OUP user menu

Generational coexistence and ancestor's inhibition in bacterial populations

Fernando Baquero , Marc Lemonnier
DOI: http://dx.doi.org/10.1111/j.1574-6976.2009.00184.x 958-967 First published online: 1 September 2009


Generational coexistence in structured environments raises the possibility of a competition between ancestors and descendents. This type of kin competition, and in particular, the possibility that descendents might actively repress the ancestor's dominance, has been rarely considered in microbial evolutionary ecology. The recent discovery of the phenomenon of stationary-phase contact-dependent inhibition of bacterial ancestor cells by late descendents provides a new theoretical perspective to analyze intrapopulational evolutionary changes. The ancestor's inhibition effect might accelerate such changes, particularly when the descendents have acquired small adaptive advantages that are insufficient to rapidly displace the well-settled ancestors in a complex niche. Besides this effect of triggering selection of small genetic differences, the opportunities for intergenerational coexistence in bacteria, where ancestor's inhibition might occur, are reviewed in this work. A theoretical analysis is provided about the explanatory possibilities of the ancestor's inhibition effect in the controversies about intraspecific (in a large sense, including intrapopulational) genetic diversification, and the discontinuities observed in such processes, giving rise to the emergence of individualities and therefore differential units of selection.

  • cell-to-cell communication
  • self–non-self- recognition
  • evolution of bacterial populations


The sequential substitution of ancestors by their descendents along the time line is the most evident consequence of replication and, therefore, of evolutionary biology. Ancestors progressively enter in senescence, displaying lower reproductive rates, and sooner or later, become extinct. However, ancestors might also reduce the reproductive rates of their descendents. During its life span, the ancestor could have reached a pre-eminent position in a particular environment, so that most of the locally available nutritional resources are obtained by the ancestor, to the detriment of the descendents, at least in the absence of vacant sites. Eventually, the descendents themselves might serve as a source of food for the ancestor. An intuitive image is a large and dense tree dropping its dehiscent fruits and seeds below its top, out of the light and the nutrients, damning them to decay and death. If a proportion of these descendents harbor potentially beneficial genetic changes for the species, this advantage will be lost. This implies a retardation of the evolutionary rate, and a consequent increase in the propensity of being invaded by more adapted foreign organisms. Surprisingly, the possibility that descendents might repress the ancestor's dominance has been rarely considered in the literature. It is likely that such a phenomenon occurs in termites, which modify their environmental conditions in such a way that the early death of established reproductives takes place, the latter being replaced by a more plastic offspring (accelerated inheritance hypothesis) (Thorne et al., 2003). If such a repression by descendents is rare in nature, then what is the real biological cost of ancestor's inhibition? Evidence of ancestor's growth inhibition by direct contact with descendents at the stationary phase has been recently presented in a model system in bacteria, possibly offering a new way of understanding the population dynamics and evolutionary trends in bacterial systems (Lemonnier et al., 2008). In this review, we consider the different scenarios in which generational coexistence and competition is expected to occur. We discuss the potential implications of ancestor's inhibition by descendents from the main standpoint of bacterial organisms.

Ancestor's inhibition in Escherichia coli

Recently published observations endorse the possibility of ancestor's inhibition in bacterial organisms. The evolution of strains that killed or inhibited the growth of the bacteria from which they were derived (their ancestors) was observed during the course of serial passage experiments using liquid cultures of a mutator gene deletion strain of E. coli K-12 (Lemonnier et al., 2008). Increased rates of mutation might occur in 20% of the cases in E. coli clinical strains (Baquero et al., 2004). This inhibition was referred to as stationary-phase contact-dependent inhibition (SCDI) because it occurs after the cells stop growing and requires physical contact between the evolved and the ancestral bacteria. The phenomenon was found to be frequency and density dependent: it does not occur when the frequency of the evolved cells in the mixture is low or when the total density of bacteria is beyond a certain threshold (c. <5 × 107bacteria mL−1). SCDI simulation results indicate that if SCDI were the only advantage possessed by evolved bacteria, their rate of ascent during serial passages would be low and they would probably not have been detected, and much less dominated the population. In other words, the ancestor's inhibition phenotype only emerges if a density- and frequency-independent fitness advantage occurs in the population, allowing reaching the SCDI threshold.

