Many soil-inhabiting bacteria are known to produce secondary metabolites that can suppress microorganisms competing for the same resources. The production of antimicrobial compounds is expected to incur fitness costs for the producing bacteria. Such costs form the basis for models on the co-existence of antibiotic-producing and non-antibiotic producing strains. However, so far studies quantifying the costs of antibiotic production by bacteria are scarce. The current study reports on possible costs, for antibiotic production by Pseudomonas fluorescens Pf0-1, a soil bacterium that is induced to produce a broad-spectrum antibiotic when it is confronted with non-related bacterial competitors or supernatants of their cultures.
Methodology and Principal Findings
We measured the possible cost of antibiotic production for Pseudomonas fluorescens Pf0-1 by monitoring changes in growth rate with and without induction of antibiotic production by supernatant of a bacterial competitor, namely Pedobacter sp.. Experiments were performed in liquid as well as on semi-solid media under nutrient-limited conditions that are expected to most clearly reveal fitness costs. Our results did not reveal any significant costs for production of antibiotics by Pseudomonas fluorescens Pf0-1. Comparison of growth rates of the antibiotic-producing wild-type cells with those of non-antibiotic producing mutants did not reveal costs of antibiotic production either.
Based on our findings we propose that the facultative production of antibiotics might not be selected to mitigate metabolic costs, but instead might be advantageous because it limits the risk of competitors evolving resistance, or even the risk of competitors feeding on the compounds produced.
Citation: Garbeva P, Tyc O, Remus-Emsermann MNP, van der Wal A, Vos M, Silby M, et al. (2011) No Apparent Costs for Facultative Antibiotic Production by the Soil Bacterium Pseudomonas fluorescens Pf0-1. PLoS ONE 6(11): e27266. doi:10.1371/journal.pone.0027266
Editor: Ramy K. Aziz, Cairo University, Egypt
Received: June 8, 2011; Accepted: October 13, 2011; Published: November 16, 2011
Copyright: © 2011 Garbeva et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work is financed by The Netherlands Organization for Scientific Research. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Interference competition is an important strategy of bacterial strains to establish and maintain themselves within microbial communities . A well-known mechanism of bacterial interference competition is the production of antibiotics . These secondary metabolites can be targeted against more or less closely related strains and species (e.g. bacteriocins) or against a wide range of competitors (e.g many polyketides) , , . Theoretical models have demonstrated that, instead of decreasing diversity by leaving only the most aggressive strains, microbial warfare could actually promote diversity, with dynamic coexistence of many strains differing in their antibiotic production and sensitivity profiles , , , . These results are obtained when it is assumed that both resistance to- and production of antibiotics come at a fitness cost, resulting in a reduced growth rate. The ecological trade-offs involved in investment in killing, resisting or outgrowing competing strains is thus predicted to maintain diversity.
Because of its profound relevance to human health, the fitness cost of bacterial resistance to antibiotics has received far more attention than the fitness cost of bacterial antibiotic production. Whereas it has emerged that the majority of bacterial antibiotic resistance mechanisms comes at a fitness cost , as predicted by theory , few studies have examined the cost of antibiotic production. Indications for biological costs of antibiotic production or antibiotic resistance are generally obtained by comparison of relative fitness of wild-type strains with that of antibiotic-negative mutant strains (e.g. , ). However, mutations causing loss of antibiotic production may cause additional changes in the bacterial phenotype , .
Previously, we have reported on competitor-induced triggering of broad-spectrum antibiotic production in fluorescent pseudomonads , . The soil isolate Pseudomonas fluorescens Pf0-1 exhibits antibiotic activity only when it is confronted with specific phylogenetically unrelated competitors (e.g. Pedobacter sp.) or their supernatant indicating that it can distinguish interspecific competition from intraspecific competition . Although the structure of the antimicrobial compound has not yet been elucidated, upregulated genes during confrontation with competitors point at the synthesis of a polyketide-like compound.. In addition, we have shown that it has broad-spectrum activity, acting against both Gram-positive and Gram-negative bacteria as well as against fungi .
