In this section I consider in greater detail the general categories of cooperative behavior. These include though are not limited to exploiting environments with efficiency rather than expediency, the potential for bacteria to evolve cooperative strategies within the context of biofilm formation in conjunction with subsequent dissemination, the social development of cooperative spore stalks by otherwise independent-living microbes, and the actual merging of genomes as a seemingly cooperative act. The latter represents an extreme though nonetheless not too uncommon limit to tendencies toward symbiosis formation, which is a subject considered in detail in the second chapter on microbial cooperation, that is, which deals specifically with the subject of symbiosis ("Symbioses"). My focus in each case will be on not just the advantages of cooperation but also on how exploitation of cooperative individuals by defectors may be avoided.
A straightforward, though not necessarily easily attained cooperative behavior is one of restraint to avoid a Tragedy of the Commons. In many microbiological examples, however, the Tragedy of the Commons is not quite an accurate metaphor since what is being described is a failure to achieve some high or maximal peak population density. By contrast, and more traditionally, a Tragedy of the Commons is avoided not by maximizing peak population densities but instead by achieving and then sustaining high environmental carrying capacities. Indeed, a high peak population density within a given environment can be well beyond the carrying capacity for a given species, resulting in sufficient environmental depletion that the ability of the environment to support further organism reproduction is somewhat reduced. Nonetheless, while reproductive restraint could lead to sustaining higher carrying capacities by some organisms, for others reproductive restraint instead can ultimately allow higher peak population densities, that is, ones that are well beyond a environment's carry capacity. Such situations can be less ambiguously described in terms of conflicts between economy and expediency or equivalently as a conflict between growth yield and growth rate, respectively. Note that economy may also be described as displaying greater efficiency in the utilization of resources. Greater economy thus can be equated with biases toward greater growth yield as well as greater efficiency in resource utilization, both as seen in comparison with greater expediency.
It is the economical organism, grown in pure culture, that is more effective at reaching higher peak densities (which, as noted, may or may not be viewed as avoiding a Tragedy of the Commons). These same organisms, however, reach these higher peak densities at a slower rate, as they more efficiently convert available resources into progeny. The expedient organism, by contrast, grows its population faster, but at a cost of efficiency, with the result being a lower peak population density. In both cases it may be a decline in resources or an increase in the density of wastes that limits population size. Either way, it is an exhaustion of the potential to exploit local environments especially that limits population growth, rather than explicit, voluntary restraint on the limits of growth such that population sizes do not come to exceed carrying capacity (that is, it is restraint during exponential growth that is displayed by economical organisms rather than necessarily limitations on the duration of exponential growth; though the latter can occur as well, it is not our emphasis). Restraint on population growth rates thus can be viewed as a cooperative behavior, and one, given mechanisms of microorganism death balancing births, that might even give rise to attainment of a higher carrying capacity, e.g., such as during growth within chemostats. How, then, might such restraint evolve?
In a chemostat containing two trophic levels – that is, consumer and resource – the more efficient organism can be defined as that which requires the least amount of resource per progeny produced. What organism is the better competitor, however, is determined either by display of higher intrinsic growth rates or higher resource affinity (see "Predicting Best Competitor "found in the chapter, "Passage Through Time"). Thus, it is easily possible for an organism to be a better competitor and to be a less efficient resource utilizer. Unlike the previous example, where maximizing peak population density was described as not necessarily equivalent to avoiding a Tragedy of the Commons, in this case the Tragedy of the Commons metaphor is quite apt since chemostat steady state densities are equivalent to a chemostat's carrying capacity. Indeed, chemostats can be highly illustrative of a Tragedy of the Commons because of this idea that the better competitor is not necessarily the most efficient competitor. The better competitor in a chemostat, that is, is not necessarily the organism that has the potential to display the higher carrying capacity.
