INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Biological traits that are not important components of fitness in a particular environment may nonetheless evolve by a variety of mechanisms. Such mechanisms include pleiotropic side effects of adaptive mutations on unselected traits (43), the random fixation of effectively neutral alleles by genetic drift (19), and "hitchhiking" of neutral mutations due to genetic linkage with an adaptive mutation (16, 31). Along with adaptation by natural selection, these more indirect mechanisms of evolution are relevant to such issues as the genetic divergence of populations within a species (8, 43) and the fate of genetically modified organisms in natural environments (25, 36, 37, 40).

Several studies have shown that isolated populations of the same species may evolve different adaptations even to the same selective conditions. Some of the clearest evidence for this phenomenon comes from laboratory experiments using fruitflies (9, 18) and bacteria (42, 43). Alternative pathways of adaptation to a particular selective environment may lead to heterogeneous changes in traits that are not important for fitness in that environment. For example, several populations of organisms that are under selective pressure to use a particular substrate more efficiently may each find a different physiological mechanism to do so. The effects of these different adaptations on the ability of organisms to use some alternative substrate may be positive in some cases, neutral in others, and negative in yet others. As a consequence of these different correlated responses, some populations may be fortuitously well adapted and others may be poorly adapted to environments that contain this alternative substrate.

Of particular relevance to these concerns is the effect that evolutionary adaptation by bioremediative organisms to their intended anthropogenic substrate has on their abilities to grow on, and compete for, naturally occurring substrates. If adaptation to an anthropogenic substrate always leads to a reduction in fitness on natural substrates, then this reduced fitness provides a measure of safety for the release of organisms modified for bioremediation. If, however, such adaptation indirectly enhances fitness on natural substrates, then this enhanced fitness increases the likelihood that the released organisms (and any adverse effects they may cause) will persist in the environment.

In this study, evolving populations of bacteria were used to investigate the effect of adaptation to one carbon source on competitive performance on an alternative substrate. The bacterial strains employed were two natural isolates of the -proteobacteria (genus Burkholderia), each capable of degrading the anthropogenic herbicide 2,4-dichlorophenoxyacetic acid (2,4-D) (32, 41). Experimental evolution occurred in two separate stages, one in which succinate was the sole carbon source and one in which 2,4-D was the limiting substrate (Fig. 1). This two-stage design was employed to allow a period during which any major adaptations to general selective conditions other than the carbon substrate could occur (stage I on succinate) prior to subsequent evolution on the selective substrate of primary interest, 2,4-D. Allowing any such general adaptation first should maximize the degree of substrate-specific adaptation on 2,4-D. Previous evolution experiments also using 2,4-D-degrading bacteria have been conducted to investigate the role of environmental structure in adaptation and the genetic divergence of multiple populations evolving independently in identical habitats (21, 22, 23).

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Derivation of bacterial strains. Horizontal lines indicate periods of evolution, with the selective regimen and sole carbon source indicated above and below, respectively. Each horizontal line represents two independently evolved replicate populations. F and S represent the fast and slow ancestral strains, respectively. For stage I-derived strains, the first letter (F or S) indicates the ancestor and the second letter (B or C) indicates whether stage I evolution occurred in batch (B) or chemostat (C) cultures. The letter r indicates streptomycin resistance (see Materials and Methods). The first two letters for stage II evolution strains indicate their stage I-proximate ancestor, and the third letter (B) indicates stage II evolution in batch culture.

In stage I of the experimental evolution, two ancestral strains were used to found replicate lines that were propagated in batch culture with succinate (a naturally occurring substrate) as the sole carbon source (Fig. 1). One of these strains (strain F) grows relatively fast under laboratory conditions, whereas the other (strain S) grows more slowly (42). In stage II of the evolution experiment, clonal isolates derived from the stage I batch culture lines were used to found replicate lines that were further propagated in batch culture with 2,4-D as the sole carbon source (Fig. 1). Additional stage II batch lines were founded with clones from stage I lines that had undergone evolution in a different selective regimen, chemostat culture (45). (These stage I lines were also involved in a sibling study of the effects of adaptation to either batch culture [an abundant-resource regimen] or chemostat culture [a scarce-resource regimen] on fitness in the alternative selective regimen) (45). The competitive performances of all of these derived lines were then measured relative to those of their immediate ancestors on both succinate and 2,4-D. The results of these experiments demonstrated that adaptation to either substrate can be associated with both improvements and losses in competitive performance on the alternative substrate.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Strains. For simplicity, the designations of the two ancestral strains in this study were changed from those used in previous literature (strain F was TFD3, and strain S was TFD20), in which their 2,4-D catabolic pathways were genetically and phylogenetically characterized (32, 41).