Competitor-dependent induction of antibiotic production allows for another possibility to examine costs of antibiotic production namely by comparing growth rates of wild-type bacteria with and without induction of antibiotic production. In the current study we used both approaches to quantify the possible fitness cost of antibiotic production in P. fluorescens Pf0-1: 1) comparison of the growth rate of the wild-type with and without induction of antibiotic production and 2) comparison of the growth rate of wild-type and non-antibiotic producing mutants under conditions that induce antibiotic production. All experiments were performed using nutrient-poor media, as soil-dwelling bacteria typically experience a scarcity of easily degradable carbon resources , . Moreover, growth limiting conditions represent a situation in which the costs of antibiotic production should be most pronounced as there is no surplus of energy resources . Costs were measured in liquid culture as well as on semi-solid medium in an incubation chamber, allowing quantification of micro-colony growth.
Materials and Methods
Bacterial and fungal cultures used
The bacterial strains used in this study are Pseudomonas fluorescens Pf0-1 (Gamma-proteobacteria) which was isolated from an agricultural loam soil in Sherborn, Massachusetts, USA  and Pedobacter sp. V48 (Sphingobacteria), which was isolated from a coastal dune site in The Netherlands . The strains were pre-cultured from frozen glycerol stocks on 1/10 strength Tryptic Soy Broth agar (TSBA) . In addition to the parental strain, deletion mutant ΔPfl01_3463-66 (with deletion of Pfl01_3463-3464: two branched-chain alpha-keto acid dehydrogenase E1 components; Pfl01_3465: branched-chain alpha-keto acid dehydrogenase subunit E2 and Pfl01_3466: dihydrolipoamide dehydrogenase) which is unable to produce the broad-spectrum antibiotic was used as well . The fungal isolate Rhizoctonia solani anastomosis group 2.2IIIB, a plant-pathogenic basidiomycete, was used as bioindicator for production of broad-spectrum antimicrobial compounds .
Preparation of cell-free Pedobacter supernatant
Cell-free supernatant was prepared by centrifugation (16,000 x g for 5 min) followed by filtration (Spin-X 0.22 µm filters; Corning Costar, Cat# 8160) of over-night cultures of Pedobacter strain V48 grown in 1/10th strength Tryptic Soy Broth at 20oC. An aliquot of the cell-free supernatant was boiled for 10 minutes and was used as a control. Boiling the cell-free supernatant for 10 minutes was sufficient to destroy the signalling compound(s) that trigger antibiotic production in P. fluorescens Pf0-1.
Determination of growth of Pf0-1 in nutrient-poor liquid media
The effect of cell-free supernatant from Pedobacter sp V48 on the growth rate of Pf0-1 was determined in a nutrient-poor liquid medium (0.5 gL−1 KH2PO4; 0.1 gL−1 BD BactoTM Yeast extract (Cat# 210934) and 0.1 gL−1(NH4)2SO4; adjusted to pH 6.5). This medium was supplemented with 10% (v/v) boiled (control) or unboiled Pedobacter supernatant, respectively. P. fluorescens Pf0-1 was inoculated to an optical density (OD) (600 nm) of 0.02 which corresponds to 2.6×106 cells/ml−1. The 50 ml cultures were incubated for 24 hours at 20°C shaken at 200 rpm and 1 ml samples were taken every hour for OD measurements and viable counts. Additionally, the growth rate of P. fluorescens Pf0-1 wild-type and mutants ΔPfl01_3463 (both supplemented with 10% (v/v) boiled (control) or unboiled Pedobacter supernatant) were measured in 96-well plates (Greiner bio-one, Cat# 655180) using Synergy Microplate Reader. The OD measurements were performed every 30 minutes for total period of 8.5 hours.
