Supplementary Materials
Figure S1. Growth and contour-stepped differential response surfaces for descendent (blue) and ancestor
(yellow) strains across a carbon and nitrogen gradient. The descendent strains, Low-N (a-b) and Low-C
(c-d), were predicted to be competitively dominant over the ancestral strain everywhere except for low
nitrogen concentrations, driven by increased RK values and increased affinity for both carbon and
nitrogen.
METHODS
Saccharomyces cerevisiae Isolates
The strains were isolated from a “no captive cultures” winery (Kumeu River, Auckland, New
Zealand). For the work presented here, we used two wild-type, diploid isolates with differing
genotypes based on micro-satellite typing of five loci; each isolate being <50 generations
removed from the wild. These isolates are referred to as P1 and P2 following the original
identification by Goddard (2008). The isogenic strains were stored at −80 °C with 1.7 mL of
dense culture and 0.3 mL glycerol (15% v/v) in 2 mL Eppendorf tubes; this is standard practice
for storage of S. cerevisiae cultures. For the purposes of our work we considered an isogenic
strain a population, which enabled unambiguous identification of niche requirements,
competitive outcomes and evolutionary responses. Specifically, use of isogenic strains removes
within-population variation associated with the presence of different genotypes (polygenic
populations). Isolates were sub-sampled from cryo-storage into standard media for culturing S.
cerevisiae. They were inoculated into yeast extract-peptone-dextrose broth (YPD; Y1375,
Sigma-Aldrich, St. Louis, MO, USA), which at 50 g L-1 comprises 20 g L-1 bacteriological grade
peptone, 10 g L-1 yeast extract, and 20 g L-1 glucose.
Chemostat Cultures
Experiments were performed in Multitron II incubators (Infors HT, Bottmingen, Switzerland)
with DasGIP glassware (DasGIP AG, Jülich, Germany) and “200 Series” pumps (Watson-
Marlow, Wilmington, MA, USA). Each of these chemostats concurrently maintained sixteen
250-mL culture flasks at 30°C. Chemostat culture flasks were inoculated from starter cultures of
Saccharomyces cerevisiae and grown at 30°C for 48 h in batch-culture mode with non-limiting
nutrients, after which they were inoculated into the chemostats and maintained under log-phase
growth with constant resource-supply rates.
Initial tests with different incubation times suggested that populations had stable biomass values
after ≥25 generations; we therefore mapped population niche requirements after 25 generations
growth in the chemostat. This number of generations also limited the possibility that a beneficial
mutation might sweep through any one population (Goddard & Bradford 2003).
Initial tests with different carbon (C) and nitrogen (N) supply rates suggested that concentrations
of C ranging from 0.09 to 0.75 g C L-1 gave ~10-fold differences in population biomass when N
was not limiting (0.064 g N L-1), and that concentrations of N ranging from 0.004 to 0.032 g N L1
gave ~12-fold differences in population biomass when C was not limiting (1.5 g C L-1).
The medium (CYN5501 yeast nitrogen base without amino acids, (NH4)3PO4, KH2PO4 and
Inositol; ForMedium Ltd, Norwich, UK) for overnight cultures at non-limiting nutrient
conditions was combined with 10 g L-1 D-glucose monohydrate (49159; Sigma-Aldrich, St.
Louis, MO, USA), 5 g L-1 anhydrous (NH4)2SO4 (A4418; Sigma-Aldrich, St. Louis, MO, USA),
and 1 g L-1 anhydrous KH2PO4 (P9791; Sigma-Aldrich, St. Louis, MO, USA). This gave
concentrations of 4.00 g L-1 carbon, 1.06 g L-1 nitrogen, and 0.228 g L-1 phosphorous. Each
medium was adjusted to pH 6.0±0.1 using NaOH. We established in preliminary experiments
and development of our ‘Characterization Protocol’ that culture densities in the chemostat with
this medium was equivalent at the stated and double concentrations of C, N and P, confirming
the formulation was not growth-limiting. To invoke resource limitation we then fed the
chemostat cultures with the same base medium but with reduced concentrations of C and N.
The reference, “rich” formulation of our experimental culture medium provided 3.20 g C L-1,
1.00 g N L-1 and 0.200 g P L-1; as well as the manufacturer’s recommended concentration of
YNB without amino acids. The P and base concentrations were held at these concentrations for
the entire process (including under the evolution experiments). The C supply concentration of
3.20 g L-1 was below the 8 g L-1 threshold considered to drive experimental populations into
fermentative, anaerobic respiration.
Biomass was determined by ceasing the chemostat agitation and allowing the cultures to stand
for 5 min to permit the non-cellular residue to settle to the base of the flasks, at which point ~45
mL of culture was removed. The sample’s exact volume was determined before it was
centrifuged at 4 °C to collect the biomass pellet. The supernatant was then removed and the
pellet re-suspended by vortex agitation in 1 mL DI water and transferred to a 2 mL microcentrifuge tube. The process of centrifugation and re-suspending was repeated twice more to
wash the biomass clean of media. The cleaned pellet was dried at 65 °C to stable mass (~24 h),
and then weighed. From the mass and volume of culture sampled we estimated the population
mass density to be used in the second step of the characterization protocol.
