We can differentiation selection acting on microorganisms in terms of the evolution of resistance into three basic categories. These are selective benefits, selection for increased integration with the carrier's metabolism (i.e., reduced costs of resistance), and that resistance often is an only episodically useful property. The first of these is the primary issue when considering directional selection/adaptive evolution. The second is an issue we will return to often, that is, the ongoing integration of especially new sequences and gene products into an organism's existing metabolism. The latter gives rise to issues of control of gene expression as well as selection, especially on the genes themselves, for mobility.

These considerations are consistent themes in this text and things that one should always keep in mind when considering adaptations: What are the benefits? What are the costs and how may those costs be reduced? To what extent might the underlying genes be mobilized? One in addition can consider interactions between these different facets of adaptation, which often can be described in terms of what are known as tradeoffs. That is, do increases in benefits in one respect result increases in costs in others? Do decreases in costs give rise to decreases in benefits? And do these changes in the fitness associated with an adaptation impact its potential to be transferred horizontally to other organisms? These types of questions, as noted, are general themes of this text.

Selection for Resistance

In many ways the question of why be resistant is quite simple: The phenomena to which resistance is selected otherwise reduce organism fitness in some manner. By displaying resistance to these costly phenomena, associated reductions in fitness can be either partially or fully avoided. Thus, for example, a bacterium could replicate while in the presence of an otherwise debilitating antibiotic. In fact, the impact of a phenomenon to which resistance is manifest can be said to be impacting a population in terms of hard selection: Something that can reduce microorganism fitness in absolute terms. Evolution of resistance thus results in an increase in the absolute fitness of the organisms involved, at least so long as the antagonistic phenomenon persists (note, though, that resistance to the actions of conspecifics as would instead be associated with soft selection is possible as well). An alternative perspective is that resistance will not evolve unless the consequence of resistance is an increase in absolute fitness. That is, lack of increases in fitness, despite a display of resistance, can be due to the mechanism of resistance itself being so costly that, overall, these costs offset the benefits.

One can consider selection for resistance in terms of concepts of periodic selection and genetic hitchhiking. A microorganism that is resistant to an environmental antagonist will, upon application of that antagonist, increase in frequency within a population relative to competitors that are not resistant. This, in clonal populations, is the phenomenon known as periodic selection. That is, to the extent that populations are clonal then the result can be dominance of the population by a single genotype, the genotype in this case that is associated with resistance. The fitness of this genotype will be a consequence, at a minimum, of carriage of a resistance allele.

All other alleles associated with a genotype that displays antibiotic resistance will increase in prevalence given sufficient antibiotic exposure. In fact, this increase may occur even if those associated alleles are otherwise detrimental to the organism, which is the concept of genetic hitchhiking. The resistance allele too may be detrimental to the carrier but nonetheless overall give rise to a phenotype that is beneficial (re: tradeoffs). For example, carriage of the resistance allele could result in reduced growth rates, as measured in the absence of the antagonist, but nonetheless result in a fitness advantage, such as a greater growth rate in the presence of environmental antagonist. It is even possible for mutations to arise which compensate for the fitness costs associated with carrying or expressing a resistance allele. Such genes, for example, could be ones that result in greater control by an organism of the expression of resistance alleles.

All of these factors combine to give rise to the idea that while selection for resistance can be a trivial matter, the details of that selection may not be quite as simple as one might expect. A portion of those complications arise as a consequence of microorganisms often displaying relatively low-levels of sex, such as in terms of gene-exchange within a species or population. The result is that it is not just a resistance allele that comes to prominence within a population but instead a resistance-associated genotype. Additional complications arise because adaptations generally do not occur within a vacuum but instead can impose various costs on organisms (i.e., tradeoffs), and those costs in turn may be mitigated by the occurrence of additional, somewhat compensating adaptations.