SCDI evolved independently in eight different cultures through seven different mutations in the same gene, thus providing an interesting example of convergent evolution in action. The precipitous decline in the density of the ancestral cells was shown to occur when naïve cells were mixed with evolved bacteria obtained after 62 passages (c. 412 generations) (Lemonnier et al., 2008). However, the phenomenon was also observed when naïve cells were mixed with evolved bacteria isolated as early as the 14th (1/100) transfer of one culture (c. 93 generations), and evolved bacteria with this property dominated all eight cultures by the 35th transfer (c. 233 generations) (our unpublished results). It has been suggested that historical contingency is especially important when it facilitates the evolution of key innovations (Blount et al., 2008). Mutations in the glgC gene, which codes for the enzyme ADP-glucose pyrophosphorylase of the glycogen synthesis pathway, were found to be responsible for both the ability of the evolved bacteria to inhibit or kill their ancestors and their immunity to that inhibition or killing. No clear mechanistic connection, however, could be established between the accumulation of glycogen in evolved cells and the observed bactericidal/bacteriostatic effect that they induced. The glgC mutations may be considered as suppressors of one or more genes responsible for the sensitivity of E. coli to SCDI. An attractive hypothesis is that the accumulation of glycogen by evolved cells might interfere with the regulation of chromosomal toxin–antitoxin (TA) systems, thus preventing in evolved cells the death or the growth inhibition triggered by TA systems in response to nutritional stress (Hayes, 2003). Interestingly, cell death triggered by TA systems has been shown to be depend on quorum sensing and hence on cell-to-cell communication (Kolodkin-Gal et al., 2007). According to this model, the hyperglucogenic trait would only be effective in case of nutritional dearth, and so would be neutral in times of welfare and could be rapidly lost in fast-growing cells. This is consistent with our findings that SCDI was only active during the stationary phase. Recent data document what seems to be a different case of contact-dependent inhibition that is detected in late stationary-phase cultures of E. coli. However, ancestor's inhibition could not be invoked in this case because mutations in the uup gene, encoding a soluble ATP-binding cassette ATPase, rendered the bacteria sensitive to their uup+parents (Murat et al., 2008). In conclusion, SCDI offered what is, to our knowledge, the first example of a genuinely de novo evolution of an allelopathic mechanism, interestingly targeting the ancestor's population.

Aging and generational coexistence in bacterial populations

Bacterial populations are inherently multigenerational. Typically, the E. coli population obtained under standard laboratory growth conditions, as in an 18-h broth tube or in a bacterial colony in a Petri dish, is composed of cells of at least seven different generations. In practice, while an unlimited number of generations can be readily yielded and monitored when studying bacterial populations, generational coexistence has been consistently overlooked by microbiologists. Arguably, a cause for this situation resides in the long-standing perception that the bacterial cell cycle acts like a perfect perpetuating machine that replicates identical individuals and faithfully segregates the genetic material to daughter cells. From this standpoint, it was obviously unclear whether ancestors and descendents could actually be discriminated, and the technical limitations to experimentally approaching this issue seemed intractable. It has been recently shown that populations of genetically identical Bacillus subtilis comprise numerous distinct cell types, exhibiting a spectrum of phenotypes (López et al., 2009), and therefore distinct reproduction rates are expected to occur. Moreover, recent studies have challenged the vision of bacterial immortality and provided evidence for bacterial aging and death (discussed in Nystrom et al., 2003, 2007). Model rod-shaped bacteria such as E. coli, which were once thought to undergo finely regulated symmetric divisions on the basis of morphological criteria and refined genetic evidence, actually carry out physiologically asymmetric divisions. Progeny cells that inherit the old pole grow more slowly, divide less frequently and are more subject to death than their new pole counterparts (Stewart et al., 2005). It has been reported that cellular damage (manifested by protein aggregates) is asymmetrically segregated to descendents that inherit the old pole. This would in turn provide the new pole cells with a rejuvenation mechanism that would promote the perpetuation of the population (Lindner et al., 2008). From the perspective of generational coexistence, it is interesting that bacterial aging is associated with mechanisms whose apparent function is to suppress aging by dooming senescent individuals to cell death. Yet, from existing data, it seems unlikely that aging-associated death depends on specific interactions between aging and juvenile cells. Hence, current approaches to bacterial aging have not addressed the possibility of generational competition. In addition, there has been a great deal of recent contributions of utmost elegance reporting competition phenomena within bacterial populations. ‘Cannibalism’, ‘fratricide’, contact-dependent inhibition in mixed bacterial populations as well as other mechanisms to kill cells that are genetically identical (or nearly identical) have been reported and their physiological roles have been discussed (reviewed in González-Pastor et al., 2003; Slechta & Mulvey, 2006; Claverys & Havarstein, 2007). In many ways, these mechanisms are supportive of a multicellular mode of life of bacterial populations, and can be interpreted as adaptive behaviors, either collaborative (even altruistic) or competitive, in response to specific environmental stresses. Then again, the impact of generational coexistence in these competition phenomena remains essentially unaddressed. Doubtless, however, populations composed of overlapping generations occur in nature. Some common natural scenarios are discussed below in which multigenerational bacterial interactions take place and where generational competition can be hypothesized.