Agar-incubation chambers for observation of bacterial growth on semi-solid medium
1 ml of P. fluorescens Pf0-1 overnight culture was centrifuged for 3 min at 16,000 g. The cells were resuspended in 10 mM phosphate buffer (pH 6.5) containing 10% non-boiled or boiled cell-free supernatant of Pedobacter sp V48 to an optical density of 0.002 at 600 nm; so, the initial number of cells per ml was 10 times lower than in the liquid growth experiments. The lower number was used as it allows distinguishing individual bacteria on the agar slices (see below). An aliquot of 280 µl of P. fluorescens Pf0-1 bacterial suspension was equally spread on a thin layer (2 mm) of water agar containing 5 g L−1 NaCl; 1 g L−1 KH2PO4 and 0.1g L−1 (NH4)2SO4 adjusted to pH 6.5 as previously described . Since bacterial growth was observed on this water-agar medium no extra substrate (yeast extract) was added . Apparently, the substrates that were added with the 10% Pedobacter supernatant and the substrates that became available after autoclaving of agar were sufficient for growth. After inoculation the plates were dried for about 5 minutes in a flow cabinet and 3 slices of 1 cm2 were cut and carefully transferred to incubation chambers. The incubation chamber described in Figure 1 is adapted from Reinhard and Van der Meer  and Robin Tecon (unpublished data).
Between the spacers 3 pre-inoculated water-agar (WA) slices of 1 cm2 are placed. A glass coverslip is slightly pressed on top of the agar slices. The sides of the incubation chamber are sealed with parafilm.
Microscopical counting and data analysis
Twenty randomized pictures at 400-fold magnification representing an area of 1 mm2 were taken of each slice with an Axio Imager M1 microscope (Carl Zeiss, Oberkochen, Germany) under phase-contrast illumination and an AxioCam MRm camera. Microscopic enumerations were performed every hour for a period of 9 hours at room temperature. During this period microcolonies remained two-dimensional, i.e. no stacking took place. Digital images were analyzed using the AXIO VISION v4.7 Software (Carl Zeiss Imaging Solutions GmbH, Germany) for enumeration and area determination of bacterial colonies.
Extraction of antimicrobial compound and test for inhibition
P. fluorescens Pf0-1 overnight liquid cultures exposed to 10% boiled or unboiled supernatant of Pedobacter sp V48 (initial volume 35 ml) were acidified with trifluoroacetic acid (0.1% (v/v)), mixed with 2 volumes of ethylacetate and shaken vigorously for 5 min. as described by Raaijmakers et al. . After overnight incubation at −20°C the unfrozen liquid (ethylacetate) fraction containing the active compounds was carefully transferred to a new flask and dried under constant air-flow. The dried extract was dissolved in 150 µl 50% (v/v) methanol and tested for inhibition of the bacterial isolate Pedobacter sp V48 and the soil borne plant-pathogenic fungus Rhizoctonia solani as described by Garbeva et al. .
RNA extraction, cDNA synthesis and real-time PCR
RNA was extracted from P. fluorescens Pf0-1 grown in nutrient-poor liquid medium that had been exposed to 10% boiled or non-boiled supernatant of Pedobacter sp V48 as described above. Cells for RNA extraction were collected at 4 time points (3 h, 6 h, 8 h and 24 h) and diluted with sterile phosphate buffer to the same optical density (600 nm) to obtain equal amounts of cells. The RNA extraction and cDNA synthesis were performed as described previously by Garbeva et al .
For real-time PCR assessment of expression of genes involved in antibiotic production two different primer combinations were used: (1) primer combination 3463F835 (GAT TTT TAC GCG GTC TAC GC) and 3463R1036 (TGA TCA GGT TGC TGT TTC AGG) amplifying 206 bp from gene Pf01_3463 encoding the two branched-chain alpha-keto acid dehydrogenase E1 component and (2) primer combination 3465T1F (CAG GGC CCG ATG GTT GC) and 3465T1R (TTG CTT TTT GTG CCG CGC TCG) amplifying 348 bp from gene Pf01_3465 encoding branched-chain alpha-keto acid dehydrogenase subunit E2. As a control, a 210 bp fragment of the house keeping 16S rDNA gene was amplified using primer combination 16SPf01F (TTG GGA GCC TTG AGC TCT TA) and 16SPf01R (AAG GCA CCA ATC CAT CTC TG). Real-time PCR was performed using a Corbett Research Rotor-Gene 3000 thermal cycler (Westburg, Leusden, The Netherlands) with the following conditions: an initial cycle of 95°C for 15 min followed by 40 cycles of: 95°C for 15 sec; 56°C for 50sec and 72°C for 50 sec. Standard curves were established for each primer combination.
