Parallel evolution of alternate morphotypes of Chryseobacterium gleum during experimental evolution with Caenorhabditis elegans

Abstract Microbial evolution within polymicrobial communities is a complex process. Here, we report within-species diversification within multispecies microbial communities during experimental evolution with the nematode Caenorhabditis elegans. We describe morphological diversity in the target species Chryseobacterium gleum, which developed a novel colony morphotype in a small number of replicate communities. Alternate morphotypes coexisted with original morphotypes in communities, as well as in single-species experiments using evolved isolates. We found that the original and alternate morphotypes differed in motility and in spatial expansion in the presence of C. elegans. This study provides insight into the emergence and maintenance of intraspecies diversity in the context of microbial communities.


Introduction
How micr obes ada pt within networks of biotic inter actions is an ongoing question.During experimental e volution, micr obial communities in a given environment tend to follow similar trends in composition over time, where the same or similar taxa are present acr oss comm unities (Mor ella et al. 2019, Estr ela et al. 2021, Taylor et al. 2022 ).Across experiments , conditions , communities , and contexts, the comm unity-le v el tr ends observ ed during e volution ar e similar: a r a pid initial shift in comm unity composition is follo w ed b y mor e gr adual c hanges, gener ating a c har acteristic fastslow pattern in compositional div er gence fr om the initial community (Goldford et al. 2018, Estrela et al. 2021 ).This suggests that e volution of micr obial comm unities follows rules, suc h that similar selective pressures lead to community-level structural similarities (Meroz et al. 2021 ).
Ho w e v er, it does not follow that individual member taxa must e volv e similar tr aits .T he genetic and phenotypic traits of specific taxa within communities often div er ge and div ersify (Castledine et al. 2020 ).One example is within-host evolution in hostassociated microbial communities, where a given microbe of interest will evolve over time in situ.In some cases, selective sweeps result in different communities harboring a different dominant off-shoot of the same common ancestor (Didelot et al. 2016 ); the result is observed as diversity between hosts (Didelot et al. 2016 ) or across sites within the same host (Conwill et al. 2022 ).Alternatel y, v ariants fr om a common ancestor can coexist, pr oducing within-population diversity (Rainey et al. 2000, Chu et al. 2021, Turner et al. 2023 ).A m ultispecies comm unity context can pr omote within-species diversity (Chu et al. 2021 ), but as diversity can emerge from homogeneous populations (Cano et al. 2003, Hoffman et al. 2006, Blount et al. 2008, Pestrak et al. 2018, Chen et al. 2019, Obeng et al. 2023 ), a multispecies community is not r equir ed.The conditions for ada ptiv e div ersification ar e not yet clear.
Adaptation is driven by pressures in the abiotic and biotic envir onment, including inter actions with eukaryotes.Her e we use Caenorhabditis elegans , an established model system for microbial coevolution experiments (Meneely et al. 2019, Ford and King 2020, Gibson et al. 2020 ), as a host and predator during microbial community evolution with the worm.As in pr e vious work, we use combinatorial, defined starting communities with a tractable number of members (Taylor et al. 2022 ), allowing us to follow evolution within member taxa.
In this study, we r eport par allel e volution of within-species diversity in bacteria from C. elegans -associated communities.We focus on c har acterization of a morphological variant in the Gramnegative opportunistic pathogen Chryseobacterium gleum .The variant emerged at similar times in different communities and was maintained through the end of the experiment.The alternate mor ph differ ed in motility when compared to the original morphotype, as well as in population expansion on solid media in the presence of C. elegans .Our results provide insight into the selective pr essur es that can produce and maintain intraspecies variation during community-based microbial evolution.

Strains and culture conditions
Bacterial strains in Table 1 were obtained from the USDA Agricultur al Researc h Service (ARS) Cultur e Collection (NRRL) or fr om the American Type Culture Collection (ATCC).Strains sho w ed characteristic morphologies on salt-free nutrient agar (NA; 3 g yeast extract, 5 g peptone, and 15 g/l a gar), whic h was used for identification of colonies (Taylor et al. 2022 ).Combinatorial communities were set up with mixtures of seven strains as shown in Table 2 .N2 Bristol C. elegans were provided by the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructur e Pr ogr ams (P40 OD010440).Stoc ks of ancestr al N2 w orms w ere cultivated on 10 cm NGM agar plates at 25 • C with E. coli OP50 as a food source according to standard protocols (Stiernagle 2006 ).