In a chemostat consisting of two types of competing organisms, selection favors those organisms that can best sustain their populations in the face of low nutrient densities in combination washout/outflow, rather than the one that can sustain the highest population density in the absence of competition. That is, while sustaining a population at a lower resource density defines which organism is the better competitor, greater competitive ability, as noted, does not have to give rise to higher carrying capacities. This seeming paradox occurs because carrying capacity itself is determined as a balance between rates of resource utilization, rates of cell division, and rates of washout from the chemostat. Competitive ability by contrast is a function of growth rate and resource affinity, but not of rates of recourse utilization. As a result, a less efficient organism that displays a lower carrying capacity, but nonetheless either a higher growth rate or a higher resource affinity, can still be the better competitor, thus driving more efficient organisms to extinction. In essence, the less efficient but nonetheless better competitor is fueling that greater competitive ability at the expense of the environment. The resulting environmental degradation, the Tragedy of the Commons, affects all competing organisms, but the better competitor still wins so long as its lower efficiency is coupled with an ability to sustain its population at lower resource densities than can its competitors.
Considerations of what cooperation entails in chemostats points not to how such cooperation can necessarily evolve but instead to what such cooperation in fact means, and that is that the more cooperative organism is that which displays the more efficient population growth. Thus, it is not so much organisms that display intrinsically faster population growth rates or higher resource affinities that are the defectors, in terms of a chemostat Tragedy of the Commons, but instead that those organisms may achieve those qualities at the expense of their efficiency of growth. In other words, defection in a chemostat can have at least two guises, maximizing peak population density at the expense of carrying capacity (previous section) or, alternatively, maximizing competitive ability also at the expense of carrying capacity. Defection therefore is not so much about replicating faster as it instead is about enhancing one's population growth rate through utilizing more than one's "share" of the commons.
An additional consideration is that in and of itself decreased efficiency of resource utilization is not necessarily a beneficial quality since it can come about in the absence of some compensating utility such as faster growth rates—we might describe such inefficient but also ineffective organisms as maladapted. Thus, even among bacteria, resource "hogs" are not necessarily intrinsically also better competitors. We instead can view resource utilization efficiency along with any benefits that might be associated with this lower efficiency as potentially representing a pleiotropy, but one where it is not resource inefficiency that is selectively advantageous but instead what beneficial phenotypes can be manifest in association with this inefficiency. In other words, it is not what resources one can accumulate that's important so much as how one is able to utilize those resources (it's not what you have that counts but instead what you do with it!). Evolutionarily, in chemostats, that utility can be measured in terms of competitive ability.
While there may be no practical means of restraining inefficient resource utilization in continuous culture, if less efficient organisms still are the better competitors, restraint nonetheless may evolve as a consequence of between-culture competition. In such competition, the better competitor may be the individuals that display the higher carrying capacities, higher peak densities, or better dissemination abilities. To the extent that the better cooperator, within cultures, is able to achieve these things, and sufficient bottlenecking occurs – along with other mechanisms assuring clonality, such that within-culture cooperators need not compete directly with cheaters – then cooperation may yet evolve. Such evolution, experimentally, is more easily followed employing batch cultures.
Demonstration of the selective advantages associated with greater efficiency, and thereby greater peak density absent within-culture competition, has been achieved employing bacteriophages. One example comes from my own work (Abedon et al., 2003) and involved batch-culture competition between phages displaying economy (i.e., cooperators) versus phages displaying expedience (i.e., defectors). That is, one phage produced more progeny per bacterium infected, but took longer to do so, whereas the second, cheater phage produced fewer progeny per bacterium, but did so faster (i.e., shorter latent period and smaller burst size equals expedience versus longer latent period and larger burst size which is efficiency). The resulting observations were that when resources (bacteria) were plentiful, then the expedient phage could achieve multi-fold greater peak population densities than the efficient phage, but only if the two phage types were allowed to compete in the same culture. By contrast, when phages were grown separately, then it instead was the efficient phage that attained multifold superiority in terms of peak density. To the extent that dissemination ability to new environments is enhanced by greater peak densities (that is, if more organisms at the start of dissemination means more organisms at the end of dissemination), and peak densities can be attained prior to such dissemination, then it is clear that economy would be advantageous over expediency. These ideas have since been corroborated by Kerr et al. (2006) and both studies are discussed in greater detail elsewhere (Abedon, 2008; Kerr et al., 2008; Abedon, 2009a) .