All experiments were performed in triplicate. Differences in optical densities of liquid cultures between treatments were tested for significance for each time interval by one-way analysis of variance. Bacterial viable count data that were used to calculate the maximum yield (maximum number of colony forming units per ml) were also log-transformed prior to one-way analysis of variance. For the analysis of microcolony development the data were log transformed and the slopes of the regression lines were compared in GraphPad Prism 5 (GraphPad Software, Inc., CA, USA) using a two tailed t-test. The statistical analyses of quantitative real-time PCR data were carried out with XLStat 2010 (Addinsoft, New York, USA) using a two-tailed t-test. Data were considered to be statistically different at p≤0.05.
P. fluorescens Pf0-1 antibiotic production and growth rate
The OD measurements indicated that Pf0-1 grew only a short period (1–2 hr) exponentially in nutrient-poor liquid media followed by linear increase (Figure 2; Figure S1). This rapid decline in growth rates occurred for all treatments and stationary phase was reached after 8 hours. At none of the time intervals (performed on 96-well plates every 30 min) were there significant differences in OD between cultures exposed to boiled and unboiled Pedobacter supernatant (Figure 2). The growth rate of Pf0-1 mutant Δ3463, which is deficient in the production of the broad-spectrum antimicrobial compound triggered by Pedobacter supernatant, was also compared with that of the wild-type strain in the presence of Pedobacter supernatant. Again, no differences in OD were observed (Figure 2).
: Pf01wt+CFS - P. fluorescens Pf0-1 wild type with 10% cell-free supernatant from Pedobacter sp. V48; Pf01wt+BCFS - P. fluorescens Pf0-1 wild type with 10% boiled cell-free supernatant from Pedobacter sp. V48; Δ3463+CFS – P. fluorescens Pf0-1 mutant Δ3463 with 10% cell-free supernatant from Pedobacter sp. V48 and Δ3463+BCFS – P. fluorescens Pf0-1 mutant Δ3463 with 10% boiled cell-free supernatant from Pedobacter sp. V48. The measurements were performed every 30 min for total period of 8.5 h. Symbols represent means of 3 replicate measurements; error bars represent standard deviations.
Similar observations were made for the 50 ml cultures where OD measurements were performed every hour (Figure S1). For the latter cultures, the maximum yield (highest number of viable cells per ml of growth medium) was calculated and again no differences were found between the treatments (Table 1.).
The liquid cultures used for growth rate determinations were extracted at the end of the incubation to confirm that the wild-type Pf0-1 was induced by Pedobacter supernatant to produce antibiotics. The presence of antibiotic activity in the extracts was tested by determining the effect of the extracts on growth of Pedobacter sp V48 and the soil-borne pathogenic fungus Rhizoctonia solani. Inhibition of Rhizoctonia and Pedobacter was observed only with the extracts from wild type Pf0-1 cultures exposed to 10% cell-free Pedobacter supernatant (Figure S2A, B). There was no such inhibition by the extracts obtained from Pf0-1 cultures exposed to 10% boiled cell-free Pedobacter supernatant or from Pf0-1 mutant Δ3463 cultures exposed to 10% cell-free Pedobacter supernatant.
Expression of genes involved in antimicrobial compound production
At four time points (t = 3 h, 6 h, 8 h and 24 h) during growth in liquid medium, P. fluorescens Pf0-1 cells were collected for RNA isolation, cDNA synthesis and quantitative RT-PCR. Primers targeting genes Pfl01_3463 and Pfl01_3465 were used for quantitative RT-PCR. Genes Pfl01_3463 and Pfl01_3465 encode branch-chain alpha-keto acid dehydrogenase E1 components and branched-chain alpha-keto acid dehydrogenase subunit E2 that were previously demonstrated to be essential for the production of broad-spectrum antimicrobial activity by P. fluorescens Pf0-1 . Quantitative RT-PCR revealed that triggering of Pfl01_3463 and Pfl01_3465 genes expression by Pedobacter supernatant was already apparent after three hours, as there was a significant (3.16 and 2.57 fold respectively) increase in comparison with the expression of these genes in cultures of Pf0-1 exposed to 10% boiled Pedobacter supernatant (Table 2). The expression of genes Pfl01_3463 and Pfl01_3465 was always at least two-fold higher in the Pf0-1 cultures exposed to 10% cell-free Pedobacter supernatant than in the controls (Pf0-1 cultures exposed to 10% boiled cell-free Pedobacter supernatant).