Spatially structured environments and generational competition

While coexistence among generations occurs in organisms located within colonies grown on agar media or in broth cultures under standard laboratory conditions, a more relevant experimental model for this phenomenon can be found in structured environments such as biofilms. These surface-associated communities are physiologically distinct from the free-swimming (planktonic) bacteria that are found in broth cultures grown on shaking incubators. In a simplified view, biofilms are characterized by dense clusters of cells attached to a surface and embedded in an extracellular polymer matrix that maintains the community together. The surface may be either abiotic (such as indwelling medical devices) or living like the mucosal surfaces that are colonized by human pathogens during chronic infection (Parsek & Singh, 2003; Hall-Stoodley et al., 2004). Biofilms are characterized by their spatial heterogeneity and may be seen as genuine ‘cities of microorganisms’ (Watnick & Kolter, 2000). Access to nutrients, oligoelements and oxygen is not uniformly available to biofilm ‘inhabitants’; channels eventually separate clusters of cells, creating chemical gradients, which in turn generate genetic and physiological heterogeneity (Stewart & Franklin, 2008). These physical or chemical gradients might also trigger intraspecies quorum-sensing systems (Shank & Kolter, 2009), potentially altering the local multicellularity patterns. Different types of cells (such as motile or matrix-producing ones) localize to distinct regions within the biofilm (Vlamakis et al., 2008). Such a self-promoted functional and spatial diversity is expected to protect communities from unstable environmental conditions (principle of ‘insurance hypothesis’) (Boles et al., 2004). Occasionally, the biofilm-associated bacteria detach from the biofilm matrix. Fluid-driven clumps of bacteria may then contribute to the dissemination of the infection within the host (Hall-Stoodley & Stoodley, 2005). Hence, in these compartmentalized (structured) communities, coexistence between bacterial generations might occur as a result of local migration. It is easy to imagine a series of spatially close environmental patches in which a particular bacterial population has been seeded. In each one of these relatively isolated patches, the bacteria have been able to replicate for a number of generations until a threshold is reached. The determinants of this threshold and the number of generations needed to reach it might differ in each patch depending on the local availability of nutrients, oligoelements, quorum-sensing signals, the nature of the colonizable surfaces and/or the protection it may provide against deleterious factors. A close relation between motility and biofilm formation has been shown (Lemon et al., 2008) in which, frequently, the planktonic-motile cells were those that exhibited faster rates of multiplication, thus leading to the interaction of their later generations with the more ancestral ones within the biofilm. Connectivity among biofilm patches might also be restored after a given period of time as a consequence, for instance, of swarming motility, thereby allowing the interaction between different generations.