Community-based experimental evolution
Comm unity-based e v olution w as carried out as in (Taylor et al. 2022 ), with minor modifications.To observe the outcomes of community-based host-associated evolution in a simple microbial community, we initiated experimental evolution of combinatorial minimal communities of bacteria (see Tables 1 and 2 ) with and without the nematode C. elegans as predator and host (see Fig. 1 for w orkflo w schematic).Starting cultures of bacteria were pr egr own individuall y in 1 ml LB cultur es for 48 h at 25 • C, tr ansferred to 1.5 ml Eppendorf tubes, and centrifuged 2 min at 9000 × g in a tabletop centrifuge (Eppendorf) to pellet.Supernatant was r emov ed and bacteria were resuspended in 1 ml S medium, then diluted to 10 8 CFU/ml in S medium and combined in equal ratios to form the communities in Table 2 .Worm + plates were initiated with sync hr onized L1 larv ae of WT Bristol N2 C. elegans ( ∼100 L1 larvae per plate).Worms were added to 6 cm NGM agar plates with 50 μl of the indicated community, with UV-killed lawns of E. coli OP50 as a starting food source.All communities were established in duplicate.
Every 7 da ys , plates were scored on a 0-4 scale for number of total worms, number of larvae, and plate cov er a ge by bacterial lawn.After scoring, plates were washed once with 1 ml M9 worm buffer + 0.1% Triton X-100 (M9TX01) to r emov e the bulk of the worms and the bacterial la wn.T he resulting suspension was centrifuged briefly in a small benchtop centrifuge to pellet w orms, then w ashed again with 1 ml M9TX01.The second wash w as drawn do wn to 100 μl and supernatant was discarded; worms wer e r esuspended by flic king, and 10 μl of the worm-and bacteriacontaining liquid was tr ansferr ed to a fr esh r oom-temper atur e plate of NGM + UV-killed OP50.Adult worms from the remaining sample were used for community analysis; target was 50 individual worms, but if total adult count was insufficient, all adults in the sample were counted and digested.Worms were washed, surface bleached, and mechanically disrupted as batches using a motorized pestle according to standard protocols for this lab (digests in 20 μl M9TX01 in 0.5 ml tubes, Kimble Kontes) (Taylor and Vega 2021 ).
Samples of live worms and bacterial contents fr om batc h digests wer e individuall y fr ozen on the day of passaging.Samples of the C. elegans population wer e pr eserv ed at −80 • C according to standard protocols (Stiernagle 2006 ).Half of the volume from worm batch digests was used for plating on NA to determine community composition; the remaining half was mixed 1:1 with 40% gl ycer ol and frozen at −80 • C. Single colony picks were taken from Comm unities ar e sho wn left to right in or der of community ID and replicate number (community A replicate 1, A1; A replicate 2, A2…; total n = 24).Bacterial strains are abbreviated: "AA", A. aurescens ; "BS", B. subtilis ; "CX", C. gleum (CG) or C. indologenes (CI); "MO", M. oxydans ; "OA", O. anthropi ; "RE", R. erythropolis ; "SX", S. spiritovorum (SS), or S. thalpopium (ST) (Tables 1 and 2 ).Chryseobacterium gleum (CG) is the Chryseobacterium isolate in communities A-B, E-F, and I-J (columns 1-4, 9-12, and 17-20).(E) and (F) Morphological diversity in worm + communities.(E) Emergence of alternate morphs.Passages where alternate morphotypes were observ ed ar e color ed according to the r elativ e abundance of the alternate mor ph, as a fr action of that species' total r epr esentation (fAlt).For passa ges where alternate morphs were noted but not counted, or where alternate morphs were retrieved from glycerol stocks but not from the original worm digest plates, r elativ e abundance is unknown and is assigned a v alue of −1 (blue) for visualization.Bars (blac k) indicate comm unities wher e the indicated isolate (CI or CG) was not part of the initial comm unity.Comm unity r eplicates ar e on the y -axis (in r e v erse order top to bottom, L2 → A1).(F) Colon y mor phologies in C. gleum.On the salt-free nutrient agar (NA) used for community plating, the original morph is brightl y color ed, glossy, domed, smooth, and slightly mucoid.The alternate morph is smaller and flattened, with less opacity and duller coloration.Morphologies are distinctive despite minor variation (right image, from community F1 passage 10).Two colonies with variations on the alternate morphology are circled in the center of the image ; one original morphology colony is circled on the right side of the plate.
worm digest plates (up to six isolates per species per replicate), grown in 100 μl LB in 96-well plates for 48 h at 25 • C, then mixed 1:1 with 40% gl ycer ol and cov er ed with aluminum sealing foil befor e fr eezing at −80 • C. When additional isolates wer e needed, the a ppr opriate cryopr eserv ed worm batc h digests wer e str eaked out fr om gl ycer ol on N A plates; single colonies w er e r estr eaked on fresh NA plates to confirm morphology, then single colony isolates from homogeneous plates were grown to late log phase in LB or NGM liquid medium and cryopr eserv ed at −80 • C.
For comm unity-based e v olution in the absence of w orms, a slightl y modified pr otocol w as used.Plates w er e initiall y established with ancestral communities as already described, but on 6 cm NGM plates without UV-killed OP50 as this initial worm food sour ce w as not needed.Every 7 da ys , plates w ere w ashed with M9TX01 to r emov e the entir e bacterial lawn, centrifuged for 2 min at 6000 × g to pellet, and resuspended in 1 ml S medium by pipetting.The resulting suspension was diluted in a 10-fold series for plating and CFU counting, and 10 μl of a 1:10 dilution of each suspension was aliquoted onto the center of a fresh 6 cm NGM plate to seed communities for the next passage.

Pairwise bacterial interactions on solid media
Isolates of CG from passage 6 and 10 worm + communities A1, A2, F2, and F2, as well as the ancestr al str ain ( n = 3 isolates per morph per condition), were grown up separately in 1 ml LB cultures for 24 h at 25 • C and diluted to a uniform cell density of 10 8 CFU/ml in S medium.Isolates of the original mor ph r epr esenting a given comm unity + passa ge wer e combined.For pairwise competitions, eac h original mor ph was mixed with individual alternate morph isolates.Mixed pairs were diluted 1:10 in S medium, then 10 μl of each mixture was transferred to the center of two 6 cm NGM plates (1.5% a gar, pour ed on the day of experiment), and plates wer e mov ed to 25 • C to grow.At indicated time points (2, 7, and 14 da ys), la wns were washed off plates (see Community-based experimental evolution ) and resuspended in 1 ml 1X PBS.These resuspensions were serially diluted 10-fold in 1X PBS (200 μl final volume per well) and plated in 100 μl aliquots on 10 cm NA plates for CFU counting and identification of original and alternate morphs.On day 7, fresh plates were inoculated with 10 μl of the first (1:10) serial dilution of the resuspended plate contents and incubated as described for 7 days for sampling on day 14.To quantify the effects of experimental factors (comm unity, gener ation, str ain ID) on fAlt morph, beta regressions with fAlt as the dependent variable were carried out in R using function betareg() from package betareg.
For pairwise interactions in the presence of worms, plates were established as alread y described, exce pt with the addition ∼10 gravid N2 hermaphrodites to each plate on day 0. Propagation and quantification of C. gleum populations was carried out as already described.
To compare bacterial populations within plates, a stainless steel circular clay hole cutter (diameter 0.8 cm) was used to remov e a gar plugs fr om the edge and center of each 6 cm plate (one plug at each location per plate sampled).Plugs were deposited dir ectl y into 1 ml 1X PBS in 1.5 ml microcentrifuge tubes and vortexed to resuspend bacteria.The resulting suspension was dilution plated onto NA as already described.

Chryseobacterium gleum mortality assays
As we observed host mortality in monocolonization experiments with these isolates, we sought to determine whether C. gleum was pathogenic to C. elegans , and whether e volv ed and ancestral isolates differed in pathogenicity.Isolates of C. gleum ancestral strain (ANC) as well as isolates obtained from worm + community plates in passages 6 and 10 w ere gro wn separately in 1 ml LB cultures for 48 h at 25 • C and diluted to a uniform cell density of 10 8 CFU/ml.The isolates used in these experiments were the same randomly selected isolates as in the pr e vious section.A volume of 50 μl of each isolate were plated onto separate 6 cm NGM plates in triplicate.Plates were grown up at 25 • C for 48 h to de v elop a lawn.Using a BioSorter (250FOCA, Union Biometrica), 50 sucrose washed adult N2 w orms w ere plated onto the resulting la wns .Plates were incubated at 25 • C, and worms wer e scor ed e v ery 24 h for 4 da ys .Worms that were not moving independently and did not respond to prodding with a worm pick were scored as dead.