Effect of antibiotic production on P. fluorescens Pf0-1 growth rate on agar and microcolony morphology
Using water-agar incubation chambers we determined growth rate and size of Pf0-1 colonies with and without induction of antibiotic production by Pedobacter supernatant. Growth (increase of surface areas micro-colonies) was exponential during the period of examination (Figure 3). There was no significant difference in growth rates between the two treatments (Figure 3A). No significant differences in growth rates were observed between the colonies of wild type and mutant Δ3463 strains either (Figure 3B). In fact, the average colony size of Pf0-1 supplied with 10% cell-free supernatant from Pedobacter sp V48 was slightly (but not significantly) bigger than that of Pf0-1 supplied with boiled cell-free supernatant. However, there was a clear difference in colony morphology. After 7 hours of incubation, wild-type Pf0-1 exposed to 10% Pedobacter supernatant started to form spherical colonies, a phenomenon not observed for Pf0-1 exposed to boiled Pedobacter supernatant or for Pf0-1 mutant Δ3463 exposed to unboiled Pedobacter supernatant (Figure 4).
(A) white squares -wild type P. fluorescens Pf0-1 exposed to 10% cell-free supernatant from Pedobacter sp. V48 and black squares- wild type P. fluorescens Pf0-1 exposed to 10% boiled cell-free supernatant from Pedobacter sp. V48; (B) white squares -wild type P. fluorescens Pf0-1 exposed to 10% cell-free supernatant from Pedobacter sp. V48 and black squares- mutant Δ3463 exposed to 10% cell-free supernatant from Pedobacter sp. V48. Squares represent the means of micro-colony sizes and the error bars represent the standard deviations. Statistical analysis revealed no significant differences (p>0.05) between the growth rates (slopes) of the treatments.
Many, if not most, bacteria produce a range of secondary metabolites that can target competing microorganisms , . Natural bacterial populations have been found to consist of a wide variety of genotypes that differ in their ability to both suppress and withstand conspecifics , , , . A trade-off between an advantage in growth (resource or scramble competition) and an advantage in ‘killing capacity’ or ‘resistance capacity’ (interference competition) lies at the basis of theoretical models attempting to explain the coexistence of strains differing in antibiotic production and sensitivity , , , , , . The reason that not all strains evolve to produce antibiotics thus is explained by the fact that the production of these compounds (and their corresponding immunity factors) incurs a metabolic cost. In competition in a structured environment, antibiotic producing cells will displace sensitive (non-antibiotic producing) cells, whereas sensitive (non-antibiotic producing) cells have a growth rate advantage over resistant (non-antibiotic producing) cells that in turn displace antibiotic producing cells because they do not carry the cost of antibiotic production .
Previously, it was hypothesized that the facultative- rather than the constitutive production of antibiotics represents a cost-effective strategy, as the antibiotic compound is only produced in situations where it is needed . Our finding that the cost of antibiotic production in the P. fluorescens Pf0-1 system is not significant is not in line with the cost-based assumption on basis of which theoretical models aim to explain how microbial warfare can promote microbial diversity. If these costs are truly insignificant then why do not all strains produce antimicrobial compounds constitutively? Two alternative ecological trade-offs could be envisaged to be at work. First, facultative antibiotic production could prevent competitors evolving resistance to the antibiotic by reducing exposure . It is evident from clinical studies that increased exposure to antibiotics (through patient consumption) can result in higher resistance levels in a pathogen population (e.g. ). Second, it has recently emerged that many bacteria can actually subsist on antibiotic compounds , . Although antibiotic production could inhibit the growth of some strains competing for resources, it could promote the growth of others. It is presently unknown how important both mechanisms are in bacterial populations in soil but they certainly seem worthy of future attention.