Bacterial populations at the stationary phase frequently segregate two types of subpopulations upon dilution of stationary-phase cultures into a fresh medium: one that started dividing and another that did not (Roostalu et al., 2008). The later one are nongrowing persisters, temporarily dormant bacteria. Probably, a fraction of non-negligible proportion of growing cultures is composed by these slow- or nongrowing cells, which are tolerant to antibiotics (Wiuff et al., 2005). As a consequence, these cells can be selected by antibiotic action, and, after a period of time, when in the absence of drug exposure they resume growth, they could be exposed to their later descendents, in other patients or human communities, generational coexistence.

Another example of a natural circumstance in which generational coexistence might occur is the epidemic behavior of a bacterial pathogen. Different hosts are successively colonized and, in turn, different population sizes might be reached in each of them. Even different bursts of colonization or infection might occur in the same host, as in the case of recurrent urinary tract infection episodes, which are thought to be triggered by the intermittent dissemination of uropathogenic E. coli from biofilm-like reservoirs present in bladder epithelial cells (Rosen et al., 2007). Finally, endemic bacterial clones (ecotypes) are maintained for years in hospital institutions (Coque et al., 2008). The dispersal of the original clone in different compartments (like patients or wards) favors its diversification, by having different access to the mobile gene pool through horizontal gene transfer, and by being exposed to different selective landscapes. The adaptive advantages provided by acquired genes (such as antibiotic resistance genes) could produce an asymmetric selection of particular ecotype subpopulations that will increase its number of generations. In other compartments, the ancestor will remain less efficiently selected. Eventually, with time, such ancestor and evolved subpopulations will be relocated in proximity as they share the same vernacular habitat.

Generational mixing is therefore an expected issue under these and probably other circumstances (Fig. 1). How could ancestor inhibition impact the developmental programs of such dynamic, spatially structured and heterogeneous communities? Rapid changes might eventually appear as a kind of sudden anagenetic replacement of the ancestor following the branching of a descendent, microevolution within clones or clonal complexes. In practice, this might be observed as events of clonal replacements, which have been described (Caugant et al., 1981; Akiba et al., 1999, 2000; Zhang et al., 2000; Baquero et al., 2002; Shima et al., 2006). For instance, particular bacteria acquiring an advantageous trait in a hospital (such as an increased transmission rate or antibiotic resistance) could produce an extensive progeny that might contribute to the replacement of their susceptible ancestors in the hospital's patients, in a way that could not be explained by deficit in hygiene or increase in antibiotic use.


Four cells arising from a single ancestor might give rise to lineages with a different number of generations over the same period of time, if they are separated in different environments or by different processes. In the first line, generations occur in a regular way. In the second line, in another environment, fewer generations occur during the same period of time. In the third line, a process such as formation of persister cells or spores assures that essentially the same cell (same individual, see Fig. 4) is recovered after an extended period of time, when growth resumes. In the fourth line, there is a slow-growing lineage (as in the second line), but occasionally, this lineage is involved in an epidemic burst, with a sudden increase in the number of generations in short periods of time. At the right end, cells might coexist that correspond to different generations counted from the common ancestor.