There are two basic scenarios for the evolution of intraspecific cooperation as they may occur within biofilms, those that directly involve the formation of extracellular polymeric substances and those that do not. In either case, the extracellular polymer, often a polysaccharide, serves as the "glue" that holds together the biofilm, a population or community of microorganisms either adhering to a surface or which exist as flocculations within aquatic environments. This glue, too, may serve as a barrier to penetration by would-be defectors into microcolonies that otherwise consist of cooperating bacteria. Especially in the first scenario, as discussed immediately below, I take that lack of penetration simply as a given.
Kreft (2004) provides a model for the evolution of cooperative traits within biofilms, specifically the evolution of more economical rather than more expedient growth strategies. His model depends on a combination of the propensity of bacteria within biofilms to exist within microcolonies, the tendency too for those microcolonies to be clonal, i.e., to resist invasion by other bacteria, and, lastly, the tendency of dissemination from microcolonies to involve single rather than aggregates of bacteria. The result is sufficient genetic similarity among cells, within microcolonies, that successful dissemination by one cell is evolutionary equivalent to successful genetic dissemination by all cells making up the single microcolony (the equivalent in us is for the cells making up our bodies to contribute to the production of gametes that go on to succeed in producing successful offspring). Furthermore, genetic similarity is assured through the single-celled genetic bottlenecking involved in dissemination. Thus, any mutants that are more effective cheaters will tend to lose out upon dissemination if their better competitive ability (the cheating tendencies) does not correspond to effective biofilm formation, as a clonal population, and subsequent dissemination.
This again is an antagonistic pleiotropy argument, where gains at one life stage have the potential to impact abilities at another. In particular, gains in competitive ability within microcolonies can come at the expense of dissemination ability towards formation of new microcolonies. In other words, so long as cheating provides only short-term gains in fitness, which are then followed by fitness declines associated with dissemination, then cheater genotypes may not increase in frequency over time within the larger population. One possible mechanistic route toward reduced dissemination ability can come if cooperators simply produce larger microcolonies that possess a greater number of potentially disseminating bacteria. This explanation, as with that made in the previous section, assumes that selection for cooperation occurs as a consequence of between-microcolony competition rather than within-microcolony competition. That is, it assumes that cooperators are competing with defectors within what are essentially separate cultures (different microcolonies). Note also that as with competition between phages, cooperation may be more advantageous given culture (microcolony) maturation prior to dissemination, particularly if cheaters have an initial growth-rate advantage that comes at the expense of overall microcolony fecundity at the point of maturation.
Kreft (2004) provides a more specific scenario for this between-microcolony competition, one that is still dependent on microcolony maturation. Here, again, the initial cheater growth-rate advantage is lost as the microcolony matures, which has the effect of reducing the number of cheater cells available to contribute progeny to the disseminating pool, that is, at the point of microcolony maturation that this dissemination occurs. This scenario too qualifies as an antagonistic pleiotropy, one that specifically involves early-stage growth rate advantages that come at the expense of later-stage growth rate advantages. Such a tradeoff is also seen in the Blount et al. (2008) study where faster early-stage glucose utilization by planktonic E. coli, was associated with a failure at later stages to fully exploit available resources (in this case citrate as a carbon source). As above, gains in competitive ability do not necessarily come at the expense of dissemination ability.
As Kreft (2004) suggests, a polar opposite scenario likely operates during the broth-culture enrichment techniques often employed during microorganism isolation. That is, those organisms that display especially rapid, planktonic population growth will tend to come to dominate broth-based cultures. This would imply a bias among environmental isolates of bacteria towards those which, perhaps, are not the most effective biofilm formers but, instead, better planktonic growers. This argument also, though implicitly, may be one of antagonistic pleiotropy, where better growth under planktonic conditions is assumed to not be consistent with a more effective biofilm-based life cycle. In fact, bacteria that are faster growers during planktonic growth in effect initiate dissemination prior even to the imitation of microcolony/biofilm formation.
