UBC Faculty Research and Publications

Experimental demonstration of ecological character displacement Tyerman, Jabus G; Bertrand, Melanie; Spencer, Christine C; Doebeli, Michael Jan 30, 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12862_2007_Article_603.pdf [ 261.68kB ]
JSON: 52383-1.0167816.json
JSON-LD: 52383-1.0167816-ld.json
RDF/XML (Pretty): 52383-1.0167816-rdf.xml
RDF/JSON: 52383-1.0167816-rdf.json
Turtle: 52383-1.0167816-turtle.txt
N-Triples: 52383-1.0167816-rdf-ntriples.txt
Original Record: 52383-1.0167816-source.json
Full Text

Full Text

ralssBioMed CentBMC Evolutionary BiologyOpen AcceResearch articleExperimental demonstration of ecological character displacementJabus G Tyerman*1, Melanie Bertrand1, Christine C Spencer2 and Michael Doebeli1,3Address: 1Dept. Zoology & Centre for Biodiversity, University of British Columbia, 6270 University Blvd., Vancouver, BC, V6T 1Z4 Canada, 2Dept. Ecology and Evolutionary Biology, University of Toronto, 25 Harbord St., Toronto, Ontario, M5S 3G5, Canada and 3Dept. Mathematics, University of British Columbia, 6270 University Blvd., Vancouver, BC, V6T 1Z4 CanadaEmail: Jabus G Tyerman* - tyerman@zoology.ubc.ca; Melanie Bertrand - bertrand@zoology.ubc.ca; Christine C Spencer - cspencer@eeb.utoronto.ca; Michael Doebeli - doebeli@zoology.ubc.ca* Corresponding author    AbstractBackground: The evolutionary consequences of competition are of great interest to researchersstudying sympatric speciation, adaptive radiation, species coexistence and ecological assembly.Competition's role in driving evolutionary change in phenotypic distributions, and thus causingecological character displacement, has been inferred from biogeographical data and measurementsof divergent selection on a focal species in the presence of competitors. However, directexperimental demonstrations of character displacement due to competition are rare.Results: We demonstrate a causal role for competition in ecological character displacement.Using populations of the bacterium Escherichia coli that have adaptively diversified into ecotypesexploiting different carbon resources, we show that when interspecific competition is relaxed,phenotypic distributions converge. When we reinstate competition, phenotypic distributionsdiverge.Conclusion: This accordion-like dynamic provides direct experimental evidence that competitionfor resources can cause evolutionary shifts in resource-related characters.BackgroundWhen populations of different species occur in sympatry(together), they often have trait values that are moreextreme than the values occurring in allopatric (isolated)populations [1]. For traits associated with resource acqui-sition or metabolism, this phenomenon is called ecologi-cal character displacement, to distinguish it fromreproductive character displacement, which describesshifts in traits associated with reproduction. Ecologicalcharacter displacement is observed in Galapagos finchesally believed to be caused by resource competition. The-ory [9-12] predicts that character displacement will resultfrom competition selecting and maintaining extreme phe-notypes to minimize phenotypic overlap and thus mini-mize interspecific competition.Experiments also support the hypothesis that competitioncan select for divergence in resource-related traits. Schluter[13] measured selection in sticklebacks and demonstratedthat growth rates and survival were depressed in the pres-Published: 30 January 2008BMC Evolutionary Biology 2008, 8:34 doi:10.1186/1471-2148-8-34Received: 25 September 2007Accepted: 30 January 2008This article is available from: http://www.biomedcentral.com/1471-2148/8/34© 2008 Tyerman et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 9(page number not for citation purposes)[2-4], plethodontid salamanders [5], sticklebacks [2],Anolis lizards [6], and spadefoot toads [7,8], and is gener-ence of competitors, and that selection was frequencydependent [14]; and Bolnick [15] showed that competi-BMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34tion could generate disruptive selection regimes in naturalpopulations of sticklebacks. However, selection is notevolution, and few studies have shown that interspecificcompetition for resources leads to evolutionary shifts inphenotypic distributions of resource-related traits [16].Taper [16] demonstrated character shifts using bean wee-vils, however he failed to detect trade-offs associated withthe observed shifts, thus the divergence may have evolvedfor reasons other than interspecific resource competition.Microbes have been employed to great advantage in stud-ying the generation and sorting of adaptive variation [17-21]. Using microbes to test evolutionary hypotheses ispossible because microbes evolve quickly in response toenvironmental conditions set and controlled by theresearcher. Additionally, replicate populations can bestudied in order to determine the repeatability of evolu-tionary response, and microbes can be stored indefinitelyat -80° so that assays between ancestors and descendentscan be conducted [reviewed in [22]]. MacLean et al. [18]used biolog plates to characterize diversification of Pseu-domonas bacteria in response to resource competition.This