As yet, ancestor's inhibition has been shown to evolve in planktonic E. coli grown under standard laboratory conditions. Therefore, evidence that this mechanism plays a role in the evolution and dynamics of structured communities will have to be provided. For this evidence to be conclusive, we believe that four determinant criteria should be fulfilled: (1) the evolution of the community should depend on direct interactions between the species within this community; (2) a mechanism for the self-recognition of ancestors and their progeny should be operative in these communities; (3) the evolution of the community should involve the killing or the inhibition of ancestors by the evolved progeny; and (4) mechanisms promoting intergenerational contact should exist and should play a significant role in the evolution of bacterial communities. From the existing literature, it might well be the case that each of these four criteria has already (yet independently) been documented. There is, for instance, compelling evidence that bacteria evolve specific cell-to-cell interactions that affect the structure and the relative abundance of species within a compartmentalized community. Such interactions eventually sculpt a descendent community that is more stable and productive than its ancestor (Hansen et al., 2007). Furthermore, the genetic basis of self–non-self recognition, a pivotal characteristic of the SCDI mechanism of ancestor's inhibition, has recently been clarified in a model experimental system using swarming colonies of Proteus mirabilis strains. Mutants that form boundaries with their parents have been isolated, and the molecular identifiers have been characterized (Gibbs et al., 2008). Moreover, the role of bacterial cell death during biofilm development has been documented. This involves (but is probably not limited to) the release of genomic DNA from lysed cells that favors cell cohesion and biofilm stability (Bayles, 2007). Although the mechanisms that govern cell death in biofilms are thought to be related to TA systems [also called programmed cell-death modules; Engelberg-Kulka et al. (2006)], it cannot be ruled out that fratricide, cannibalism or even parricide (ancestor's inhibition) mechanisms could contribute to cell death and hence to biofilm development. Finally, cell-to-cell communication plays a capital role in the developmental program of biofilms (Davies et al., 1998). Interestingly, quorum-sensing is required for the detachment of cells in Staphylococcus aureus biofilms, indicating that cell-to-cell communication plays a capital role as a dispersal mechanism to colonize new niches (Boles & Horswill, 2008). Hence, while the mechanism for the inhibition of ancestors in bacterial populations (SCDI) awaits elucidation, underlying mechanisms for contact-dependent bacterial death, self–non-self recognition, cell-to-cell communication and cell dispersal have been described. We anticipate that the molecular basis for ancestor's inhibition will become clear as hypotheses bridging these apparently different phenomena are tested. As it was stated before, the role in ecology and evolution of molecular tools produced by microbial ‘thieves, assassins and spies’ (Vlamakis & Kolter, 2005), leading to a sort of ‘unity from conflict’ (Rainey, 2007), should not be underestimated.

Bridging gradualism and discontinuities: selection of very small differences by ancestor's inhibition

On the basis of the survival of superior individuals and the gradual change of populations, the students of diversity are expecting continuity in nature (Mayr, 1982). What is found is all kinds of discontinuities not only among species, but also among subspecific taxons. Gradualism implies the selection of very small differences. A classical problem in evolutionary biology is to determine the minimal selectable phenotypic difference among individuals or populations. Indeed, it is frequently assumed that if the phenotypical effect of a mutation is low, its contribution to the selective advantage should be similarly low. The problem, rooted from the Darwin's times, is that many significant events in evolution occur by step-wise selection of minor changes (gradual evolution, including geographic variation) (Mayr, 1982). For instance, duplication in genes or genome regions might be at the origin of a number of adaptive events in bacterial organisms, starting by a minimal phenotype change. Of course, most of these extremely frequent duplications (as many as one in 103 bacterial cells) (Pettersson et al., 2005) are never selected.

The acquisition of a mutational trait (including sequence duplication) might provide a minimal potential fitness advantage, that it is only converted in real advantage when the host variant harboring the trait is hooked by selection. Selective multiplication of the variant is the touchstone of the trait's adaptive value. However, this multiplication does not necessarily imply persistence or evolutionary success. If the selective advantage is very small, and particularly if the environment fluctuates (and so the selective advantages), the fixation of such a trait into the population might require extensive periods of time or eventually will never occur. Periodic selection, a rapid and drastic change of a particular genetic variant in a population by another one, that is, the sequential dominance of most-fit variants, will produce the abortion of such potentially adaptive traits (Atwood et al., 1951). Even starved cultures of E. coli undergo successive rounds of population takeovers by mutants of increasing fitness as was illustrated by the growth advantage in the stationary-phase phenotype (Zambrano & Kolter, 1996; Zinser & Kolter, 1999). Because of the complexity and variability of natural environments, such a monodimensional success of a particular trait might eliminate the possibilities of a more successful multivectorial evolutionary path. Indeed, in large asexual populations, multiple beneficial mutations, some of them providing small advantages, might occur in different coexisting lineages, causing competition among them (clonal interference). How to assure the permanence of subdominant traits of potential adaptive value in bacterial populations?

