can have the potential of driving a population to extinction, and particular so if the population does not adapt. If extinction is averted, however, then the average fitness of the population, as well as that of individual members of populations, becomes greater in comparison to the average fitness of the population at the point where hard selection was imposed. This increase in fitness is explicitly a consequence of acquisition of whatever adaptation allowed the population to survive. This increased fitness, though, may be and indeed likely is limited only to the environmental conditions which imposed the hard selection to begin with.

Contrasting hard selection, soft selection is imposed by conspecifics. Basically, organisms can evolve to become better competitors within their own populations. Such greater competitive ability can result from adaptations that are not directly in response to conspecifics, but soft selection in particular is a consequence of adaptations that are in direct response to the actions of conspecifics. For example, an animal may become more attractive to the opposite gender. In the end, this is not going to directly impact the absolute fitness of the population, though certainly those "more attractive" animals obtain fitness benefits in terms of increased or otherwise qualitatively better mating opportunities relative to the fitness of their conspecifics. Note that such mechanisms are not things that have a potential to lead to population extinction if not addressed.

Periodic Selection, Selective Sweeps, and Hitchhiking

Natural selection is simply a biased movement of alleles through time, with some alleles thereby deterministically increasing in frequency while others deterministically decline. Periodic selection throws an additional twist on this otherwise rather simple, and elegant(!) idea. Like clonal interference and Muller's ratchet, periodic selection is a consequence especially of a relative lack of sex within populations, that is, it is seen particularly within clonal populations. While Muller's ratchet is a consequence of drift and clonal interference the result of inefficiencies in natural selection that "interfere" with allele fixation, periodic selection in fact is exactly what you expect to see from natural selection: fixation of beneficial alleles. It is more complicated than that, however. First, though beneficial alleles become fixed, it is actually the whole genotype which serves as the target of selection over multiple organism generations. Thus, just as clonal interference shows what can happen when a beneficial allele finds itself "trapped" within an otherwise mediocre genotype, periodic selection shows what happens to genotypes that are particularly fit.

So far this is not that out of the ordinary since a particularly fit genotype could very well be the consequence of containing a particularly beneficial allele, and indeed that is often the case. Because it is the genotype that is being selected, rather than just the beneficial allele, periodic selection however can lead to the selection also for detrimental alleles, that is, so-called genetic hitchhikers which are linked to beneficial alleles (just as in general mutators, alleles conferring higher mutation rates benefit by being linked to the beneficial alleles they generate). The end point of periodic selection, and also what distinguishes it from clonal interference, is that periodic selection is a mechanism that can result in allele fixation. Because it is the genotype that is being fixed, rather than individual alleles, hitchhiking alleles will be fixed as well.

If fixed alleles happen to be detrimental, then what one observes is the seeming paradox of natural selection selecting for declines in the frequency and even the extinction of beneficial alleles. Indeed, the result can be natural selection that is acting essentially in a manner that is similar to the action of genetic drift. To the extent that it is luck whether an allele finds itself hitchhiking along with a beneficial allele, however, then the process, for the hitchhiking allele, is indeed stochastic despite its association with a process involving natural selection. That is, while periodic selection is deterministic (that is, a product of natural selection), to a large extent genetic hitchhiking can be stochastic (that is, basically another form of genetic drift). Note that this scenario is essentially equivalent to the idea of a 'negative' hitchhiking where beneficial alleles are linked with sufficiently detrimental (low-fitness) alleles that the beneficial alleles are driven to extinction by the association. Sex thus can be viewed, at least in part, as an adaptation that increases the potential for beneficial alleles to survive in populations by explicitly unlinking those alleles from detrimental alleles with which they otherwise by necessity would have to share fates. This idea in turn is not so different from the perhaps more familiar one that sex can serve as a means of bringing beneficial alleles from multiple sources into the same individual or individuals.

Selective sweeps can be viewed either as equivalent to periodic selection or instead serving as a more general term which allows for at least some linkage equilibrium (that is, recombination associated with sex that separates alleles from each other). Periodic selection, that is, gives rise to selective sweeps, of varying effectiveness, as those sweeps can occur within clonal populations. Note that in either case, periodic selection or selective sweeps, what is being described is positive selection acting on populations, with fitter individuals outcompeting less-fit organisms. With selective sweeps it is simply potentially a smaller fraction of an organism's alleles that are determining fitness whereas with periodic selection it is all of an organism's alleles that are determining fitness.

Neutral Mutations

A neutral mutation is one that displays fitness benefits, or costs, relative to a wild-type allele, but ones that are sufficiently small that natural selection acting on that mutation is weak in comparison to genetic drift. More generally described as neutral alleles, such changes can be quite common, particularly in light of the degeneracy of the genetic code as well as the existence of non-coding and "junk" regions within genomes. Important to keep in mind is that while a given allele may be selectively neutral in one environment, the same will not necessarily be true for all possible environments. Thus, just as natural selection is dependent upon environmental context, so too is whether an allele is or is not selectively neutral.

Nonsynonymous to Synonymous Substitution Ratios

Recall that synonymous substitutions are mutations that represent changes in nucleotide sequence but not in the amino acid sequence of encoded polypeptides. To at least a first approximation these changes are neutral since, at a minimum, they result in no change in protein structure. Nonsynonymous substitutions, on the other hand, do result in changes in protein structure, though that too is a simplification since changes in protein structure that are the result of amino acid substitutions can vary in their degree of impact. Still, synonymous substitutions are more likely to be neutral. Non-synonymous substitutions, and particularly ones that have been subject to natural selection, e.g., such towards allele fixation, by contrast are more likely to be adaptive.