study also demonstrated how diverse genotypes weremaintained by frequency dependent interactions likelyresulting from competition for resources. Similarly, Bar-rett et al. [20] showed that diversification of Pseudomonasgenerated imperfect generalists in response to competi-tion for substitutable resources. These studies nicely illus-trate how metabolic diversification occurs in the face ofresource competition. While they show divergence in phe-notype space, the phenotypes measured are not function-ally linked to competitive performance in theenvironment experienced during evolution. Therefore,the importance of the phenotype for competition remainsunclear. For example, it is not clear whether in experimen-tal populations seeded with only two phenotypes, compe-tition would lead to divergence, i.e. to an increase in thephenotypic distance between these two strains. Similarly,it is unclear what the effects of removing competitionfrom other strains would be on a single focal phenotype.Using diversified Escherichia coli B populations, we showin this paper that competition for resources can lead tophenotypic divergence of competing strains, i.e., to eco-logical character displacement, and that absence of com-petition can lead to phenotypic convergence. We evolvedE. coli for 1000 generations in liquid batch cultures withglucose and acetate as sole carbon resources. Ten replicatepopulations diversified into cultures consisting of twoecotypes that specialized on glucose or acetate [See Addi-tional file 1]. When E. coli grows in batch culture, glucoseis consumed first, followed by acetate [23]. This generatesa two-phase (i.e., diauxic) growth profile within a single24 h batch cycle (Figure 1a). Diauxic growth profilesthe first is exhausted. Resource exploitation can thus bedescribed as a metabolic reaction norm [24], and differentmetabolic reaction norms correspond to different 24 hgrowth profiles.Our evolved cultures had diversified into two ecotypes,identifiable by different 24 h growth profiles (Figure 1a).These growth profiles were assayed in the absence of com-petitors of the opposite ecotype and hence are not a plas-tic response to the presence of a competitor. Instead, theyreflect genetically distinct metabolic reaction norms,because offspring clones generate similar 24 h growth pro-files as parental clones from which they descend [24]. Wenamed the two distinct ecotypes Slow-Switchers (SS) andFast-Switchers (FS) after differences in their relativeswitching lags (lagace) between diauxic growth phases (Fig-ure 1b). We extracted lagace and nine additional quantifia-ble traits from diauxic growth curve profiles. Thesephenotypic traits carry the signatures of different strategiesfor metabolizing resources and have been shaped andmaintained by competition for resources [[24], See Addi-tional file 2].24 h growth curves reveal resource usage differences between ecotypesFigur  124 h growth curves reveal resource usage differences between ecotypes. (a) Examples of 24 h growth curves for the ancestor and derived ecotypes (Slow-switchers and Fast-switchers) from strain dst1018 after 1000 generations of evolution. (b) Histogram of lagace reveals two phenotypic clusters (Fast-switchers = black and Slow-switchers = white).0 5 10 15 20-2.4-2.0-1.6aTime (h)ln(Optical Density)FSAncestorSSblagaceFrequency0 20 40 60 80 100051015Page 2 of 9(page number not for citation purposes)reveal how bacteria consume one resource (e.g., glucose)and switch to a second resource (e.g., acetate) only whenBMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34Results & DiscussionWe envisage ecotypes occupying different regions of mul-tidimensional phenotype space, characterized by particu-lar values of resource-related traits, z. We can measure thedistance between ecotypes, ∆z, under transitions fromsympatry to allopatry (or vice versa), and ask whether thatdistance changes due to character displacement as theorypredicts [9,11,12,25]. Under competitive release, i.e.,moving from sympatry to allopatry, phenotypic distribu-tions should evolve towards intermediate values and thusappear closer in phenotype space, so that the distancemeasured in sympatry, ∆zSYM is larger than the distance inallopatry, ∆zALLO (i.e., ∆zSYM - ∆zALLO > 0). We tested thisprediction by evolving FS and SS ecotypes (from threepopulations) in isolation (i.e., under competitive release)for ~200 generations. Growth curve parameters wereextracted at T1 (generation 0), corresponding to sympatry,and at T30 (generation 200), corresponding to allopatry.We measured evolutionary response as the difference intrait value (T1-T30) for each ecotype from each popula-tion. We reduced the number of traits by conducting aprinciple components analysis (PCA, see Additional file3) and characterized SS and FS ecotypes in composite phe-notype space (Figure 2a). We calculated the distances ∆zA-LLO and ∆zSYM, and tested whether ∆zSYM - ∆zALLO > 0.Under competitive release, we found strong support forphenotypic convergence (Figure 2b) between ecotypesfrom all three populations (randomization test, dst1018:P = 1.0 × 10-6, dst1019: P = 1.0 × 10-6 and dst1020: P = 1.0× 10-6). Convergence occurred primarily