Selection for Reduced Costs of Resistance Carriage

The principle of allocation states that an organism cannot simultaneously allocate a unit of resource to both survival and reproduction, where resources typically are considered particularly in terms of energy. Thus, any gene that must be replicated, or protein that must be synthesized, is potentially burdensome for an organism since resources devoted to generating as well as using those molecules cannot also be used towards reproduction. Exceptions to this statement are threefold: (1) The molecule may actually be contributing to reproduction, (2) the molecule may contribute to the acquisition of additional resources, or (3) the molecule may provide a selective benefit that offsets its cost. Furthermore, the lower the cost of something then smaller the selective benefit needed to offset the cost.

Given that overall, that is, net costs are a consequence of a combination of both "gross" costs and benefits, an organism can achieve net benefits by either enhancing utility or by reducing costs. Enhancement of utility can be a consequence of a molecule interacting more effectively with other molecules within an organism. Reducing costs, by contrast, can be the result of a molecule interfering to a lesser extent with the functioning of other molecules. In either case, mutations that enhance benefits or reduce costs can directly impact targets other than the focus molecule. Such modifications can be described as epistatic, which more generally is described as an interaction between loci towards the generation of a given phenotype. The acquisition of a resistance mechanism thus can be costly to an organism due to a combination of the resource costs of generating the adaptation and because the adaptation potentially could interfere with the functioning of other adaptations within the same organism. Mutations that either better control the focus adaptation or which allow it to better interact with the rest of the organism when expressed therefore can be beneficial. Furthermore, these compensating mutations can be located in loci other than that which is directly responsible for the resistance adaptation.

On the one hand, compensating mutations can be a good thing because they give rise to a reduction in the cost of carriage of an adaptation. On the other hand, the presence of compensating mutations can result in reductions in fitness should the organism ever lose the original adaptation. This can be restated by saying that selection, given compensating mutations, can result in the retention of adaptations that are no longer useful to an organism other than in terms of retaining the fitness advantages associated with other, subsequent mutations. These issues are a variation on the concept of specialization or antagonistic pleiotropy, but as applied to an organism's interaction with disparities in its own genotype, or indeed simply another way in which fitness is context dependent, with here the context again being an organism's own genome and gene products.

An antibiotic-resistant bacterial strain, as a consequence of accumulating compensating mutations, thus could find itself unable to revert to bacterial sensitivity without cost, even if the original selective agent, the antibiotic, were no longer available in its environment. This, furthermore, could be the case whether the resistance mechanism were chromosomally or instead extra-chromosomally located. Selection thus can favor increases in the frequency of resistance alleles, opposition to loss of those alleles while the original selective agent persists in the environment, and also could prevent the loss of alleles even after the selective agent is no longer present. Note, though, that blocks on the loss of a resistance allele that could stem from the presence of compensatory mutations is valid from a perspective of intra-clonal competition, i.e., bacteria that revert to antibiotic sensitivity but which retain compensating mutations could find themselves at a selective disadvantage even given an absence of antibiotic in the environment. At the same time, however, this does not mean that a resistant organism could not simply be replaced by one that is more fit simply because it hadn't acquired the antibiotic resistance to begin with.

Selection for only Transiently Useful Traits

Often genes are only facultatively useful, that is from time to time rather than more consistently so. This episodic utility comes about because the environments found both internal and external to organisms are not static but instead can vary over time or from place to place. Antibiotics, for example, may or may not be present in an organism's environment at any given time, or place. The result is that selection is not always operating for retention of a given trait within a population. Instead, selection for a given allele can at times be reduced to zero. Key issues are how long selection is absent in combination with how costly a trait is to retain. If costs are high and intervals without selection are long, particularly longer than the life span of the carrying organisms, then we can reasonably expect that a trait will be lost from a population. This issue is also relevant to the initial presence of alleles within populations: Evolution is not anticipatory and the fitness associated with a given allele is a function of how beneficial that allele is immediately. Newly arising alleles that convey no benefits to their carrying organism, even if they possibly could someday, will tend to be lost from populations because natural selection will not immediately serve to resist that loss.