Growth curves
Bacteria isolated from evolved communities and ancestral isolates for each strain were grown up separately in 1 ml LB cultures for 24 h at 25 • C, then diluted in liquid NGM to a uniform cell density of 10 8 CFU/ml.Each strain was then further diluted 1:100 to a final cell density of 10 6 CFU/ml in 150 μl liquid NGM in a clear pol ypr op ylene 96-w ell plate (three to four replicates per strain, eac h fr om a separ ate single colon y isolate).T he plate was co v er ed with BreatheEasy Sealing Membrane and incubated on a BioTek Syner gy HTX plate r eader using a 24-h kinetic pr otocol r eading OD600 e v ery 10 min at 25 • C. Gr owth curv e data wer e anal yzed in R using the function SummarizeGrowthByPlate () in growthcurver (Spr ouffske and Wa gner 2016 ) to infer carrying ca pacity of the logistic growth model from the OD600 over time data from each individual well; maximum growth rates were inferred for each well using man y.splines .fit()fr om pac ka ge growthrates (Hall et al. 2014 ).
P ar ameter v alues wer e calculated for all r eplicates for eac h individual strain, and summary statistics (mean, median, and variance) were calculated across experiments for each replicate.

Surface motility
Surface motility w as assay ed using a canonical drop culture assay on 6 cm NGM agar plates with 0.35%, 0.5%, or 1.5% agar.Plates, 6 cm, wer e filled with pr ecisel y 8 ml of agar on the day of experiment and allo w ed to solidify lid-off for 30 min in a laminar flow hood to help ensure uniform surface moisture across plates (Tremblay and Déziel 2008 ).Bacterial cultures w ere gro wn ov ernight fr om gl ycer ol stoc ks at 25 • C with shaking at 300 RPM in 100 μl of LB in individual wells of a 96-well plate , co v er ed with a BreatheEasy Sealing Membrane (USA Scientific).To acclimate cultures to the growth medium, cultures were diluted 1:1000 into 100 μl liquid NGM the morning of the experiment and allo w ed to r egr ow under the same growth conditions for 4 h.To inoculate motility plates, 1 μl drops of NGM culture were carefully placed in the center of each plate , a voiding splatter.Plates were incubated in stacks of not more than three plates at 25 • C in the dark, with a tray of water providing ambient humidity.1.5% and 0.5% plates gr e w for 7 da ys .0.35% plates gr e w for 2-3 days due to the swarms r eac hing the edge of the agar plate.Swarm diameter was measured with a ruler at 2-3 h intervals except overnight.
In assays to determine the effects of nutrient concentration on motility, the same protocol w as follo w ed, with the exception that 0.35% NGM agar was made with peptone at 0.2X (0.5 g/l), 1X (2.5 g/l), or 4X (10 g/l) of the standard formulation.All other components were k e pt constant according to the standard formulation for NGM agar.Acclimatization was done in 0.2X NGM.Where indicated, 10-fold dilutions of NGM cultures were made immediately prior to inoculation of motility plates.
Data were fitted to linear models.For plates established with a constant inoculum, the full model (Diameter ∼ log(Time) + Peptone + Morph + Community) was compared using likelihood ratio testing (LRT) against nested models dropping each independent pr edictor individuall y to determine whic h pr edictors wer e significant.For plates initiated with a range of initial inoculum densities, the full linear r egr ession model was instead Diameter ∼ log(Time) + Peptone + Morph + Dilution; only community-A isolates appeared in this data set.LME models were fitted to data for original and alternate morphs separately to determine whether dilution of the initial inoculum had a significant effect on slope, intercept, or both in the linear model.Linear r egr essions wer e carried out in R using the function lm() from package stats in base R , and LME fits were carried out using the function lmer() from package lme4 .

DN A extr action and genome sequencing
Cultures for genome extraction were grown individually overnight in 1 ml cultures of LB (25 • C, shaking at 300 RPM).Samples were centrifuged at 13 000 RPM for 2 min to pellet cells.Cells were washed with 1 ml 1X PBS thr ee times, r esuspended in 1 ml PBS, and separated into 5 PCR tubes (200 μl each).To improve lysis efficiency, a freeze-boil protocol was used ( −80 • C freezer for 10 min, then 100 • C thermocycler for 10 min; r epeat).DNA extr action was carried out using the Promega Wizard DNA Extraction Kit, with addition of 10 μl proteinase K (20 mg/ml, MolBio or Invitrogen) and 50 μl lysozyme (20 mg/ml, MolBio) to the lysis step (incubated 30 min at 37 Quality control and assembly were performed in Bactopia (Petit and Read 2020 ) using default par ameters, with r efer ence to a species-specific dataset (Ariba r efer ence datasets, RefSeq mash sets, and GenBank sourmash) including full genome assemblies of up to 1000 genomes identified as C. gleum .Variant calling was performed against the complete r efer ence genome for eac h str ain sequenced.
Mauve was used to identify SNPs from multiple genome alignment of the e volv ed isolates and the ancestral genome .T he alignment with the r efer ence genome was used for identification of SNP-containing loci when possible.When no homolog was available in the r efer ence genome, or when homologs were not annotated with gene names or specific functions, nearest homologs were identified via nucleotide megaBLAST (high homology) or BLASTn (somewhat dissimilar sequences) against C. gleum and C. cucumeris .

Results
As in pr e vious work (Taylor et al. 2022 ), 12 combinatorial communities were constructed from combinations of 11 member strains.All communities initially contained seven members, representing the same set of genera but with different re presentati ves at the species or strain level (Tables 1 and 2 ).This design allo w ed us to compare outcomes for each strain across at least four to six compositionally similar communities.
Experiments were initiated by inoculating communities in technical duplicate on NGM agar plates with wild type (WT) N2 C. elegans (Fig. 1 A).Comm unities wer e passa ged to fr esh plates e very 7 da ys , at which time a subsample of adult worms from each plate was disrupted to release intestinal contents (Taylor et al. 2022 ) to quantify community composition (see the section "Materials and methods").Single-species e volution, wher e eac h of the ele v en str ains was passa ged individuall y on plates with worms, was carried out in a pr e vious run of this experiment (Taylor et al. 2022 ) and was not repeated here.
As pr e viousl y observ ed (Taylor et al. 2022 ) experimental passaging of combinatorial bacterial communities with C. elegans resulted in coherent trajectories of community composition, which wer e lar gel y r obust to species-le v el v ariation in the starting comm unities (Fig. 1 B-D).These tr ajectories wer e c har acterized by r a pid loss of comm unity ric hness, r esulting in comm unities comprised almost entir el y of C. gleum/indologenes (CX), Microbacterium oxydans (MO), and Ochrobactrum anthropi (OA) within the first five passa ges .Composition conv er ged to one of two states; comm unities initiated with C. gleum came to be dominated by that species, wher eas comm unities initiated with C. indologenes wer e dominated by M. oxydans and O. anthropi (Fig. 1 D).While compositional conv er gence was also observed in communities passaged without C. elegans (Fig. 1 C), comm unity composition was v ery differ ent without worms, indicating that C. elegans played an important role in shaping these communities.
We observ ed div ersification within member taxa in wormassociated communities, in the form of morphological colony variants of Chrysebacterium and M. oxydans (Fig. 1 E and F; Figure S1 , Supporting Information ).No communities produced variants of both Chryseobacterium and M. oxydans .Not all communities produced variants; in most cases, only one of a given pair of replicate communities did so .W ith some exceptions, variants emerged nearly in synchrony across communities around passage 6, and mor photypes fr equentl y coexisted to the end of the experiment.We did not observe colony variants in communities passaged without worms (Fig. 1 ) or during passaging of single bacterial strains on plates with worms but without the community (Taylor et al. 2022 ).Although there was some variation within each morphotype (Fig. 1 F; Figure S1 , Supporting Information ), alternate mor phs fr om differ ent comm unities wer e visuall y similar, leading us to classify colonies into "original" (same morphology as ancestor) and "alternate" morphs.Timing of Chryseobacterium variant emer gence corr esponded r oughl y with time-series minim ums of compositional diversity ( Figure S2 , Supporting Information ).
To better understand the selective pressures underlying diversification in these populations, we chose to focus on c har acterization of within-species diversity in C. gleum (CG).Morphological diversity in Chryseobacterium emerged only twice, but from two different starting communities, allowing us to compare isolates acr oss these differ ent initial conditions.C. gleum alternate morphs wer e observ ed in comm unity A r eplicate 2 (A2) and comm unity F replicate 2 (F2); the remaining replicate of both communities A and F (A1, F1) gener ated onl y colonies of the original morphotype.
Taxonomy of e volv ed isolates was confirmed via 16S sequencing.Isolates from both morphotypes retained full pathogenicity to w ar d the worm ( Figure S3 , Supporting Information ), consistent with our prior experience with these Chryseobacteria (Taylor et al. 2022 ) and with other work in this clade (P a ge et al. 2019 ).