Whilst we did not observe significant costs of antibiotic production here, biological costs associated with antibiotic resistance have been reported to vary from significant (e.g. , , ), to no-cost (e.g. , , , ) to even enhanced fitness (e.g. ). This variation in costs might be explained by the fact that the genetic systems underlying antibiotic resistance are diverse and furthermore might not be readily comparable to those that underlie antibiotic production (including the production of a molecule conferring autoimmunity). Mutations conferring resistance arising in sensitive cells often modify the molecule targeted by the antibiotic in such a way that, although the cell is protected from the antibiotic, its function is severely compromised . Such non-additive, pleiotropic fitness costs often are severe. With antibiotic production on the other hand, it could be hypothesized that the cells mainly bear the additive, metabolic cost of the production of the antibiotic and the immunity molecules (which must be small compared to the sum of all other molecules produced by the cell) and suffer less from pleiotropic costs. In addition, compensatory mutations that mitigate pleiotropic fitness costs have been identified in resistant bacteria . It could be that compensatory mutations are more readily accessible for genetic systems mediating antibiotic production than they are for resistance mutations thus lowering the fitness cost in the former.
Although we did not observe costs for antibiotic production, clear changes in P. fluorescens Pf0-1 colony morphology during antibiotic production were apparent from the agar chamber experiments. Antibiotic producing P. fluorescens Pf0-1 cells formed spherical colonies whereas the non-producing mutant as well as wild-type strain in the control situation did form irregular shaped colonies. Recently it was reported that different P. fluorescens colony morphology variants have distinct metabolic profiles . It seems plausible that the spherical colony-shape of Pf0-1 here is a response to the produced antibiotic and not to a signal of Pedobacter, as the mutant deficient in the production of the antibiotics did not produce spherical colonies in the presence of Pedobacter supernatant. Formation of such spherical colonies may coincide with antibiotic production to obtain the highest protection against antagonising organisms. Further studies are needed to understand the mechanism and the biological relevance of changes in colony morphology of P. fluorescens Pf0-1 during antibiotic production.
Bacterial growth (Optical Density 600 nm) in 50 ml nutrient-poor liquid cultures.: Pf01wt+CFS - wild type P. fluorescens Pf0-1 with 10% cell-free supernatant from Pedobacter sp. V48; Pf01wt+BCFS - wild type P. fluorescens Pf0-1 with 10% boiled cell-free supernatant from Pedobacter sp. V48; Δ3463+CFS – P. fluorescens Pf0-1 mutant Δ3463 with 10% cell-free supernatant from Pedobacter sp. V48. The measurements are presented as mean ± SD (n = 3 replicates).
(A) Antagonist assay against the fungus Rhizoctonia solani with extracts from liquid cultures (Figure S1) 1- Mutant Δ3463 with 10% cell-free supernatant from Pedobacter sp. V48; 2- wild type P. fluorescens Pf0-1 with 10% cell-free supernatant from Pedobacter sp. V48; 3- wild type P. fluorescens Pf0-1 with 10% boiled cell-free supernatant from Pedobacter sp. V48 and C- control 50% methanol. (B) Effect of different extracts from liquid cultures (Figure S1) on growth of Pedobacter sp. V48 (growing on 1/10 TSBA for 24 h) 1.Ext.wt- extract from wild type P. fluorescens Pf0-1 with 10% cell-free supernatant from Pedobacter sp. V48; 2. Ext.mut- extract from mutant Δ3463 with 10% cell-free supernatant from Pedobacter sp. V48 and 3.Con- extract from wild type P. fluorescens Pf0-1 with 10% boiled cell-free supernatant from Pedobacter sp. V48. 4. PdV48- Pedobacter sp. V48 without any extract. Data are presented as mean ± SD (n = 3 replicates). * - Indicates significant reduction of Pedobacter sp. V48 CFU as compared to other treatments (p<0.05) as analyzed by one-way ANOVA followed by Tukey`s HSD test.