In the model of Kreft (2004) , the evolution of cooperation in biofilms is assumed to be a function of late-stage growth advantages that come from more economical rather than more expedient growth strategies. Xavier and Foster (2007) also provide a model for the evolution of cooperation in biofilms, though their model is based explicitly on the production of a common good, in this case an extracellular polymer. In their model, polymer production is not without cost, and therefore a more expedient growth strategy might include a forgoing of or at least reduction in extracellular polymer production. The cost of such a strategy, however, comes later in biofilm development, i.e., upon greater biofilm maturation. Here, just as with Kreft's model, the cheater is predicted to display less robust later-stage growth. Unlike Kreft's model, Xavier and Foster provide a mechanistic explanation for the less-robust growth: Without extracellular polymer production, microcolonies are less able to project themselves into nutrient-containing currents, and indeed may be shadowed from those currents by more robust, polymer-supported microcolonies. Thus, the cooperator bacteria – those producing their extracellular polymer – as with Kreft's model, ultimately are better competitors even within microcolonies than bacteria producing less polymer.
Xavier and Foster (2007) suggest that their scenario is analogous to competition among plants for light, where the plant (as a clonal group of cooperating cells) can be more effective in the long run if they invest, early on, in costly wood production. Woody growth, that is, serves as a support for subsequent plant (e.g., tree) vertical growth, as well as shading of plants (e.g., "weeds") displaying more expedient growth strategies. Note that stromatolites, which are macroscopic mats of cyanobacteria, and which can form in places which are devoid of disruption by animals, too may be viewed as a form of vertical growth which allows better access to sunlight (the animals, sponges, also can be viewed as benefitting from cooperative thrusting of extended bodies into currents). The bottom line is that cooperators ultimately possess a selective advantage because cooperator microcolonies come to shade defector microcolonies in terms of access to nutrients along with oxygen that is flowing past the surface upon which the microcolonies are located. Notwithstanding these advantages, if cheater mutants are able attain equivalent access to extracellular polymer as the polymer producers, then the polymer producers may not ultimately display a reproductive advantage. To the extent that this public good actually is less publicly available but instead is actually more available to the producer than it is to non-producers, despite its extracellular location, then the producers may retain their late-stage growth advantage. Note the equivalence of this argument to previous considerations of the evolution of individuality: Selective advantages exist only to the extent that associated phenotypes are not too extensively shared with would-be cheater individuals.
In both the Kreft (2004) and Xavier and Foster (2007) models, no allowance is made for the penetration of otherwise clonal microcolonies by cheaters. Such penetration can occur either from without (by otherwise unrelated individuals) or from within (by mutants). Thus, in both cases, there may be limits to the benefits of maturation in the selection for cooperators since, with time, the resulting structures (microcolonies) may come to acquire an increasing cheater load. This occurs because biofilms may provide an exploitable common good, resulting in a Tragedy of the Commons if not successfully policed, where means of combating such cheating could be viewed as occurring within a context that represents a primitive multicellularity. On the other hand, without that policing, simple multicellularity may be destined to failure, as illustrated by the experimental biofilm model of Rainey and Rainey (2003) .
Perhaps the classic example of microbial cooperation involves spore-stalk formation by aggregating social amoeba or bacteria, i.e., the cellular slime molds (Dictyostelium) and analogous bacteria (i.e., Myxococcus). In both cases a free-living state is followed by aggregation of cells and then the formation of spores (where the latter represent the single-celled dissemination state). The formation and successful dissemination of these spores requires cooperation among the aggregating cells, and this is especially so since not all aggregating cells are able within the resulting spore stalks to go on to produce spores. That is, some cells are destined to exist only as spore-supporting stalk cells. This is a situation that is analogous to the employment of extracellular polymers to support mature (and potentially disseminating) biofilm microcolonies (Xavier and Foster, 2007) , except with the added complication that the cellular aggregation step that comes prior to stalk formation reduces the potential for the quasi-multicellular entities to be clonal.