along the firstprincipal component axis, with parallel shifts occurringon the remaining axes. Patterns of evolutionary responsediffered among populations. For example, convergence intwo populations (dst1019 and dst1020) consisted of bothecotypes moving towards one another in phenotypespace, but in population dst1018, convergence was due toa shift of both ecotypes in the same direction, but with SSchanging to a larger extent (Figure 2a). We suspect thatinitial differences in position in phenotype space(dst1018 vs. dst1019 and dst1020) accounted for differ-ences in evolutionary response of these ecotypes whenreleased from competition.Next, we investigated whether adding competition wouldinduce phenotypic divergence. For this we selected inter-mediate, convergent genotypes (SS' and FS'), which weisolated from T30 cultures (Figure 3, see Methods for fur-ther description). We competed SS' vs. FS' for 200 genera-tions, after which the frequency of FS'-derived genotypeswas <0.1% in 4 of the 10 competition replicates, suggest-ing that SS'-derived ecotypes were often able to competi-tively exclude FS'-derived ecotypes. Thus, we isolatedgenotypes derived from SS' (SSSYM) or FS' (FSSYM) from anCharacter displacement under competitive releaseFigure 2Character displacement under competitive release. (a) Symbols reflect mean ecotype evolutionary response from replicates (n = 20) evolved from each of three source populations (dst1018 = circles; dst1019 = triangles; dst1020 = squares), and arrows show evolutionary trajectories from sympatry to allopatry. Black symbols are FS ecotypes, white symbols are SS ecotypes. The ancestor (+) to the original evolution experiment is illustrated for comparison. Pheno-types are projected into two dimensions using the loadings from PC1 and PC2. (b) Mean distance in trait space, ∆z, between ecotypes in sympatry (black) and allopatry (white) during competitive release, for replicates (n = 20) from three aZPC1ZPC2lagace LongShortrmax−ace LowHighODmaxHighLowr max−gluHighLowANC1819201819201.∆zdst1018dst1019dst1020Page 3 of 9(page number not for citation purposes)earlier time point (generation 100, when FS was stillpresent in an appreciable frequency in all cultures), andpopulations. Error bars are 95% confidence intervals.BMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34calculated the mean growth-curve parameters for SSSYMand FSSYM. We projected these parameters using the samecomposite trait space characterized during competitiverelease (Figure 4a). Thus, we explicitly tested whethercompetition could induce evolutionary divergence bydirectly reversing the changes that occurred during com-petitive release. Indeed we found that competitioninduced divergence (t = 2.73, df = 9, p < 0.02, Figure 4b).Interestingly, divergence did not exactly retrace the evolu-tionary trajectory of convergence (Figure 4a vs. Figure 2a).Both convergence under competitive release, and subse-quent divergence due to competition, occurred along thefirst composite trait axis. However, under competitiverelease, both ecotypes contributed to convergence,whereas only the SS' phenotype contributed to diver-gence. Moreover, the magnitude of the evolutionaryresponse during the divergent phase was smaller than dur-ing the convergent phase (Figure 4b vs. Figure 2b). Thisdifference in magnitude may be because we assayed char-acter displacement after 200 generations in the first phaseand only 100 generations in the second phase, allowingCharacter displacement after competition was induced between intermediat  eco ypes (SS' vs. FS')Figur  4Character displacement after competition was induced between intermediate ecotypes (SS vs. FS). (a) Phenotypes are projected and scaled as in Figure 2, and gray symbols and arrows illustrate the evolutionary trajecto-ries that occurred during "competitive release" (first phase of study) in the relevant populations for comparison (see Figure 2a). Mean ecotype trajectories for SS' (circles) and FS' (trian-gles) from allopatry to sympatry. The black arrow shows the mean evolutionary trajectory of SS'-derived genotypes during competition, while the FS'-derived genotypes did not change substantially. (b) Distance in trait space, ∆z, between pairs of SS' and FS' competitors in sympatry (white) and allopatry aZPC1ZPC2lagace LongShortrmax−ace LowHighODmaxHighLowr max−gluHighLowANC18 SS19 FSSS'FS'b0.' vs. FS'∆zAfter 200 generations (T30) of isolated evolution, "conver-gent" cultur s (SSALLO and FSALLO) w re assayed for interme-diate genotyp sFigure 3After 200 generations (T30) of isolated evolution, "convergent" cultures (SSALLO and FSALLO) were assayed for intermediate genotypes. (a) SS' ecotype derived in an ara- culture (dst1018), with FS (dotted) and SS (dashed) ecotypes shown for comparison and (b) FS' ecotype derived from an ara+ culture (dst1019) with FS (dotted) and SS (dashed) ecotypes shown for comparison.0 5 10 15 20-2.4-2.0-1.6aTime (h)ln(Optical Density)SS'FSSS0 5 10 15 20-2.6-2.2-1.8-1.4bTime (h)ln(Optical Density)FS'FSSSPage 4 of 9(page number not for citation purposes)less time for evolution. However, the difference in magni-tude of evolutionary response may also have