Logistic growth
We first sought to explain intraspecies diversity in C. gleum from an ecological standpoint, to determine (1) how intraspecies diversity is maintained in these populations and (2) whether the emergence of diversity in one replicate of eac h div ersified comm unity has an ecological explanation, or simpl y r eflects the low frequency at which new variants arise.We began with the simplest hypothesis, wher e intr aspecies div ersity is maintained or lost due to interactions with conspecifics.
Bacterial gr owth par ameters (maxim um gr owth r ate and saturation density) are often used as markers of relative fitness (Concepción-Ace v edo et al. 2015, Momeni et al. 2017, Saav edr a et al. 2017 ).We, ther efor e measur ed gr owth par ameters for original and alternate morphotype isolates of C. gleum, from passages 6 (earl y emer gence) and 10 (end of experiment) and communities A and F. Liquid NGM medium and growth at 25 • C were used to approximate the conditions in the original experiment.Individual isolates were grown in separate wells for 24 h in a plate reader to record absorbance (OD 600 ) o ver time , and cultures were dilution plated at the end of the run to determine colony forming units (CFU/ml).
Ov er all, original and alternate morphotypes sho w ed minimal differ ences in gr owth ( Figur e S4 , Supporting Information ).Maxim um gr owth r ate w as lo w er in alternate morphs than in the ancestr al str ain; onl y one set of original morphs (from community F2 pass 10, F2o10) sho w ed this decrease relative to ancestor.CFU measur ements fr om stationary phase cultures indicated incr eased satur ation density in alternate morphs as compared with the ancestor.Ho w e v er, man y original mor phs (A1o10, A2o10, F1o6, F2o6, and F2o10) also sho w ed higher saturation density than the ancestor, which was not always accompanied by a reduction in maxim um gr owth r ate.
These results did not explain the observed morphotype diversity.While a tradeoff between decreased exponential growth r ate and incr eased satur ation density in alternate mor phs might have allo w ed coexistence of morphotypes, the improved saturation density and mor e r a pid gr owth of e volv ed original mor phs should have pushed the balance to w ar d competitive exclusion.Further, these data did not explain the absence of alternate morphotypes in one replicate (A1, F1) of each community.

Pairwise competition on solid media
Growth of isolates in liquid media may not be a good representation of performance in mixed cultures in a structured environment.We, ther efor e next sought to assess intraspecies competitive ability of alternate and original morphs directly, using pair-wise competitions on solid media.In these experiments, cultures of alternate and original mor phs wer e gr own separ atel y in liquid media, diluted to fixed density, then mixed 1:1 and inoculated into the center of NGM agar plates as was done during experimental ev olution.No w orms w ere added during these experiments, to focus only on intraspecies interactions between morphotypes.At indicated time points, plates w ere w ashed with buffer to resuspend the entire lawn, and the resulting suspension was plated to determine counts of alternate and original morphologies .P opulations wer e passa ged to fr esh plates after 7 days as befor e.
Coexistence was maintained in all pairs of alternate and original morphs (Fig. 2 A).Populations were similar at the end of the first (day 7) and second (day 14) passages for most combinations ( Figure S5 , Supporting Information ) and extending a subset of conditions for one additional passage had no further effect on ratios or abundances (not shown), suggesting that a local ecological equilibrium had been r eac hed for most pairs.
The final fraction of alternate morph (fAlt) varied across pairs (10%-99% at day 14), but some trends were evident (Fig. 2 A; Table S1 , Supporting Information ).The ancestral original-morph str ain (WT) gener all y sho w ed higher r elativ e abundance (lo w er fAlt) than e volv ed original morphs (beta regression with fAlt as dependent variable, P ≈ 5.8e −6 ).Alternate morphs from the final passage tended to show higher r elativ e abundance than passage-6 isolates ( P ≈ 1.3e −5 ).Community of origin (A1/A2/F1/F2) did not predict outcomes, and alternate morphs did not consistently perform differ entl y in competition with original mor phs fr om the same community and passage, as would be expected with Red Queen-type arms-race dynamics (all indicated terms not significant after correction) (Morran et al. 2011 ).Qualitatively, across strains and pairs, we observed that alternate morphs tended to incr ease fr om r ar e (fAlt < 10%) but r ar el y decr eased fr om common ( > 80%).This indicated that intr aspecies competitiv e exclusion does not explain the absence of alternate morphs in communities A1 and F1.
Much of the variation in outcomes remained unexplained.A model using strain IDs as independent variables for fAlt (beta regression pseudo-R 2 = 0.43 with 13 terms) performed somewhat better than the model using community + passage (pseudo-R 2 = 0.32 with eight terms), while an extended model using strains and combinations (37 terms) accounted for m uc h of the variation in fAlt (pseudo-R 2 = 0.91) ( Table S1 , Supporting Information ), suggesting that pair-specific interactions were important for describing this v ariation.We, ther efor e sought to determine whether differ ences in gr owth par ameters within competing pairs could explain this variability.To do so, we compared differences in gr owth r ate (r Alt -r Ori ) and differences in saturation log 10 (CFU/ml) in NGM medium (K Alt -K Ori ) against fAlt in in vitro pairwise competitions for all pairs of isolates.We observed no correlation betw een gro wth parameters and outcomes in pairs ( Figure S6 , Supporting Information ).