The agents that microorganisms acquire resistance to are not always present in environments and therefore, given the above considerations, we have to wonder not only how resistance mechanisms persist in populations but also how they resist extinction following their initial occurrence within a population. The short answer in addressing these questions is that a great number of alleles in fact do go extinct, following either their mutational generation or horizontal transfer. Two things can counter this tendency, however, and these are, essentially, a combination of luck and hitchhiking, where hitchhiking, actually, can be viewed simply as a subset of luck. Furthermore, we can distinguish mechanisms into those that are driven by natural selection versus those that are not. Both luck and hitchhiking can have natural selection components and this is what we will focus on.

An allele encoding some form of resistance, and which newly enters a population, might, with luck, immediately solve a problem for the carrying organism. That is, the allele might be selectively beneficial right from the start, even if the selective agent is present in the environment only transiently. Hitchhiking is a variation on this idea, only instead of the focus allele being selectively beneficial right from the start, that allele is simply linked to a second allele (or alleles) that provides the selective benefit. In the case of resistance alleles, particularly those carried by plasmids, the dependence on luck for allele survival, including in terms of genetic hitchhiking, appears to be part of the retention strategy.

A resistance plasmid is one that carries one or more genes that encode bacterial resistance to something. Most notably, at least as far as humans and medicine is concerned, this resistance can be to antibiotics. Interestingly, often antibiotic resistance genes are found in multiples per plasmid. The issue of luck and persistence when encoding functions that are only occasionally useful may be the reason for multiply encoding resistance factors, and we can consider why this is so both from the perspective of the plasmid serving as the unit of selection (that is, rather than the carrying bacterium serving as that unit) and in terms of genetic hitchhiking (again with the plasmid serving as the unit of selection).

The idea actually is a relatively simple one, and equivalent to placing multiple, different bets on outcomes. In most instances none of these bets pay off, that is, the multiple genes carried by a given plasmid all may not supply any benefit following acquisition by a bacterium. The result, either due to genetic drift or because the plasmid otherwise carries some cost of carriage, is a loss of the plasmid from the population. Occasionally, however, one of the plasmid genes will supply a benefit prior to its loss from a population, resulting, potentially, in fixation of the plasmid within the population. Similarly, if the plasmid is already established, then occasional usefulness of its genes may assure a retention within the bacterial population. Indeed, we can easily envisage a steady state where increases in plasmid frequency due to occasional usefulness is exactly balanced by plasmid absence of usefulness. Furthermore, the likelihood of plasmid usefulness should increase as more genes are added to the plasmid, though so too should costs of carriage. Each individual gene, when it is not useful, is retained within the population due to genetic hitchhiking with that gene (or genes) that currently is useful. This hitchhiking is with regard to both plasmid persistence and selection acting on the bacterium.

As with the plasmid overall, we can envisage a resistance gene steady state on plasmids. Here the genes increase their presence in conjunction with supplying a selective benefit to the plasmid, but if a gene rarely or never supplies such a benefit then selection for retention of that gene on the plasmid may be insufficient to counter loss due to random mutation and genetic drift. This likelihood of loss of the gene may be larger to the extent that the gene is costly to carry. Note in these discussions that the benefit of the gene to the plasmid, in terms of encoding resistance functions, is mediated through the carrying bacterium, where the bacterium either does or does not survive within its environment (and/or successfully compete within its population). Plasmids nonetheless can carry other genes that play other roles in enhancing plasmid fitness without directly enhancing bacterial fitness, and these latter genes are involved in plasmid retention within a bacterium as well as plasmid transfer between bacteria. What function serves as the unit of selection and what unit of selection a given gene may be benefiting (or hurting) – gene, plasmid, bacterium, even bacterial population – can differ depending on circumstances as well as the nature of the gene and even in terms of what perspective one takes in considering such issues. Far from being a bizarre, overly complicated take on how evolution hypothetically might operate in this specific system, in fact we will repeatedly return to similarly complex scenarios throughout the text.