ecological(black).BMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34reasons. Schluter [2] argued that the speed of divergenceduring character displacement is greatest when pheno-typic distance (i.e., degree of similarity) between compet-ing species is intermediate. In particular, while verysimilar species experience intense competition, the speedof divergence is not expected to be high, because anincrease in phenotypic distance may not substantiallydecrease competitive intensity. Instead, divergencebecomes faster only after it has progressed considerably(See Figure 6.1 in ref [2]). Since the phenotypes we com-peted were rather similar (Figure 3), this effect may havedelayed the response in our divergence treatments.Finally, the evolutionary response under competition maybe different because divergence may have occurred in phe-notypic dimensions not captured by the composite traitspace defined by the PCA analysis of the competitiverelease experiment. We conducted an independent PCAanalysis on the data from only the divergence phase of ourexperiment, which yielded a different composite traitspace. In this new trait space divergence is also significant,but the response is of similar magnitude to the evolution-ary response initially identified (data not shown).ConclusionEcologists [2,9,11,12] continue to emphasize a causal rolefor competition in ecological character displacement.However, other factors, such as predation [21,26,27] canalso affect adaptive processes of diversification. Grant &Grant [3] have therefore recently called for a definitivedemonstration of competition's causal role in ecologicalcharacter displacement. Here, we answer this call usingexperimental tests in bacterial populations. Our evidencefor character convergence after competitive release is par-ticularly compelling, and our work supports the trust thatecological theory [2,11,28,29] has placed on competitionfor resources as an important driver of character diver-gence.Our study demonstrates that interspecific competition forresources can cause resource-related phenotypes to shift asexpected in response to competition. The initial adaptivediversification generating SS and FS ecotypes, followed byour manipulations of interspecific competition (byremoving and subsequently adding competitors) revealscompetition's role in driving accordion-like shifts on dis-tributions of resource-related phenotypes: divergence fol-lowed by convergence followed by divergence.Coexistence in the face of interspecific competition forshared resources may demand such an evolutionaryresponse, with the exclusion of the inferior ecotype as analternative outcome [30].MethodsDescription of evolved strainsTen replicate populations of E. coli B were alternately ini-tiated from two isogenic lines [31], which differed withrespect to a neutral marker. The isogenic lines differed intheir ability to utilize arabinose (ara+/-), which weexploited to discriminate between lineages in mixed cul-tures (see "Fitness assays" and "Competition induced"sections below). We followed the protocols of Lenski et al.[31] and others [24,32-36] with minor variations. Weused large, loosely-covered test tubes, filled with 10 mL ofDavis Minimal Salts media (DM) supplemented with 250µg/mL glucose and 575 µg/mL acetate as the sole carbonsources. These resources were selected because diversifica-tion in their presence has been shown previously[24,33,36]. Cultures were incubated at 37° and vigorouslyshaken (250 rpm) for 24 h. Each day (i.e., after 24 ± 1 hof growth), 100 µL of culture was transferred to 10 mL offresh media (~1/100 dilution) and thus the seasonal cyclewas reset. Each batch cycle yielded on average log2100 =6.7 generations.To test whether adaptation had occurred, we competedthree populations (dst1018, dst1019, dst1020) againstthe ancestor of opposite marker type, and calculated rela-tive fitness as done previously [31] (see below). Thesethree populations were selected from the initial ten popu-lations because there was a high correlation between col-ony morphology variation (large vs. small) and ecotype(SS vs. FS), which we exploited for purposes of identifica-tion in mixed culture assays. Fitness increased by ~14%[See Additional file 1] in all three populations. This sug-gests that adaptive evolution occurred over the course of1000 generations.By generation 1000, two discernable E. coli ecotypes, Fast-switching (FS) and Slow-switching (SS), were identified inall ten replicate populations [See Additional file 1], andthere was extensive variation in frequency of the two eco-types. We view the parallel emergence of diversity in eachpopulation as an indication that the divergence was adap-tive [10].To show that there is a functional (i.e., adaptive) explana-tion for the divergence in our E. coli populations, weassessed whether trade-offs in resource usage were detect-able between SS and FS strains. From previous work[24,33,36] and this study, it appears that SS was function-ally similar to the ancestor, while FS had diverged toexploit acetate earlier in the 24 h growth cycle (indicatedby reduced lagace, Figure 1b). Presumably this enhancedperformance on acetate is associated with reduced per-formance on glucose. Such a