Evolution experiments. (i) Stage I batch evolution. Two clones each from strains F and S were inoculated independently into 10 ml of a mineral salts medium (45) containing 500 µg of succinate per ml and grown to stationary phase while being shaken at 120 rpm. These lines were designated FB1, FB2, SB1, and SB2, with the first and second letters indicating ancestral strain (F or S) and selective regimen (batch), respectively, and with the numbers distinguishing replicate lines (Fig. 1). The four lines were then diluted daily into fresh medium for 75 days (see Table 1 for dilution factors and numbers of generations evolved for all selection experiments). Dilution factors were set to allow complete population replacement in 24 h with minimal time spent in stationary phase. All cultures were maintained at 25°C in all experiments.

(ii) Stage I chemostat evolution. Two initially clonal lines from each ancestor were also evolved in chemostats for 75 days during stage I, where fresh medium containing succinate as the sole carbon source continually flowed into culture vessels and culture volume was kept constant by continuous vacuum removal of excess culture (45). These lines were designated FC1, FC2, SC1, and SC2, where C indicates a chemostat selection history. These chemostat-evolved lines were competed against their ancestors in a different study (which tested for a trade-off between fitness at high and low substrate concentrations [45]) but not in this study. For logistical reasons, only strains that experienced the batch culture regimen during their most recent stage of evolution were competed against their ancestors in this study. Thus, the stage I chemostat-evolved strains served as the proximate ancestors for half of the stage II batch lines but not as evolved strains for analysis in their own right. Use of these strains as stage II ancestors (rather than simply initiating more replicate lines from the stage I batch clones) allowed a greater chance of detecting patterns of adaptation specific to distinct selective histories.

(iii) Stage I clones. Single clones were isolated from each of the eight stage I evolved lines. The four batch-evolved clones (FB1, FB2, SB1, and SB2) were compared to their ancestors in competition experiments described below. In addition, a spontaneous streptomycin-resistant sister clone was selected from each of the four clones FB1, FC1, SB1, and SC1 (denoted by the letter r). Each of these eight clones (FB1, FBr, FC1, FCr, SB1, SBr, SC1, and SCr) was then used to initiate two new lines (distinguished by the numbers 1 and 2) for stage II evolution.

(iv) Stage II batch evolution. Stage II batch lines were propagated as described above for stage I lines except that the growth medium contained 500 µg of 2,4-D per ml rather than succinate and dilution factors varied. After evolution, single clones from each line were isolated for use in competition experiments. Stage II clones were named by adding the letter B to the name of their stage I-proximate ancestor, indicating stage II evolution in batch culture (Fig. 1; Table 2). All strains were stored in medium containing 10% glycerol at 80°C. The identity of each clone as a true evolved descendant of the founding clone (rather than as a contaminant) was confirmed both by streptomycin marker type and by repetitive-extragenic-palindromic-PCR diagnostic fingerprints (46).

Competition experiments. The competitive performance of each clone that evolved in batch culture during its most recent stage of evolution was measured by directly competing a clone with its proximate ancestor of the opposite marker type. These experiments were performed on both selective and alternative substrates under the same conditions in which the clone evolved. Both strains in each pairwise competition were preconditioned for two growth cycles in the competition medium. The strains were then mixed and grown together for several generations (one to three daily dilution cycles). Initial and final relative frequencies of competing strains were estimated by dilution plating onto selective and nonselective agar plates. These frequencies were then used to calculate the selection rate constant (SRC) for each evolved clone (see below), which expresses the amount of change in competitive performance undergone by a clone relative to the competitive performance of its proximate ancestor (see reference 45 for a more detailed description of competition experiments).