From these generalizations, we can infer the extent of adaptive evolution that has occurred in two lineages by determining the ratio of nonsynonymous to synonymous substitutions. Synonymous substitution rates provide an indication of mutation accumulation rates within lineages. Nonsynonymous substitution rates, by contrast, are suggestive of how much adaptive evolution has occurred. The ratio of nonsynonymous substitutions to synonymous ones thus gives an indication of evolution that is adaptive (higher ratios) versus more neutral (lower ratios). This evolution, in turn, are changes in allele frequencies that are presumably due to natural selection versus that presumably due to genetic drift, respectively.

Selectively Neutral Markers

Selectively neutral markers can be important to experimental evolution studies. They are alleles that in some manner are distinguishable from equivalent alleles found at the same locus or, alternatively, that represent additional genes carried by organisms. For example, antimicrobial resistance genes can serve as selectively neutral makers. The key is that these alleles make it relatively easy to tell different organisms apart such as with one displaying one color upon plating and the other a different color, or one successfully forming colonies versus not.

The other half of being selectively neutral is that these markers should not impact the fitness of the carrying organisms under experimental conditions. Note, though, that this criterion isn't entirely strict since it often is possible to measure organism fitness, in competition with tagged strains, using multiple competitions rather than just one. That is, with ancestral strains competing against the tagged strain along with, in separate assays, descendant strains competing against the same tagged strain. The result is that differences in fitness with the reference strain can be compared rather than the fitness between two strains directly compared during growth within the same culture. Strictly, though, such strains would not be described as neutrally marked, though they can serve a similar experimental purpose. In other words, it is possible to design analyses that demand not so much that genetic markers be neutral as that they at least impart a consistent impact on fitness going from assay to assay.

Divergent Evolution and Historical Contingency

Divergent evolution is a product of positive or directional selection, rather than a product of stabilizing selection, though it can also result from diversifying selection as well as genetic drift. Divergent evolution can result in the fixation of alleles, thereby eliminating polymorphisms, though more generally represents simply change in the genetic makeup of populations. Typically this change is adaptive and thereby presumably a product of natural selection, though random change also can be divergent. Divergent evolution is most readily observed when comparing two lineages that literally have diverged from a common ancestor, where divergent evolution simply is non-mutual change, that is, such that the two lineages come to resemble each other less.

Though divergent evolution represents change, it is important to keep in mind that there are limits to the pace with which organisms can change, particularly mutationally. That is, what an organism currently consists of genetically will always have some bearing on what that organism can change into, again genetically. What an organism currently consists of genetically, in turn, is always a function of what that organism consisted of genetically in the past. The future state of an organism thus is contingent on its present state which is contingent on its past state. As past states represent an organism’s history, the term historical contingency is used to describe the limitations that exist on organism evolution as imposed by current organism characteristics.

Plants cannot easily evolve into animals and this is due to historical contingency, and numerous adaptations are not available to specific microorganisms due to a combination of a lack of genetic sequence from which such adaptations could form and existing adaptations that are incompatible with various possible new adaptations. Strictly anaerobic processes, for example, cannot evolve in strictly aerobic organisms. Limitations due to historical contingencies, particularly due to a lack of necessary genetic sequences, can in many cases be overcome via horizontal gene transfer. This acquisition of new genetic material, particularly genetic material that has already stood the test of natural selection, though in a different organism, can allow for substantial increases in the potential for organisms to evolutionarily diverge, though only if those organisms are to some degree sexual rather than strictly clonal. An important example of such divergence is the evolution of otherwise benign microorganisms, such as commensal bacteria, into pathogenic organisms, such as bacterial pathogens, which in many cases occurs directly as a consequence of horizontal gene transfer. In the modern world an important aspect of such divergence is the acquisition of resistance to antibiotics.

As an aside, note that it is reasonable to argue that historical contingencies are more important the smaller the population size and therefore the more mutation-limited a population. An infinite population with an infinite amount of available mutational variation should be able to evolve anything that is physically, biochemically, and genetically possible anytime since all possible sequence combinations are available to the population. Alternatively, the evolution of very small populations tends to be mutationally limited, which can provide opportunities to, for example, general mutator alleles. From this perspective, one can view the impact of horizontal gene transfer on evolutionary innovation as one of reducing the impact of historical contingencies via an expansion of effective population sizes. Specifically, mutation rates are less limiting to evolution if the mutations occurring in more than one population can contribute to the genetic variation seen within a specific population, and that expanded access to genetic variation essentially is the consequence of horizontal gene transfer. Of course, even with horizontal gene transfer the size of "populations" mutationally contributing to evolution still is not infinite, so historical contingencies continue to exist, that is, since not all of sequence space can possibly be explored by a finite population. Nevertheless, historical contingencies explicitly may be overcome via horizontal gene transfer, and the resulting evolution explicitly can be viewed as a consequence of larger "population" sizes resulting in greater levels of variation upon which natural selection can act. Thus,

If genomes adapt successfully and freely enough, their idiosyncrasies perhaps can be ignored, despite or even because of the complexity of their mechanisms. However, if some traits cannot change or if limitations imposed by pleiotropy and other constraints, lack of mutations, or small population size greatly influence evolution by changing the tradeoff surface or preventing adaptation along it, we must take genetic details into account.