trade-off has previously beenPage 5 of 9(page number not for citation purposes)found in diversified strains that have evolved under simi-lar conditions [24,32]. To test for trade-offs in resourceBMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34use, we competed SS and FS in environments that wereskewed to having either more glucose or more acetate (seeFitness Assays below). A tradeoff would imply that in aglucose-enhanced/acetate-reduced environment, SS -hav-ing a metabolic profile geared towards efficient glucoseuse – would have higher fitness, while in an acetate-enhanced/glucose-reduced environment, FS – having ametabolic profile geared towards enhanced acetate use –would have higher fitness. Indeed, we found support forthe hypothesis that tradeoffs in resource use underlie themaintenance of diversity in metabolic profiles (t = 4.305,p < 0.0005 [See Additional file 2]). This tradeoff inresource use strongly supports the hypothesis thatresource competition was the selective cause for the diver-gence into SS and FS ecotypes.Asexual nature of our linesE. coli exchange DNA via conjugation, passing plasmidsbetween donor and recipient cells. However, E. coli B hasno plasmids and can thus be considered asexual [31]. Weensured that the ancestral lines (rel606 and rel607) andevolved lines used in this study had no plasmids with astandard mini preparation of genomic DNA isolated fromcells grown from each culture (Sigma GenElute PlasmidMiniprep Kit). No plasmids were detected in the ancestorsor evolved cells.Fitness assaysFitness of each evolved line was determined relative to theancestor using competition experiments as described in[31]. Briefly, evolved cultures (mixed sample of SS and FS)from the endpoint of our evolution experiment (genera-tion 1000) and cultures from the ancestors (both markertypes) were inoculated from frozen stock into evolution-ary media, and grown for 24 h. Evolved culture and ances-tor (of opposite marker type) were mixed in equalproportions (by volume) and inoculated into freshmedium (~1/100 dilution) in ten replicates; plated onTetrazolium agar with arabinose to determine densities atinoculation (T0), and then grown and transferred for twodays before being plated to yield T2 densities. Relative fit-ness was calculated as ln(EVT2/EVT0)/ln(ANCT2/ANCT0)(modified from [31]), where EV is the density of evolvedculture and ANC is the density of the ancestor (at times T0and T2). To determine fitness of SS and FS in skewedresource environments (i.e., 90% [glucose]-10% [acetate]or 10% [glucose]-90% [acetate]), we isolated 10 SS and 10FS genotypes from strain dst1018, inoculated them indi-vidually into fresh medium (50% [glucose] - 50% [ace-tate]) for 24 hours, and then arbitrarily selected pairs of SSand FS to mix in equal proportions (by volume). We inoc-ulated ten pairs into both extreme environments. Weplated T0 and T2 on tetrazolium agar plates (without ara-binose) and used colony morphology (large or small col-onies, see [24]) to aid us in determining the densities ofboth SS and FS ecotypes at each time point. We calculatedrelative fitness as above, substituting SS and FS for EV andANC.Growth parameter extractionGrowth curves were obtained by inoculating ~1.5 µL ofconditioned culture into 150 µL of fresh evolutionarymedium (see above) in individual wells of a 96-wellmicroplate. Microplate cultures were grown in a Biotek808UI Optical Density reader, under similar conditions tothe original evolutionary environment (37°, wellshaken). Measurements consisted of optical densities(OD, 600 nm) obtained every 10 min over the course of24 h. Data files were converted to a usable format usingMicrosoft Excel, and growth curve parameters wereextracted with a program written in object oriented C++.Table 1 summarizes the parameters extracted from growthcurves. These were modified from [24]. The StartTime wasTable 1: Description of parameters extracted from growth curves and summary data for SS and FS ecotypes (isolated from strain dst1018).Parameter Explanation Slow-switcher (SS)Mean (95% CI)Fast switcher (FS)Mean (95% CI)StartTime Time where optical density (OD) can be easily detected (OD = 0.08 at 600 nm) – not directly included in the analysis, but included in calculation of rmaxgluTP, timeToRmaxglu, SP, rmaxaceTP, and ODmaxTP.-- --rmaxglu Maximum growth rate during "glucose phase" of diauxie. 0.081 (0.076–0.086) 0.083 (0.077–0.089)rmaxgluTP rmaxglu time point – StartTime. 23.5 (22.9–24.2) 27.6 (26.7–28.5)SP Switching (Time) Point from glucose to acetate phase – StartTime. 32.2 (31.8–32.7) 31.9 (31.4–32.4)ODSP OD of switching point. 0.24 (0.23–0.25) 0.25 (0.24–0.26)Lagace Switching lag (time) from glucose to acetate growth 82.7 (76.3–89.0) 17.2 (10.4–24.0)rmaxace Maximum growth rate during "acetate phase" of diauxie. 0.0020 (0.0015–0.0024) 0.028 (0.023–0.034)rmaxaceTP rmaxace time point – StartTime 114.9 (108.7–121.2) 49.0 (42.2–55.9)ODmax Maximum OD. 0.24 (0.23–0.25) 0.36 (0.34–0.38)ODmaxTP ODmax timepoint -StartTime 49.2 (35.2–63.2) 80.5 (72.6–88.4)Page 6 of 9(page number not for citation purposes)ODfinal Yield or OD at the end of the 24 hour period. 