Pairwise competition on worm plates
These in vitro r esults differ ed fr om those observ ed during experimental e volution.Alternate mor phs wer e pr esent at higher r elative abundance during in vitro pairwise competition (Fig. 2 ) than in the original samples (Fig. 1 ).In many pairs, the alternate morph was > 50% of the population (maximum alternate morph ∼99%), as compared with a maximum of ∼30% alternate morph during experimental evolution.
As the presence of the worm had a marked effect on community structure (Fig. 1 ), we hypothesized that the worm might also alter interactions between morphs .We , therefore repeated pair-wise competitions with C. elegans on plates to determine whether the worm could alter the r elativ e abundance of alternate morphs.Pairs of alternate and original morphs were mixed to create a range of initial frequencies of the alternate morph, then spotted onto standard 6 cm NGM + OP50 plates as before, but with the addition of ∼10 gravid N2 hermaphrodites to each plate .Here , we used onl y CG fr om the final passa ge, r epr esenting pr esumabl y the most adapted isolates from these experiments.
The presence of worms altered C. gleum populations.In the pr esence of worms, fr action alternate mor ph (fAlt) was gener all y lo w er than in the absence of worms, with a stable point around 10% alternate morph across all pairs (Fig. 2 B).For community A2 alternate morphs, points at higher initial fAlt fell above the 1:1 line, suggesting a second equilibrium at a fAlt above the range of these data; this is consistent with in vitro results where these isolates fr equentl y comprised 90% + of the population (Fig. 2 A).For community F2 alternate morphs, the data suggested a second unstable equilibrium around fAlt = 0.4.
Further, the presence of worms altered the spread of C. gleum ov er the a gar surface.Unlike worm-fr ee plates wher e CG gr e w onl y ar ound the ar ea of the initial spot, CG lawns e v entuall y covered the plate surface when w orms w ere present.Day-3 population diameter differed across original morphs (Kruskal-Wallace P = 2.8e −4 , df = 4) and was greatest on plates with the ancestral isolate and lo w est on plates wher e alternate mor phs wer e competed a gainst nonmatc hed e volv ed original mor phs (A1o10 vs. A2a10, F1o10 vs. F2a10; Fig. 2 C).
We hypothesized that alternate and original morphs might play differ ent r oles in worm-dependent population expansion on these plates.If so, the pr e v alence of alternate mor phs at the edge of the plate (farthest point of population expansion) should be different than at the site of the original inoculum.To determine the role of alternate morphs in expansion of C. gleum populations on worm plates, we measured population composition at the plate center (near the original inoculum) and the plate edge after 7 da ys .In most pairs , alternate mor phs wer e ov er-r epr esented at the plate edge r elativ e to the center, and the magnitude of the ov er-r epr esentation was gr eatest in matc hed pairs of original and alternate morphs (A2o10 vs. A2a10, F2o10 vs. F2a10) (Fig. 2 D).

Motility
As motility of community members was altered in an earlier, similar experiment (Taylor et al. 2022 ), we next sought to determine whether morphotypes could be characterized by differences in this tr ait.C. gleum lac ks fla gellar motility (Steinber g and Burd 2015), but like many other Flavobacteriales , shows gliding motility on the surface of soft (0.35%) agar plates (McBride and Zhu 2013, Shriv astav a et al. 2013, McBride 2014, Sato et al . 2021 ).Surface motility assays were therefore conducted on soft (0.35%), swarming (0.5%), and solid (1.5%) NGM agar to determine whether the alternate morphology was associated with changes in motility (Fig. 3 ).All isolates were motile on soft agar, and alternate mor phs wer e consistentl y less motile than original morphs (Fig. 3 A; Wilcoxon tests of original vs. alternate morphs: 0.35% agar, P = 1.47e −8 ; 0.5% agar, P = 5.65e −7 ; and 1.5% agar, P = 5.84e −6 ).Early (pass 6) and late (pass 10) isolates of eac h mor photype did not differ in motility (all comparisons not significant after correction) and we did not observe any trend in motility over passages (Fig. 3 B).These data do not indicate directional selection on motility ov er all.Further, these r esults cannot explain enric hment of alternate morphs at the population frontier on standard (1.5%  B) and (C) Surface motility of original mor phs fr om passa ges 1 to 10 of communities (B)-A and (C)-F, on soft 0.35% NGM agar ( n = 4 isolates per community + passage).Isolates from passages 6 and 10 are the same as in panel (A); missing data reflect passages and communities for which colonies could not be retrieved from glycerol stocks .T he horizontal line indicates the median diameter of ancestral WT isolates ( n = 4) from the same day.Tests are Kruskal-Wallace of motility diameter at ∼1.8 da ys vs .passage .Note that community-A and community-F isolates were assayed on separate days with separate batches of freshly prepared plates, and the raw diameters cannot be compared across days due to inherent variability in these assa ys .agar) plates with C. elegans , as alternate morph colonies spread less e v en on this firm substr ate.

Sequencing
We next sought to identify genetic differences that would provide insight into the phenotypic differences between alternate and original mor phs.Ele v en str ains (Ancestor, A1o10, A2o6, A2a6, A2o10, A2a10, F1o10, F2o6, F2a6, F2o10, and F2a10) were sequenced to identify genomic changes during selection.Ov er all, we observ ed v ery little genetic div er gence between isolates ( Table S2 , Supporting Information ).All of the putative "SNPs" identified in our initial anal ysis ma pped to low-complexity regions and/or the ends of contigs ( Table S2 , Supporting Information ) and may, therefor e, r epr esent sequencing or alignment errors rather than real sequence differences.No putative mutations were unique to and conserv ed ov er alternate mor phs.
Chryseobacterium and other Flavobacteriales show gliding motility (McBride andZhu 2013 , Gavriilidou et al. 2020 ), which depends on a type 9 secretion system (T9SS) (Trivedi et al. 2022 ).All sequences sho w ed complete sets of homologs to known components of gliding motility, specifically SprA, PorV, PorP, PorT, and all gld genes.As is typical for this clade, gldN , gldM , gldL , and gldK are adjacent to one another and appear to be a single operon; the same is true for porV , porU , and gldJ , and for gldB and gldC .We did not find homologs of SprB or other secreted adhesins in these genomes; BLAST search of the F. johnsoniae operon sprCDB (Gen-Bank EF111026.1)against Chryseobacterium generated no hits, indicating that the genus in general does not have the SprB adhesin.These C. gleum genomes do contain an annotated LolA outer mem-br ane lipopr otein, whic h can be used for spreading on soft agar in the absence of adhesin activity (Sato et al . 2021 ); as with other motility genes, no SNPs wer e observ ed in the lolA sequence or adjacent regions.
When compared with the ancestral genome, no SNPs were observed in known motility or secretion genes or in regulatory regions associated with any of these genes in any evolved isolates.The lack of genetic differences between original and alternate morphologies in motility encoding regions indicated that differences between morphotypes did not arise from simple loss or modification of these components .T his aligns with pr e vious findings in other Fla vobacteriales , wher e alter ed mor phologies associated with reduced motility were not linked to mutations in motility genes (Penttinen et al. 2018 ).