The authors thank Robin Tecon for his help in designing the agar-incubation chamber and Saskia Gerards for her help in antimicrobial compound extraction. This is publication No. 5049 of the Netherlands Institute of Ecology (NIOO-KNAW).
Conceived and designed the experiments: PG WdB. Performed the experiments: OT MNPR-E. Analyzed the data: PG OT. Contributed reagents/materials/analysis tools: AvdW MS. Wrote the paper: PG MV WdB.
- 1. Hibbing ME, Fuqua C, Parsek MR, Peterson SB (2010) Bacterial competition: surviving and thriving in the microbial jungle. Nature Reviews Microbiology 8: 15–25.
- 2. Raaijmakers JM, Vlami M, de Souza JT (2002) Antibiotic production by bacterial biocontrol agents. Antonie Van Leeuwenhoek International Journal of General and Molecular Microbiology 81: 537–547.
- 3. Riley MA, Wertz JE (2002) Bacteriocin diversity: ecological and evolutionary perspectives. Biochimie 84: 357–364.
- 4. Riley MA, Goldstone CM, Wertz JE, Gordon D (2003) A phylogenetic approach to assessing the targets of microbial warfare. Journal of Evolutionary Biology 16: 690–697.
- 5. Challis GL, Hopwood DA (2003) Synergy and contingency as driving forces for the evolution of multiple secondary metabolite production by Streptomyces species. Proceedings of the National Academy of Sciences of the United States of America 100: 14555–14561.
- 6. Czaran TL, Hoekstra RF (2003) Killer-sensitive coexistence in metapopulations of micro-organisms. Proceedings of the Royal Society B-Biological Sciences 270: 1373–1378.
- 7. Czaran T, Hoekstra RF (2007) A spatial model of the evolution of quorum sensing regulating bacteriocin production. Behavioral Ecology 18: 866–873.
- 8. Gardner A, West SA (2004) Spite and the scale of competition. Journal of Evolutionary Biology 17: 1195–1203.
- 9. Andersson DI, Hughes D (2010) Antibiotic resistance and its cost: is it possible to reverse resistance? Nature Reviews Microbiology 8: 260–271.
- 10. Coustau C, Chevillon C, ffrench-Constant R (2000) Resistance to xenobiotics and parasites: can we count the cost? Trends in Ecology & Evolution 15: 378–383.
- 11. Binet R, Maurelli AT (2005) Fitness cost due to mutations in the 16S rRNA associated with spectinomycin resistance in Chlamydia psittaci 6BC. Antimicrobial Agents and Chemotherapy 49: 4455–4464.
- 12. Ender M, McCallum N, Adhikari R, Berger-Bachi B (2004) Fitness cost of SCCmec and methicillin resistance levels in staphylococcus aureus. Antimicrobial Agents and Chemotherapy 48: 2295–2297.
- 13. Bjorkman J, Hughes D, Andersson DI (1998) Virulence of antibiotic-resistant Salmonella typhimurium. Proceedings of the National Academy of Sciences of the United States of America 95: 3949–3953.
- 14. Bjorkman J, Andersson DI (2000) The cost of antibiotic resistance from a bacterial perspective. Drug Resistance Updates 3: 237–245.
- 15. Garbeva P, de Boer W (2009) Inter-specific Interactions Between Carbon-limited Soil Bacteria Affect Behavior and Gene Expression. Microbial Ecology 58: 36–46.
- 16. Garbeva P, Silby MW, Raaijmakers JM, Levy SB, de Boer W (2011) Transcriptional and antagonistic responses of Pseudomonas fluorescens Pf0-1 to phylogenetically different bacterial competitors. ISME Journal 5: 973–985.
- 17. Alden L, Demoling F, Baath E (2001) Rapid method of determining factors limiting bacterial growth in soil. Applied and Environmental Microbiology 67: 1830–1838.
- 18. Demoling F, Figueroa D, Baath E (2007) Comparison of factors limiting bacterial growth in different soils. Soil Biology & Biochemistry 39: 2485–2495.