Thus, from the start there is a problem of invasion by cheaters which is equivalent to the problems of invasion given sufficient maturation as discussed in terms of biofilms, above. Indeed, mutations in a large number of genes have been found to give rise to cheating behavior, though often with context-dependent plasticity. In other words, sporulation by these organisms can be viewed as a cooperative, biofilm-like state, only one where the clonality of projecting structures presumably is less assured.
The genome of an organism is some representation of the complement of its hereditary material. For viruses, that material can be DNA or RNA, single-stranded or double. Here I look at an example of experimental evolution in which two viruses were induced to merge by providing an advantage to the genetic linkage of their genomes, which in turn gave rise to a somewhat forced symbiosis. This example might be also viewed as a viral equivalent to the evolution of endosymbiosis, i.e., two organisms sharing a single cell or, in this case, both a single cell (upon infection) and a single capsid (prior to infection).
The merging (or integration) of phage genomes appears to be fairly common, and is the most important result obtained from studies of comparative phage genomics. That is, phage genomes, especially those of tailed phages, appear to be mosaics where different parts have come from different phage lineages. Such high levels of horizontal gene transfer appear to approach the kind of genetic integration achieved with endosymbioses. The occurrence of these integration events, however, appear to be fairly rare and therefore are difficult to detect other than post hoc; see Labrie and Moineau (2007) , though, for examples of large-scale mosaicism developing in the course of experimental evolution.
Sachs and Bull (2005) provide an alternative approach using two filamentous, ssDNA phages, each carrying a selectable marker (antibiotic resistance). These phages establish chronic infections and therefore can effect a long-term antibiotic resistance on infected hosts. In the presence of both antibiotics, only given the coinfection of a bacterium by both phages will either phage or host survive. At low phage multiplicities, however, the likelihood of coinfection is low. A possible and interesting way around that limitation would be for the two phages, upon coinfection, to genetically recombine so that one phage carries both markers.
Filamentous phages are not highly recombinogenic, but are quite flexible in terms of the size of the genomes that they are able to package. The solution reached by the phages in the Sachs and Bull experiment appears to be one that takes advantage of the latter property: Rather than one phage acquiring both markers, instead the two phages are able to co-package, that is, two phages become packaged within a single capsid so that one phage adsorption event results in two phage infections. With co-packaging, the low-multiplicity barrier to coinfection is eliminated. Therefore the likelihood of survival in the presence of both antibiotics is enhanced.
The co-packaging event, as noted, is very similar to the occurrence of endosymbiosis, where one cell becomes "packaged" within a second cell. As will be covered in greater detail in the following chapter (Symbioses), with endosymbiosis there often is a genetic degradation of one of the partners, the endosymbiont. So too did this occur with the phage co-packaging, where one of the two phages lost genes employed in its morphogenesis since those genes were no-longer-necessary following the evolution of co-packaging.
One can generalize these events as consisting of the following steps: (1) There exists a quandary stemming from the lack of genetic linkage of two (or more) essential genes. (2) An increased association of essential genes, however, is subsequently achieved through horizontal gene transfer, co-packaging, endosymbiosis, or some other solution that assures genetic linkage during co-vertical transmission of genes. (3) Following this linkage, there is then a streamlining of the association, one that reduces redundancy. This streamlining also involves a coevolution, i.e., physiological integration, of the various factors carried by the original organisms. The latter is an example of cooperation evolving once gene fates become intertwined, that is, once genes are linked it is more difficult to sustain diverging interests. As noted, such genetic integration is not uncommon, i.e., mosaicism of tailed phage genomes appears to be rampant. Phage co-packaging, by contrast, though highly illustrative to the process of genome integration, may be somewhat contrived. Symbioses and endosymbioses, on the other hand, appear to be very ordinary conditions.