0.19 (0.19–0.20) 0.32 (0.31–0.33)BMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34extracted but not used directly in the analysis; it was usedindirectly in the calculation of other variables (see Table1). StartTime was the time where the OD (600 nm) of thegrowing culture first reached 0.08. Slopes were extractedusing a moving window algorithm (i.e., linear regressionthrough nine successive time points), and were used forthe calculation of rmaxglu, switching OD, and rmaxace. ODmaxand ODfinal were the maximum and final optical densitiesduring the 24 h growth period. Means for each ecotype inTable 1 were calculated from twenty SS and twenty FSclones isolated from population dst1018.Character Displacement experiments1) Competitive ReleaseWe selected three of ten diversified populations for thisexperiment (dst1018, dst1019, dst1020). From the 1000-generation mark (maintained at -80°) we conditionedthese populations in fresh evolutionary media for 24 h,and plated on tetrazolium agar to isolate genotypes. Weselected 20 SS and FS genotypes (initially by colony mor-phology and confirmed by growth profile) from eachpopulation, and used these genotypes to initiate allopatriccultures (i.e., no interspecific competition). 1.5 µL of eachculture was inoculated into a single well containing 150µL evolutionary media of a 96-well microtitre plate (~1/100 dilution). Growth conditions and protocols mirroredthe evolutionary conditions, with the exception of differ-ences in volume between test tubes (10 mL in the originalevolution experiment) and microplate wells (150 µL inthis experiment). Separate microtitre plates were used forSS and FS cultures to prevent the possibility of cross-con-tamination between ecotypes. Although growth curveswere measured on ecotypes grown in isolation (i.e., nointerspecific competition), we assumed that initial growthparameter values (T1) had no mutations, and thusreflected the evolutionary signal of each ecotype undersympatry. This assumption is conservative, because we areactually measuring the parameters in isolation for thesympatric values to compare to later measures in allopa-try. All cultures were propagated in isolation (allopatry)for ~200 generations by transferring 1/100 of the cultureto fresh media in a new microplate every 24 hours for 30days. After 30 days of evolution, the values obtained fromgrowth curves were assumed to reflect the mean evolu-tionary response for each replicate to the treatment ofallopatry. A detailed analysis of individual genotypes(beyond this study) is ongoing. Growth parameters fromall derived cultures were log transformed.2) Competition inducedFrom T30 cultures generated in the first phase of this study(see above) we noted that all cultures were genetically het-erogeneous (as determined by variation in growth curve5 genotypes in SSALLO (i.e., cultures derived from SS). Inmany SSALLO cultures, we noted one particular recurringgenotype that had a decreased lagace, here labeled SS' (Fig-ure 3a). Similarly, in FSALLO cultures, we noted one partic-ular genotype with reduced maximum yields (ODMAX) ineach phase of diauxic growth (relative to the ancestral FSgenotype), here labeled FS' (Figure 3b). We consideredthese novel genotypes as intermediate between SS and FSecotypes, and relatively convergent towards the oppositeecotype, when compared with the ecotype from whichthey were descended (SS' derived from SS and FS' derivedfrom FS). A single SS-derived genotype (SS') was selectedfrom one of the twenty dst1018 replicates, and a single FS-derived genotype (FS') was isolated from one of thedst1019 replicates. We used single genotypes for eachnovel ecotype because we wanted to focus on the role ofcompetition (as opposed to extant genetic makeup of ini-tially variable populations) in ecological character dis-placement. Additionally, a fully replicated design with allpossible complimentary pairs of isolated novel genotypesin competition would be impractical.We initiated ten mixed cultures of SS' vs. FS' (1:1, by vol-ume) and inoculated these treatments into microplatewells (as above). We also inoculated pure SS' or FS' cultureto determine zALLO for each ecotype. We propagated themixed cultures for 30 days (200 generations) to determineif competition would cause the SS' and FS' to diverge inresource-related phenotype space. Because the frequencyof FS'-derived clones <0.1% by T30, we assayed our pop-ulations at T15 (See main text). We plated all replicatepopulations onto Tetrazolium agar (with arabinose) andidentified descendent clones by their ara +/- status. Four-teen clones for each of SS'-derived and FS'-derived sub-populations from each replicate mixture were isolatedand conditioned for 24 h before being assayed for growthcurve parameters. We then calculated the mean parametervalue from descendents from each ecotype from eachcompetition replicate for statistical analysis (see below).Statistical analysisIn both phases of the experiment, we tested the hypothe-sis that competition caused character displacement suchthat ∆zSYM - ∆zALLO > 0.1. Competitive releaseWe calculated the evolutionary response to competitiverelease (i.e., sympatry to allopatry) for each of ten traits bytaking the difference in log-transformed trait valuesbetween T1 