Phenotype × environment for alternate and original morphs
We next sought to c har acterize differ ences between isolate phenotypes by changing the environment.Concentration of a rich nutrient source (e.g.peptone) has been shown to alter motility in Chryseobacterium and related Flavobacteriales (Gavriilidou et al. 2020, Sato et al . 2021, Khare et al. 2022 ).This suggested that differences in the relationship between motility and nutrient levels in original vs. alternate morphs could be used to gain further insight into the differences between morphotypes.
We, ther efor e assessed motility on 0.35% NGM agar where nutrient concentration was altered by changing the concentration of peptone (0.2X, 1X, or 4X of the standard formulation).NGM contains peptone at a standard concentration of 2.5 g/l, compared with Difco LB medium at 10 g/l tryptone; 4X NGM is, ther efor e equivalent to 1X Difco LB in pe ptone-equi v alent concentr ation.The range of NGM peptone concentrations used here recapitulates the range of LB concentrations over which earlier studies sho w ed changes in motility in Chryseobacteria .
Motility in these assays had a weak but significant dependence on peptone concentration, and a strong and significant dependence on morphology of the isolate (Fig. 4 A; Table S3 , Supporting Information ).In all r egr essions, comm unity (A2 vs. F2) was a small but significant factor; ho w e v er, as isolates fr om differ ent comm unities were assayed on different da ys , this ma y be a batch effect.Consistent with expectations, original mor phs wer e mor e motile than alternate mor phs acr oss all conditions ( Table S3 , Supporting Information ).An intermediate le v el of peptone (1X) produced the greatest motility in original morphs (linear regression, diameter ∼ log(hours) + Peptone + Community, β Peptone1X, ORI =+ 0.41 P < 2e − 16 ; β Peptone4X, ORI not significant).Whereas original morphs sho w ed expansion rates which declined over time (linear regr ession, [expansion r ate] ∼ log(hours) + Peptone + Community, β ln(time), ORI = −0.04,P < 2e − 16 ), alternate morphs expanded at effectiv el y fixed rates ( β ln(time), ALT and β Peptone, ALT not significantly differ ent fr om zer o, all P > .05)(Fig. 4 B; Table S3 , Supporting Information ).Additionall y, mor phology of the colon y fr ontier, and particularl y the br anc hing pattern, was alter ed by peptone concentr ation ( Figur e S7 , Supporting Information ) and was specific to morphotype.
Flav obacteriaceae are kno wn to sho w density-dependent and density-independent phases of gliding motility (Álv ar ez et al. 2006, Penttinen et al. 2018, Gavriilidou et al. 2020, Sato et al . 2021, Khare et al. 2022 ).We hypothesized that the time-dependence of expansion rate in original morphs, and the lack thereof in alternate morphs, was related to density-dependence of motility .Specifically , we hypothesized that expansion rates in original morphs should show dependence on initial inoculum density, whereas this interaction should not occur for alternate morphs.To test this, motility plates were established with inoculum densities across four orders of magnitude (Fig. 4 C and D).The relationship between surface motility (diameter) and inoculum density differed between alternate and original morphs (Fig. 4 C, likelihood ratio test (LRT) P < 2.2e −16 for nested linear r egr essions with vs without morph as a predictor).For alternate morphs, inoculum density altered diameter early in expansion [intercept in linear mixed effects model (LME), CI for all terms excluded 0] but not expansion rate (LRT P = .92).Inoculum density had a larger effect on early time diameter for original than alternate morphs (LME coefficients for alternate vs original morphs: dilution −3, intercept −2.2 vs. −3.6;dilution −2, −2.1 vs. −3.1;and dilution −1, −1.7 vs. −2.6).Ov er all, expansion r ate of original mor phs was not significantly affected by inoculum density (LRT P = .09),but the lowest dilution ( −3) is significantly different from the rest of the data ( P = .012),indicating a threshold for density dependence .T his is true particularly at early times, when higher-density inocula show a phase of r a pid expansion (Fig. 4 D).
Peptone concentration significantly affected both expansion rates on soft agar (LRT P = 1.1e −15 ) and overall biomass pro-ductivity ( Figure S8A and D , Supporting Information ) in both morphs.Ho w ever, the observed motility differences between morphotypes were not explained by differences in gr owth par ameters ( Figure S8 , Supporting Information ).Even in nutrient conditions where original and alternate morphs sho w ed indistinguishable growth rates and/or carrying capacities, the original morph was substantially more motile .T his indicated that morphotype differences in the phenotype x envir onment inter action for motility could not be explained simply by differences in cell density or rate of production of new cells.