We have previously explained the selection of very small differences in bacterial populations as a result of the exposure of bacterial variants to deleterious environmental gradients (Baquero & Negri, 1997) so that a point in a gradient could lead to the precise selection of an optimal variant adapted to this particular point. For example, a very small increase in antibiotic resistance provides a selective advantage at very low antibiotic concentrations (close to the level of resistance), but not at lower or higher concentrations, where the variant population either has no advantage or is killed, respectively (Negri et al., 2000). Another mechanism might occur by which very small differences could be selected in bacterial organisms, ancestor's inhibition. Let us imagine a large bacterial population giving rise to a beneficial mutation of small effect. As the benefit is low, the minority population might be rapidly swept by periodic selection or clonal interference, particularly under circumstances of rapid growth. At lower rates of replication, the advantageous variant could be maintained longer, and eventually might slightly increase in frequency in the population (providing more generations than the average). This variant evolves in a way assuring that when the frequency and density of the variant reaches a critical point, and under conditions of slow growth, the variant inhibits (or kills) its ancestor cells. The expected effect is a sudden increase in the possibilities of fixation of the novel trait. Of course, under conditions of high growth rates another more effective variant might arise, but after the ancestor's inhibition process, this variant now has a higher possibility to occur among cells with the small advantageous trait, which is accumulated in the new selected population.

Ancestor's inhibition and kin competition

The phenomenon of generational interference reported by Lemonnier (2008) can be considered as a case of kin competition, that is, competition between relatives with high genetic and ecological relatedness. Intergenerational competition might trigger cooperative effects among organisms of different ages, including colonization success (Poiani, 1994). Indeed, the influence of kin competition in the evolution of dispersal to avoid interferences among relatives and inbreeding depression has been proposed (Gandon et al., 1999). The image of the suicidal ventures of dehiscent seeds that fall only a few feet from the parental plant was provided by Hamilton & May (1977), who proposed dispersal as the most likely evolutionary stable strategy to avoid competition. Ancestor's inhibition constitutes an extreme of kin competition, which might arise when dispersal is not a way out to avoid competition, for instance in highly niche-specialized organisms and, in general, in niche conservatism. For instance, it has been shown that clonal interference increases with larger population sizes and low migration rates (Campos et al., 2008). Essentially, ancestor's inhibition effect will produce a ‘cutting-links’ effect with the ancestor lineage, providing renovated ancestors for the future, and accelerating the establishment of novel adaptive traits within a population (Fig. 2). Such evolutionary acceleration might be particularly critical for the survival of the population in environments in which alien competitors might be present. Indeed, the essential roles of such a type of kin competition, leading to a net increase in the inclusive fitness of the population, are based on substitution of senescent cells by members (eventually ameliorated) of the same lineage. The ancestor's inhibition-based SCDI-mediated kin competition assures that the population replacement corresponds to the offspring. Kin competition requires a kin recognition and or population viscosity (limited dispersal) (West et al., 2002). Of course, in heterogeneous or seasonal environments, ancestor inhibition-based competition (eventually leading to death and cannibalism) does not assure the full elimination of the ancestor and close-to-the ancestor genotypes, and therefore the polymorphic structure of the bacterial population might be maintained (Rozen et al., 2009).


The central circle represents the basic environmental adaptive zone for the bacterial group, occupied by the ancestor (black) population and its closer descendants; later descendants are obliged to locate in neighbor, suboptimal environments, and even later ones colonize more remote places. In the long term, this dispersion may break the optimal correspondence between the basic environment and the bacterial physiology, which is corrected by reoccupancy of the basic compartment, eventually using mechanisms for ancestor's inhibition.

Ancestor's inhibition and the evolutionary units

The question of the units of evolutionary dynamics is a hot topic in the discussion of the method of grouping organisms, not only for purposes of identification or classification but also to understand its role in evolutionary processes (Okasha, 2003, 2006; Reydon, 2005). For instance, groups (as species or clones within the species) are generally considered as products of evolution (lineages) and not units participating in processes (Kluge, 1990; Frost & Kluge, 1994). The conventional wisdom indicates that there is a critical link between the process of speciation and the environmental features that fixes a particular set of genetic sequences, so that long-term environmental isolation is an attractor for speciation. Indeed, the problem of speciation [taking it in a broad biological sense, as the rise of a genetically and ecologically autonomous entity (Mayr et al., 1982)] without relevant environmental change remains a hot question in evolutionary biology. The differentiation of extrinsic environmental barriers and intrinsic genetic factors among isolating mechanisms was already considered by Dobzhansky in 1937. However, environmental isolation is considered to be a prerequisite for the building up of intrinsic genetic isolation mechanisms.