and T30. We pooled the evolutionaryresponses for all 120 replicates (3 source populations × 2ecotypes × 20 replicates/population/ecotype). We con-ducted a PCA using the correlation matrix of the pooledPage 7 of 9(page number not for citation purposes)profiles from isolated clones). Generally, there were 2–3genotypes in FSALLO (i.e., cultures derived from FS) and 2–response data [37]. We used only the first four principlecomponents as they had eigenvalues > 1 [37], andBMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34accounted for > 81% of the variation in response to allo-patry. [See Additional file 3 and Table 2 for a summary ofthe PCA]. We used the loadings from these four compo-nents and the difference data to generate independent(orthogonal) composite trait values. Thus, our ecotypesare described as points in four-dimensional phenotypespace. From T1, we calculated:∆zSYM = zFS-SYM - zSS-SYMand from T30, we calculated:∆zALLO = zFS-ALLO - zSS-ALLOwhere z is a vector in four dimensional trait space reflect-ing mean population values for FS or SS ecotypes in sym-patry or allopatry. We analyzed the three sourcepopulations separately. We determined ∆zSYM and ∆zALLO(and 95% C.I.) by randomly sampling 20 distances 1000times from the fully permutated distance data set. Weused a randomization test procedure to determine theprobability of obtaining a test statistic (∆zSYM - ∆zALLO)that was ≥ observed data [38]. P values in the main textindicate the proportion of 100,000 analogous datasetscreated having (∆zSYM - ∆zALLO > observed data), after ran-domly reclassifying all distances into ∆zSYM or ∆zALLO data-sets.2. Competition inducedOur competition replicates comprised pairs (n = 10) of SS'and FS' derived genotypes. Thus, we used a paired t-test todetermine whether Ha: ∆zSYM - ∆zALLO > 0. This allowed usto quantify evolutionary response (i.e., divergence) ineach replicate (i.e., ∆zSYM) separately, so that divergenceacross replicates could arise even if ecotypes made differ-AbbreviationsSS: slow-switching ecotype; FS: fast-switching ecotype;lagace: acetate lag; ∆z: distance in phenotypic (trait) space.Authors' contributionsJT conceived the experiments, conducted the experiments,analyzed the data, and wrote the manuscript. MB and CCSconducted the experiments. MD conceived the experi-ments and wrote the manuscript.Additional materialAdditional file 1Figure S1. (a) Relative fitness of the ancestor (Generation 0) and three populations (Generation 1000) versus the ancestor of opposite marker type (ara+/-). The dashed horizontal line is equivalent fitness, error bars indicate 95% confidence intervals, and letters above error bars denote sig-nificantly different groups. (b) The proportion of SS (95% CI) in ten rep-licate populations evolved in glucose-acetate environment (populations in rank order). The dashed horizontal line represents the grand mean for all populations.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-34-S1.pdf]Additional file 2Figure S2. Competition experiments in skewed resource environments reveal that mean SS fitness is greater than mean FS fitness when [glucose] is enhanced (from 50% to 90%) and [acetate] reduced (from 50% to 10%) (left) and that mean SS fitness is lower than mean FS fitness when [glucose] is reduced and [acetate] enhanced (right). The horizontal line indicates equal fitness, and the error bars indicate 95% CI.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-34-S2.pdf]Table 2: Summary of PCA conducted on correlation matrix of the difference data during competitive release. See Table 1 for explanation of parameters.PC1 PC2 PC3 PC4Variation explained 31.3% 21.6% 16.2% 12.0%Eigenvalue 1.77 1.47 1.27 1.09Parameter: PC Loadingsrmaxglu 0.393 -0.411 - -rmaxgluTP -0.262 0.261 0.307 0.468SP - 0.285 0.195 0.634ODSP 0.197 -0.373 -0.382 0.448lagace 0.447 - 0.380 -rmaxace -0.420 -0.140 0.224 -0.107rmaxaceTP 0.384 0.153 0.398 0.248ODmax -0.114 0.609 - 0.149ODmaxTP 0.298 - 0.508 -ODfinal -0.330 -0.357 -0.329 0.269Page 8 of 9(page number not for citation purposes)ent contributions to divergence in different replicates.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Evolutionary Biology 2008, 8:34 http://www.biomedcentral.com/1471-2148/8/34AcknowledgementsFinancial support provided by NSERC to JGT and MD, and by the James S  McDonnell Foundation to MD. Lab assistance was provided by M. Ciou, M. Chan, G. Khaira, R. McBride, and R. Suprin. RE Lenski kindly provided our lab with E. coli B strains (rel606 and rel607) that served as the ancestors in our evolution experiment. M. Friesen helped develop protocols used in extracting parameters from growth curves. Discussions with D. Ally, A. Blachford, R. Blok, J. Fletcher, L. Harmon, M. Le Gac and D. Schluter aided us in preparing the study and manuscript.References1. Brown WL, Wilson EO: Character displacement.  Syst Zool 1956,5:49-64.2. Schluter D: The ecology of adaptive radiation.  Oxford: OxfordUniversity Press; 2000. 3. Grant P, Grant B: Evolution of character displacement in Dar-win's Finches.  Science 2006, 313:224-226.4. Schluter D, Price T, Grant P: Ecological character displacementin Darwin's Finches.  