Discussion
In this study, we observ ed comm unity-based e volution in microbial consortia on plates with C. elegans .The presence of the worm alter ed comm unity composition and the e volution of member taxa, with intraspecies morphological diversity emerging in worm-associated comm unities.Her e, we focused on c har acterization of diversity in C. gleum , which developed an alternate morphotype in one replicate of each of two communities (A and F).Alternate morphs from different communities were phenotypically almost indistinguishable, indicating conv er gent e volution of intr aspecies div ersity in C. gleum .Emergence of the alternate morphotype was rare, but intraspecies diversity was robustly maintained.Spatial population expansion on plates, a ppar entl y with C. elegans as a v ector, favor ed maintenance of the alternate morph.
The selective processes that resulted in diversification are not fully clear.It has been suggested that interspecific niche-based competition can promote diversification of focal species in microbial communities (Chu et al. 2021 ).Consistent with this idea, we did not observe alternate morphotypes during evolution of bacterial monocultures with worms during a similar experiment (Taylor et al. 2022 ).Alternate morphs of C. gleum emerged at a low point in community diversity ( Figure S2 , Supporting Information ), consistent with observations that low-diversity communities are mor e likel y to gener ate ne w div ersity due to nic he av ailability (Madi et al. 2020(Madi et al. , 2022 ) ).
In a structured environment, colonization of empty space is a component of niche competition; spatial expansion may easily be important for both inter-and intraspecies niche-based inter actions.Understanding inter actions among member species is, ther efor e of interest for understanding community-level outcomes.Ho w e v er, C. gleum mor phs coexisted in single-species populations, indicating that the multispecies community was not necessary for maintenance of diversity.Spatial structure on plates may be sufficient to explain coexistence (Lo w ery and Ursell 2019 , Gude et al. 2020, Wu et al. 2022 ).It is also plausible that our alternate morph isolates are not first-step mutants, and that propagation of early rare variants did require the community context.Additionall y, ther e may be frequency-or density-dependent interactions outside the range of our assays that might explain the importance of the community.
From first principles, it is likely that selection in C. gleum occurr ed mostl y in the envir onment r ather than in the worm host.C. gleum is a pathogen, causing substantial mortality in adult N2 worms within 1 day of exposur e ( Figur e S3 , Supporting Information ); in fact, colonization by these bacteria is difficult to measure due to mortality.Community plates with C. gleum supported small, struggling populations of worms and heavy la wns , suggesting pathogen-related mortality.Despite high relative abundance of C. gleum , total worm-associated populations of this bacteria wer e small, gener all y < 10 3 bacteria per adult worm.
Further, nearly all worms on plates toward the end of a passage wer e juv eniles, whic h ar e expected to support smaller intestinal bacterial loads than adults (P ortal-Celha y et al. 2012 ).This indicates that C. gleum populations in the collective intestines of worms on these plates were small and short-term and suggests that C. gleum was maintained lar gel y in the environment rather than in the host.
In this context, the worm can act as both a predator and a vector.The worm is well known to spread bacteria by carriage in the gut and on the external cuticle (Kenney et al. 2005, Thutupalli et al. 2017, Bermudez et al. 2018 ).Alternate morphs had an advantage in expansion on hard agar plates only when C. elegans were pr esent, suggesting that v ectoring of bacteria by the worm is rele v ant.It is plausible that alternate morphs of C. gleum were selected for during spatial expansion; as C. gleum gliding motility is ineffective on hard agar, this expansion required use of a nematode vehicle .T his is the onl y case wher e coe volution of alternate and original morphs (same community and replicate) affected the outcomes of intraspecies competition.
Ho w e v er, it is not clear what trait(s) of the alternate morph ar e r ele v ant.Gliding motility is essentiall y ineffectiv e on hard (1.5%) agar, and the alternate morph still shows smaller colonies than original morphs on hard agar.It is unlikely that the motility phenotype is dir ectl y r ele v ant for spatial expansion her e.Rather, these results may indicate selection on a trait that is coregulated with motility.
One possibility is biofilm formation.Mutants in lineages of motile biofilm-forming bacteria often show a tradeoff where increased motility is associated with decreased biofilm formation and vice v ersa (v an Ditmarsc h et al. 2013(v an Ditmarsc h et al. , P alma et al. 2022 ).The T9SS is important for both motility and biofilm formation in other Bacteroidota, and these phenotypes are linked (Penttinen et al. 2018, Ec kr oat et al. 2021 ).Changes in a ggr egation or biofilm formation might affect the ability of bacterial cells to adhere to the worm cuticle, to be ingested by the worm, and/or to survive passage through the intestine.Measuring biofilm formation and vectoring by the worm, and determining the bacterial traits that affect vectoring, remain a challenge for future work.
Although the motility phenotypes do not explain expansion of C. gleum on hard agar plates, and as suc h ar e used her e onl y as an experimentally accessible indicator of differences between mor photypes, our observ ations ar e inter esting as a source of insights into T9SS-dependent motility.Our specific results were inconsistent with other work where Chryseobacterium motility increased as the concentration of rich medium (LB or Shieh) was decreased (Penttinen et al. 2018, Khare et al. 2022 ).Further, our results are in contrast with other work (Khare et al. 2022 ) where high tryptone concentr ations pr oduced a smooth motility fr ontier in Chryseobacterium .The authors suggested that this was a result of higher cell densities at high nutrient le v els.We observ ed a nearl y smooth frontier at low peptone for all original mor phs, wher eas high peptone produces dense growth and a heavily branched frontier ( Figure S8 , Supporting Information ), suggesting that the structure of the expansion frontier is due to collective behavior rather than density alone.
T hese observations , together with our observation that an intermediate concentration of peptone maximized expansion of original morphs, suggest that the properties of the spreading frontier are controlled by a combination of individual and collective motility (P atr a et al. 2016 ) and that use of these modes differs between morphs and across en vironments .F rom theory, w e expect that br anc hing will be promoted when directional movement of individual cells (e.g.taxis into unoccupied space) is r elativ el y low (Giverso et al. 2015, Bisht et al. 2017 ).Further, expansion rate should be maximized at some value of individual motility (Bisht et al. 2017 ), which could explain why original morphs sho w ed higher motility at 1X peptone than at 0.2X or 4X.It is plausible that differ ences in dir ectional mov ement and gr owth acr oss str ains and envir onments ar e sufficient to explain these r esults; this r emains to be determined.Aggregate formation may be important as well, particularly for understanding changes in branching structure (Li et al. 2021 ).
While evolution often produces compositionally similar comm unities, individual micr obial linea ges exhibit distinct e volutionary trajectories within these comm unities.Ada ptation in real (or realistic) en vironments , with spatial structure and interspecies/interkingdom interactions, tends to select for variants that interact differently with physical space and with other biological agents (Hansen et al. 2007, Santos-Lopez et al. 2019 ).Although growth rates and carrying capacities are historically used as part of definitions of "fitness" in the labor atory, r eal world fitness of microbes is often weakly associated with these parameters (Concepción-Ace v edo et al. 2015, Momeni et al. 2017, Bansept et al. 2021 ).Instead, fitness in a spatially structured environment with biotic interactions is more often a function of physical and social inter actions, wher e spatial scales and inter action structur es ar e critical aspects of selection (Hansen et al. 2007, France et al. 2018, Liu et al. 2019, Gorter et al. 2020, Conwill et al. 2022, Hoces et al. 2022 ).Inv estigations of r ealisticall y structur ed populations, wher e temporal and spatial dynamics are considered explicitly, are important for understanding microbial evolution and intraspecies diversity.

AI usage statement
During the pr epar ation of this work, the authors used Researc hRabbit.ai to acceler ate liter atur e searc h and ChatGPT to r evise portions of the text for clarity.After using these tools/services, the authors r e vie wed and edited the content as needed and take full responsibility for the content of the publication.