The possibility of ancestor's inhibition by descendents might be theoretically considered as an intrinsic mechanism of isolation, without the need for environmental isolation (on the contrary, it might require a certain degree of sympatry, as in the contact-dependent ancestor's inhibition in E. coli). This hypothesis will support the notion of species as evolutionary dynamic units, involved in the process of evolution (evolverons) (Reydon, 2005), and not only passive products of evolution (phylons) (Kluge, 1990). The important assumption below such statement is that species might have an intrinsic evolutionary potential, giving rise to differentiated variant populations, as mutants are produced in the lineage of a particular cell. Adaptive population radiation is the rapid evolution of a single lineage into a range of genotypes or species, each adapted to a different ecological niche, following a pattern of variation that can be reiterated in a predictable sequence in replicate radiations within a single taxon under identical ecological opportunities (Rainey & Travisano, 1998; Kassen et al., 2004). Indeed, changes in a quantitative trait under weak directional selection naturally lead to macroevolutionary patterns involving recurrent adaptive radiations and extinctions (Ito & Dieckmann, 2007).

Synchronic and diachronic evolutionary units

The importance of time as a key component of the biological differentiation has been reviewed elsewhere (Baquero et al., 2005). The application of the principle of ancestor's inhibition allows a synthesis between the concepts of species (again, in its broader sense, not necessarily taxons) as synchronic or diachronic entities (Salthe, 1985). Synchronic units or entities are identified as existing in a particular, limited time slice, in which the coherence of the group is derived from its ability to interact, influencing each other's behavior, determining the composition of later generations and codetermining the behavior of the evolving entity. In this sense, ancestor's inhibition is a typical effect occurring between synchronic entities, in which descendents interfere with the behavior of the ancestors, under the condition of generational coexistence, as was considered in former paragraphs. The diachronic definition considers species as extending through evolutionary time, and because of this are considered less cohesive. Diachronic entities include all the cells from the past and the present than ever belongs to it; in this case, the coherence of the group is based in an ancestor–descendent relation. Ancestor inhibition, being operatively synchronic, depends on the diachronic relations between cells (ancestor–descendents).

Ancestor's inhibition and the ecological species: theme with variations

The classic Van Valen definition of ecological species (Van Valen, 1988; Prosser et al., 2007) includes the notion of species as ‘closely related set of lineages’ that occupies an adaptive zone, and that evolves separately from all lineages outside its range. Dobzhansky (1937) considers that the species are ‘systems of populations’. Cohan (2006) has developed the concept that bacterial species are composed by ecological species (‘ecotypes’). This implies that bacteria occupy discrete niches and that periodic selection will purge genetic variation within each niche without preventing divergence between the inhabitants of different niches. A good metaphor of the assembly of ecological species in a common integrative species is a musical ‘theme with variations’. We should conceive that the theme corresponds to the basic adaptive zone of the species, probably corresponding to the phenotypes derived from its core genome. The variations correspond to the diversification of the descendents of the basic ancestor ecotype once the basic niche has been fully occupied. These variants evolve because of the need to exploit neighbor novel niches, highly related (frequently as a result of a gradient) to the ‘ancestor’ basic niche. Indeed, a problem is how species maintain coherence among variants, preventing further diversification. The ancestor's inhibition hypothesis indicates the possibility that the original niche, the basic reproductive environment for the species, might be periodically reconquered by its variant descendents, eliminating the formerly established ancestor population and being, in such a way, readapted to its original environment (Fig. 2). Another possibility is that the descendents will evolve towards the reacquisition of their ancestor's genotypes, after a period of divergence, effacing in the long term the nature of the ‘ancestor’ or the ‘descendent’ (Fig. 3). The evolutionary conservatism of the species, the tendency to maintain the old correspondence between the species and its environment, will be assured by either of these processes.