Science 1985, 227:1056-59.5. Adams D, Rohlf R: Ecological character displacement inPlethodon: biomechanical differences found from geometricmorphometric study.  PNAS 2000, 97:4106-11.6. Losos J: Integrative approaches to evolutionary ecology: Ano-lis lizards as model systems.  Ann Review of Ecol and Systematics1994, 25:467-93.7. Pfennig D, Murphy P: A test of alternative hypotheses for char-acter divergence between coexisting species.  Ecology 2003,84:1288-97.8. Pfennig D, Rice M, Martin R: Field and experimental evidence forcompetition's role in phenotypic divergence.  Evolution 2007,61:257-71.9. Doebeli M: An explicit genetic model for ecological characterdisplacement.  Ecology 1996, 77:510-520.10. Schluter D: Ecological character displacement in adaptiveradiation.  Am Nat 2000, 156:S4-S16.11. Slatkin M: Ecological character displacement.  Ecology 1980,61:163-77.12. Taper M, Chase T: Quantitative genetic models for the coevo-lution of character displacement.  Ecology 1985, 66:355-71.13. Schluter D: Experimental evidence that competition pro-motes divergence in adaptive radiation.  Science 1994,266:798-801.14. Schluter D: Frequency dependent natural selection duringcharacter displacement in sticklebacks.  Evolution 2003,57:1142-1150.15. Bolnick D: Can intraspecific competition drive disruptiveselection? An experimental test in natural populations ofsticklebacks.  Evolution 2004, 58:608-18.16. Taper M: Experimental character displacement in the adzukibean weevil, Callosobruchus chinensis.  In Bruchids and Legumes:economics, ecology and coevolution Edited by: Fujii K, Gatehouse A,Johnson C, Mitchel R, Yoshida T. Kluwer; 1990:289-301. 17. Rainey P, Travisano M: Adaptive radiation in a heterogeneousenvironment.  Nature 1998, 394:69-72.18. MacLean RC, Dickson A, Bell G: Resource competition andadaptive radiation in a microbial microcosm.  Ecol Letters 2005,20. Barrett RDH, MacLean RC, Bell G: Experimental Evolution ofPseudomonas fluorescens in Simple and Complex Environ-ments.  Am Nat 2005, 166:470-480.21. Meyer J, Kassen R: The effects of competition and predation ondiversification in a model adaptive radiation.  Nature 2007,446:432-5.22. Elena S, Lenski R: Evolutionary experiments with microorgan-isms: the dynamics and genetics of adaptation.  Nature reviews:Genetics 2003, 4:457-69.23. Mahadevan R, Edwards J, Doyle IFJ: Dynamic flux balance analysisof diauxic growth in Escherichia coli.  Biophysical J 2002,83:1331-40.24. Friesen ML, Saxer G, Travisano M, Doebeli M: Experimental evi-dence for sympatric ecological diversification due to fre-quency dependent competition in Escherichia coli.  Evolution2004, 58:245-60.25. Abrams PA: Character displacement and niche shift analyzedusing consumer-resource models of competition.  Theor PopBiol 1986, 29:107-60.26. Rundle H, Vamosi S, Schluter D: Experimental test of predation'seffect on divergent selection during character displacementin sticklebacks.  PNAS 2003, 100:14943-8.27. Nosil P, Crespi B: Experimental evidence that predation pro-motes divergence in adaptive radiation.  PNAS 2006,103:9090-5.28. Dieckmann U, Doebeli M: On the origin of species by sympatricspeciation.  Nature 1999, 400:354-7.29. Doebeli M, Dieckmann U: Speciation along environmental gra-dients.  Nature 2003, 421:259-64.30. Hardin G: The competitive exclusion principle.  Science 1960,131:1292-7.31. Lenski R, Rose M, Simpson S, Tadler S: Long-term experimentalevolution in Escherichia coli. I. Adaptation and divergenceduring 2,000 generations.  Am Nat 1991, 138:1315-41.32. Spencer CC, Saxer G, Travisano M, Doebeli M: Seasonal resourceoscillations maintain diversity in bacterial microcosms.  Evo-lutionary Ecology Research 2007, 9:775-87.33. Spencer CC, Bertrand M, Travisano M, Doebeli M: Adaptive diver-sification in genes regulating resource use in Escherchicia coli.PLoS Genetics.  PLoS Genetics 2007, 3:83-88.34. Spencer C, Tyerman J, Bertrand M, Doebeli M: Adaptationincreases the likelihood of diversification in an experimentalbacterial lineage.  PNAS  in press.35. Le Gac M, Brazas M, Bertrand M, Tyerman J, Spencer C, Hancock R,Doebeli M: Metabolic changes associated with adaptive diver-sification in Escherichia coli.  Genetics  in press.36. Tyerman JG, Havard N, Saxer G, Travisano M, Doebeli M: Unparal-lel diversification in bacterial microcosms.  Proc R Soc Lond B2005, 272:1393-8.37. Dillon W, Goldstein M: Multivariate analysis: methods andapplications.  New York, NY: John Wiley & Sons; 1984. 38. Manly BF: Randomization, bootstrap and Monte Carlo meth-ods in biology.  2nd edition. London: Chapman & Hall; 1997. Additional file 3Figure S3. Principle component analysis (PC1 vs. PC2) on differences between sympatric and allopatric trait values for Slow-switchers (white) and Fast-switchers (black) from replicates initiated from three populations (dst1018 = circles, dst1019 = triangles, dst1020 = squares).Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-34-S3.pdf]yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 9 of 9(page number not for citation purposes)8:38-46.19. Hall AR, Colegrave N: How does resource supply affect evolu-tionary diversification?  Proc R Soc Lond B 2007, 274:73-8.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items