Figure 1 .
Figure 1.Comm unity-based ada ptation pr oduces conv er gent comm unity composition and within-species div ersity.(A) Experimental w orkflo w for comm unity-based e volution.The experimental set up consisted of 12 comm unities with se v en members eac h, similar at a genus le v el but v ariable on a species/isolate le v el.Eac h comm unity w as initiated on tw o r eplicate plates (comm unity A-L, r eplicates 1 and 2), to whic h L1 sta ge N2 C. elegans wer e added.These communities + worm populations were passaged every seven days to fresh NGM agar plates.At the end of each passage, adult worms wer e batc hed, physicall y disrupted to r elease gut contents, and plated in serial dilution to assess bacterial comm unity composition.(B)-(D) Community composition follows coherent trajectories over time, with compositional differences associated with C. elegans .For plates with worms (Worm + ), data are from batch digests of adult worms taken from plates at the end of each passage ( n = 50 adult worms per batch).For plates established without worms, data are for community lawns washed from plates.(B) PCA of community composition data over time (10 passages of experimental evolution) Each point represents one replicate (12 communities, two replicates each, n = 24 data points per passage).Data from worm + and worm − communities were combined for ordination and are plotted separately for clarity.(C) and (D) Relative abundance within microbial communities in passages 1, 5, and 10, (C) without and (D) with C. elegans ; data are the same as in (B).Comm unities ar e sho wn left to right in or der of community ID and replicate number (community A replicate 1, A1; A replicate 2, A2…; total n = 24).Bacterial strains are abbreviated: "AA", A. aurescens ; "BS", B. subtilis ; "CX", C. gleum (CG) or C. indologenes (CI); "MO", M. oxydans ; "OA", O. anthropi ; "RE", R. erythropolis ; "SX", S. spiritovorum (SS), or S. thalpopium (ST) (Tables1 and 2).Chryseobacterium gleum (CG) is the Chryseobacterium isolate in communities A-B, E-F, and I-J (columns1-4, 9-12, and  17-20).(E) and (F) Morphological diversity in worm + communities.(E) Emergence of alternate morphs.Passages where alternate morphotypes were observ ed ar e color ed according to the r elativ e abundance of the alternate mor ph, as a fr action of that species' total r epr esentation (fAlt).For passa ges where alternate morphs were noted but not counted, or where alternate morphs were retrieved from glycerol stocks but not from the original worm digest plates, r elativ e abundance is unknown and is assigned a v alue of −1 (blue) for visualization.Bars (blac k) indicate comm unities wher e the indicated isolate (CI or CG) was not part of the initial comm unity.Comm unity r eplicates ar e on the y -axis (in r e v erse order top to bottom, L2 → A1).(F) Colon y mor phologies in C. gleum.On the salt-free nutrient agar (NA) used for community plating, the original morph is brightl y color ed, glossy, domed, smooth, and slightly mucoid.The alternate morph is smaller and flattened, with less opacity and duller coloration.Morphologies are distinctive despite minor variation (right image, from community F1 passage 10).Two colonies with variations on the alternate morphology are circled in the center of the image ; one original morphology colony is circled on the right side of the plate.

Figure 2 .
Figure 2. Alternate morphs of C. gleum coexist in pairwise competitions with original morphs.(A) Fraction of alternate morphs in vitro in coculture with original morphs on 6 cm 1.5% NGM agar plates.Plates were handled identically to those in experimental evolution (see the section "Materials and methods").Ancestral C. gleum (WT) is used as a r efer ence.Data ar e fr action alternate mor ph (fAlt) after two passa ges (14 days) in pairwise cocultur e (time series in Figure S5 , Supporting Information ).Empty symbols indicate pairs where alternate and original morphologies are matched (from the same comm unity, r eplicate, and passa ge , e .g. A2o6 and A2a6).(B)-(D) fAlt morphs in the presence of N2 C. elegans .Experiments were conducted as in (A) except that populations were initiated with a range of initial fAlt (20%-50%) and ∼10 gravid hermaphrodites were added to each plate on day 0. (B) Summary of population trajectories in fAlt morph.Each point represents one pair of measurements at the start and end of a 7-day passage (days 0-7 and days 7-14).The 1:1 line (solid line) indicates ecologically stable population structure; when a population is on this line, the frequency of the alternate morph is the same at the start (initial) and end (final) of a passage.(C) Diameter (cm) of C. gleum colonies on worm + plates (1.5% NGM agar) 3 days after inoculation.Tests are Wilcoxon rank-sum for indicated pairs ( * , P < .05;* * , P < .01).(D) Fraction of the alternate morph at the center (near the inoculation point) and at the edge of 7-day worm + NGM plates.Tests are Wilcoxon rank-sum for fAlt morph at plate center vs .edge .

Figure 3 .
Figure 3. Surface motility of original and alternate morphs of C. gleum.(A) Motility on 0.35%, 0.5%, and 1.5% NGM a gar.Colon y diameter (cm) is the y -axis and community ID is on the x -axis."Pass 0" is represented by the ancestral WT.Each shape represents one community + replicate ( n = 4 isolates fr om single colon y pic ks for eac h combination of comm unity + r eplicate + passa ge).(B) and (C) Surface motility of original mor phs fr om passa ges 1 to 10 of communities (B)-A and (C)-F, on soft 0.35% NGM agar ( n = 4 isolates per community + passage).Isolates from passages 6 and 10 are the same as in panel (A); missing data reflect passages and communities for which colonies could not be retrieved from glycerol stocks .T he horizontal line indicates the median diameter of ancestral WT isolates ( n = 4) from the same day.Tests are Kruskal-Wallace of motility diameter at ∼1.8 da ys vs .passage .Note that community-A and community-F isolates were assayed on separate days with separate batches of freshly prepared plates, and the raw diameters cannot be compared across days due to inherent variability in these assa ys .

Figure 4 .
Figure 4. Nutrient concentration alters C. gleum surface motility on NGM agar.(A) and (B) Here, nutrient concentration was altered by changing the concentration of peptone in 0.35% NGM agar (0.2X, squares + solid lines; 1X, circles + dashed lines; or 4X, triangles + dotted lines) while holding all other components constant.Strains in these experiments were alternate and original morph isolates from the ancestral strain and from communities A2 and F2 pass 6 ( n = 3 isolates per mor ph, tec hnical triplicate; all data points shown).Isolates from communities A and F were assayed in separate runs; ancestor was run with community A. (A) Diameter (in cm) over time.(B) Expansion rates (cm/h) calculated from the data in (A).(C) and (D) Colony diameters (C) and expansion rates (D) for community A2 pass 6 isolates ( n = 3 isolates per morph) across a range of dilutions of the original inoculum, on NGM agar plates with varying peptone concentrations (0.2-4X, top axis).Dilution 0 r epr esents the inoculum used in (A) and (B) (1 μl inoculum containing ∼10 6 CFU), from which the indicated series of 10-fold dilutions were made (lo w est dilution contained ∼10 3 CFU/ μl).Lines show fits to data (y ∼ ln(x)).

Table 1 .
Bacterial strains used in this study.
• C in bead bath to allow lysis to proceed) (Shehadul Islam et al. 2017 ).Samples were rehydrated with 50 μl rehydration solution for 1h at 65 • C. DNA concentration of each sample was measured with both Qubit (Invitrogen Qubit High Sensitivity As-say) and Nanodrop (BioTek Synergy HTX plate reader, Take3 plate).Samples above 20 ng/ μl were sent to Microbial Genome Sequencing Center (SeqCenter) for 200Mbp Illumina sequencing with No-vaSeq 6000.