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Adaptation to selective substrate. Of the stage I lines propagated on succinate in batch culture, the two S-derived lines showed significantly greater adaptation than did the two F-derived lines (Table 2; Fig. 2). Moreover, this difference is conservative, since the S-derived lines underwent fewer than half the number of doublings (185 generations) during stage I as did the F-derived lines (500 generations). This suggests that slower-growing strain S had more room for improvement in the batch regimen than did strain F. 

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Performance changes of stage I lines. (a) FB lines; (b) SB lines. Mean SRC values are shown, along with 95% confidence intervals, for both the selective substrate (succinate, left column) and the alternative substrate (2,4-D, right column). *, see Table 2.

View larger version (23K): [in this window] [in a new window]   FIG. 3.   Performance changes of stage II F-derived lines. (a) FBB lines; (b) FCB lines. Mean SRC values are shown, along with 95% confidence intervals, for both the selective substrate (2,4-D, left column) and the alternative substrate (succinate, right column).

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Performance changes of stage II S-derived lines. (a) SBB lines; (b) SCB lines. Mean SRC values are shown, along with 95% confidence intervals, for both the selective substrate (2,4-D, left column) and the alternative substrate (succinate, right column).

Competitive performance on alternative substrate. Among the 13 lines that showed significant improvement on their selective substrate, 5 also improved significantly on the alternative substrate (Table 3). Three additional lines had nonsignificant performance increases on the alternative substrate. In contrast, 3 of the 13 significantly adapted lines experienced significant losses of competitive ability on the alternative substrate and two others were suggestive of such losses (Table 3). Overall, 10 lines either showed or suggested correlated improvements on both substrates while 9 showed or suggested fitness losses on the alternative substrate (Table 3). Among the six sets of evolved lines, only the two SB replicate lines exhibited a homogeneous pattern of performance on both selective and alternative substrates. The other five sets showed (or suggested) both performance losses and performance gains among their lines on the alternative substrate.

Substrate-specific adaptation. The lines in this study may have increased their fitness by adaptations largely specific to their growth substrate during evolution, or alternatively, they may have improved by adapting to some other aspect of the selective regimen, such as the concentrations of various minerals in the medium. The latter possibility is especially relevant to the stage I lines. If most adaptation was not substrate specific, then evolved strains should be approximately equally superior to their ancestors on both selective and alternative substrates. In this study, however, many lines showed negative responses on their alternative substrate, indicating that their adaptation was largely substrate specific. Moreover, among those lines that improved on both substrates, in several cases the degrees of improvement were quite different for the two substrates, which also belies the general adaptation hypothesis.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Population dynamics of adaptation. The evolutionary changes in competitive performance on nonselective substrates (resulting from adaptation to a specific selective substrate) may be caused either by pleiotropic effects of adaptive mutations (43) or by the random drift of effectively neutral alleles (19). The latter possibility can be readily dismissed with respect to these experiments. Only a very small minority of mutations that affect fitness in a given environment are expected to be beneficial rather than harmful. It is much easier to disrupt than to improve the performance of a complex entity. That is, if one considers mutations that are selectively neutral in one environment (say, 2,4-D medium) but that affect fitness in an alternative environment (say, succinate medium), then the vast majority of the mutations should reduce, rather than improve, fitness in the alternative environment.

Multiple adaptive pathways. The results of this study clearly show that replicate evolving populations often achieved adaptation to their selective substrate by different mechanisms. This inference follows from the facts that in three of the four stage II sets (each containing four independent replicate lines), at least one line showed significant correlated improvements on both the selective and the alternative substrate and that at least one other line showed a significant loss of performance on its alternative substrate after adaptation to its selective substrate (Table 2; Fig. 3 and 4). In other words, the variation in competitive fitness among replicate populations on alternative substrates (standard deviation = 1.3394) was significantly greater (F = 14.581, 38 df, P < 0.0001 [F test statistic to compare variances]) than such variation on selective substrates (standard deviation = 0.3508) (43).

Relevance to the adaptation of genetically modified organisms in natural habitats. In the discussion of relative benefits and risks of releasing genetically modified organisms into nature, it has often been argued that genetically altered microorganisms will not persist because they are inherently less fit than indigenous competitors (6, 10, 11). Lenski (25) reviewed various arguments for why genetically modified organisms might theoretically be less fit than their natural counterparts, and he summarized some experiments with bacteria that bear on those arguments.