An idealized phage-bacteria system, in which the ‘a’ bacterial population is inhibited by a white phage (hexagon), giving rise to an extended phage progeny. However, the bacteria give rise to a ‘white-phage-resistant’ phenotype (white square) that reconstitutes the original population ‘b’. In turn, the white phage evolves, giving rise to a phage variant ‘c’, which is able to invade the white-resistant bacterial population. The successive adaptive alternance of bacterial and phage variants might produce a black phage, able to select a black-resistant population, which now recovers its susceptibility to a white-type phage. The ‘a’ (ancestor) population is therefore inhibited by a variant produced by a descendant, but in reality there is a long-term equilibrium among bacterial and phage variants, as any bacterial population can be conceived as an ancestor or a descendant.

Ancestor's inhibition and the discontinuity of evolutionary steps

A global phenomenon of ancestor's inhibition in nature might account, at least in part, for the apparent ‘punctuated’ emergence (Eldredge & Gould, 1972) of novel evolutionary units, as species. Frequently, a new species (typically in the fossil record) emerges already fully differentiated from their ancestor species. Indeed, one of the paradoxes of the Darwinian-based gradualism (extrapolationism) is the infrequent identification of ‘intermediates’, as if these intermediates were lost in the evolutionary process. The same mode of reasoning might be applied to the major evolutionary hierarchical transitions in biology (Maynard Smith & Szathmáry, 1997), as the origins of multicellularity, which implies close physical connections among independent cells, probably in extracellular matrixes (Sachs, 2008). In the evolution from single-celled green volvocine algae (Chlamydomonas) to four celled (Basichlamys) and dense colonies with functionally specialized cells (Volvox), cooperative structures are formed but also destroyed (step backs), probably by successful ‘invasions’ of descendant cheater cells (Herron & Michod, 2008; Sachs, 2008). A critical theoretical point in this discussion is to define when descendents are not considered any more as the same type of individual, considering here the term individual to refer to a level or a unit of selection (Michod, 1997). The hypothesis of ancestor's inhibition by descendents could shed light on the concept of an evolutionary individual (Fig. 4). Indeed, more experimental and observational results are needed to support ancestor's inhibition as a general hypothesis; some work should be carried out to ensure that the intuitive appeal of this image might direct further research to problems that might not really exist (Pigliucci & Kaplan, 1996). We have smaller doubts about the underestimated importance of generational coexistence in evolutionary microbiology.


Replication fidelity in asexual organisms assures that the close descendant of a cell (gray triangle) constitute essentially the same individual. The probability of a significant difference from the master cell increases with time with the number of generations, so that after a certain threshold, occurring at a given period of time, the population has changed in such a way that these descendants cannot be considered as belonging to the same individuality of the ancestor. At this time, this new individual might constitute a differential unit of selection.


The amount of experimental data supporting a biological role of the phenomenon of ancestor's inhibition by descendants remains very scarce. Therefore, the aim of this review is not to summarize and organize dispersed information about this topic. On the contrary, our aim was to alert the scientific community of a particular finding (ancestor's inhibition in evolving E. coli populations) that might have considerable explanatory interest in the interpretation of a number of observations in microbial biology. The main idea is that ancestor's inhibition by descendants might act as an accelerator of the possible fixation of a fitness advantage in a particular lineage. Such an advantage could be built up by successive genetic changes, increasing fitness until a threshold in which the ancestor's inhibition effect would act by amplifying the advantage, thereby rapidly imposing a particularly fit variant within the descendent population. In any case, how general is the phenomenon of ancestor's inhibition in evolutionary biology is a matter that needs a general discussion, and certainly lies beyond the possibilities of the authors.


We thank Bruce Levin, the main contributor to the exploration of SCDI in E. coli, for stimulating exciting controversies and wise recommendations (not necessarily followed here) around the main concepts proposed in this review. We thank Ramón Díaz Orejas, for his intellectual generosity and continuous support to this research, and to José Luis Martinez, Victor de Lorenzo, Ginés Morata and Paul Rainey for warmly discussing with us some of the concepts presented in this work; F.B. is a recipient of the Grant FIS06/1008 from the Instituto de Salud Carlos III of the Ministry of Health in Spain.


  • Editor: Ramón Díaz Orejas


View Abstract