- 19. Anderl JN, Zahller J, Roe F, Stewart PS (2003) Role of nutrient limitation and stationary-phase existence in Klebsiella pneumoniae biofilm resistance to ampicillin and ciprofloxacin. Antimicrobial Agents and Chemotherapy 47: 1251–1256.
- 20. Compeau G, Alachi BJ, Platsouka E, Levy SB (1988) Survival of rifampin-resistant mutants of Pseudomonas fluorescens and pseudomonas putida in soil system. Applied and Environmental Microbiology 54: 2432–2438.
- 21. de Boer W, Wagenaar AM, Gunnewiek P, van Veen JA (2007) In vitro suppression of fungi caused by combinations of apparently non-antagonistic soil bacteria. Fems Microbiology Ecology 59: 177–185.
- 22. Reinhard F, Van der Meer JR (2011) Microcolony growth assays. Microbiology of Hydrocarbons, Oils, Lipids.
- 23. Raaijmakers JM, Bonsall RE, Weller DM (1999) Effect of population density of Pseudomonas fluorescens on production of 2,4-diacetylphloroglucinol in the rhizosphere of wheat. Phytopathology 89: 470–475.
- 24. Ott EM, Muller T, Muller M, Franz C, Ulrich A, et al. (2001) Population dynamics and antagonistic potential of enterococci colonizing the phyllosphere of grasses. Journal of Applied Microbiology 91: 54–66.
- 25. Vos M, Velicer GJ (2009) Social Conflict in Centimeter and Global-Scale Populations of the Bacterium Myxococcus xanthus. Current Biology 19: 1763–1767.
- 26. Davelos AL, Kinkel LL, Samac DA (2004) Spatial variation in frequency and intensity of antibiotic interactions among streptomycetes from prairie soil. Applied and Environmental Microbiology 70: 1051–1058.
- 27. Hsu SB, Waltman P (1997) Competition between plasmid-bearing and plasmid-free organisms in selective media. Chemical Engineering Science 52: 23–35.
- 28. Brown SP, Inglis RF, Taddei F (2009) Evolutionary ecology of microbial wars: within-host competition and (incidental) virulence. Evolutionary Applications 2: 32–39.
- 29. Kerr B, Riley MA, Feldman MW, Bohannan BJM (2002) Local dispersal promotes biodiversity in a real-life game of rock-paper-scissors. Nature 418: 171–174.
- 30. Bergman M, Huikko S, Pihlajamaki M, Laippala P, Palva E, et al. (2004) Effect of macrolide consumption on erythromycin resistance in Streptococcus pyogenes in Finland in 1997-2001. Clinical Infectious Diseases 38: 1251–1256.
- 31. D'Costa VM, McGrann KM, Hughes DW, Wright GD (2006) Sampling the antibiotic resistome. Science 311: 374–377.
- 32. Dantas G, Sommer MOA, Oluwasegun RD, Church GM (2008) Bacteria subsisting on antibiotics. Science 320: 100–103.
- 33. Andersson DI (2006) The biological cost of mutational antibiotic resistance: any practical conclusions? Current Opinion in Microbiology 9: 461–465.
- 34. Andersson DI (2003) Persistence of antibiotic resistant bacteria. Current Opinion in Microbiology 6: 452–456.
- 35. Kanai K, Shibayama K, Suzuki S, Wachino JI, Arakawa Y (2004) Growth competition of macrolide-resistant and susceptible Helicobacter pylori strains. Microbiology and Immunology 48: 977–980.
- 36. Kugelberg E, Lofmark S, Wretlind B, Andersson DI (2005) Reduction of the fitness burden of quinolone resistance in Pseudomonas aeruginosa. Journal of Antimicrobial Chemotherapy 55: 22–30.
- 37. Zhang QJ, Sahin O, McDermott PF, Payot S (2006) Fitness of antimicrobial-resistant Campylobacter and Salmonella. Microbes and Infection 8: 1972–1978.
- 38. Workentine ML, Harrison JJ, Weljie AM, Tran VA, Stenroos PU, et al. (2010) Phenotypic and metabolic profiling of colony morphology variants evolved from Pseudomonas fluorescens biofilms. Environmental Microbiology 12: 1565–1577.