The two chemostats were maintained with gyratory agitation at 105 rpm and 25 mm throw, 10.7
mL h-1 feed rate, 50±2 mL working culture volumes, and 4.69 h (0.213 h-1) culture turnover
times. Each flask had a separate nutrient medium supply, waste removal, and air supply. Waste
removal was siphoned at 3 times the rate of media supply, ensuring that culture volumes were
maintained at ~50 mL and that there was no back-flow of waste. Air supply (at 0.5 L min-1) was
passed through oil and water traps and filtered to 0.2 µm. The outlet filter on each flask
maintained a positive pressure within each flask, in an attempt to prevent microbial
contamination of the cultures.
The number of generations in our ‘Characterization cultures’ was below that required for a
mutation to sweep to high frequency in the experimental populations. Given estimates of the rate
and average fitness effects of beneficial mutation in yeast (Joseph and Hall 2004), multiple
beneficial mutations would arise each generation and escape Haldane’s sieve (Haldane 1927),
since population sizes are ~108 cells. However, the initial frequency of a newly arising mutation
will be small, equal to the reciprocal of the population size, i.e. ~10-8, implying that it would take
numerous generations for a beneficial allele to become common in the culture population
(Haldane 1924). We thus do not expect any beneficial mutations to go to high enough frequency
during the 25 generations of characterization growth to alter the resource utilization parameters
of the population. Previous work corroborates this expectation; only a handful of beneficial
mutations were identified following two thousand generations of selection in diploid yeast (Zeyl
2005).
Antibiotic Resistance
Antibiotic-resistant strains were constructed using the pYMN21 nourseothricin resistance
(NATR) plasmid following standard protocols (Janke et al. 2004). The goal was to insert the
NATr gene at position 49550 of chromosome IV. In other strains, this insertion does not affect
fitness (DWH, personal observation). The pYMN21 plasmid was obtained from Euroscarf
(Institute for Molecular Biosciences, Frankfurt, Germany). Briefly, primers were designed to
amplify the NATR gene from the plasmid. The upstream primer had 50 bases identical to
positions 49501-49550 of chromosome IV followed by19 bases of identity to the plasmid region
S1 (Janke et al 2004). The downstream primer had 50 bases identical to positions 49600-49551
of chromosome IV followed by 20 bases of identity to the plasmid region S4 (Janke et al 2004).
Sequences of the primers are as follows. Bases in bold at 3’ end of primers are identical to
plasmid sequences, bases in regular typeface are identical to appropriate regions of chromosome
IV.
Upstream primer:
5’GCTCCGCAACTTGCTTGAATTATGAGCTCTAAGATTCAAGAAGTAAATAGCGTACGC
TGCAGGTCGAC3’
Downstream primer:
5’TAAGATCACATGGTTCCTTTATCAAGTACTACTATCATTCCATTATATGACATCGATG
AATTCTCTGTCG3’
Resistant transformants were selected on nourseothricin plates (100 µg L-1), and insert location
confirmed by PCR. We also confirmed that the antibiotic-resistant isolates showed robust colony
growth on both plate types, and that our antibiotic-naïve isolates only grew on the standard
plates. The naïve strains retained the designations P1 and P2, and the resistant strains were
named P1R and P2R.
Determining Competitive Outcomes
To perform a competition, the two selected strains were grown separately in standard YPD
medium (described above) to between 1×107 and 5×107 cells mL-1. Each culture was adjusted
(diluted) with standard YPD medium to 5×106 cells mL-1 and 100 mL aliquots of the adjusted
cultures were mixed. The resulting mixed culture was used to inoculate the chemostat flasks for
the competition run: 10 mL of combined culture was added to each chemostat flask containing
50 mL of competition medium. This gave starting densities of ~4.2×106 cells mL-1 of each strain.
The competition medium was the chemostat base medium (described above) with concentrations
of C and N defined by the resource supply point being investigated.
After 25 generations, media flow and agitation were stopped and the chemostat flasks were
allowed to stand for 5 min to permit the non-cellular residue to settle. Next ~35 mL of culture
was removed and population density measured using a T100 Turbidity Meter (Oakton
Instruments, Vernon Hills, IL, USA). Then each mixed culture was diluted to ~2×104 cells mL-1
and 10 µl (~200 cells) of this dilution was inoculated onto four, 10-cm dia. agar plates and
distributed evenly by spreading with three, 5-mm diameter glass balls.
Adaptation to Nutrient Limitation
We investigated the ability of our model to identify how traits and competitive abilities differ
among genetically related individuals. To do this, we selected P1 as an ancestral population and
adapted it to two separate resource-limited regimes. The “Low-C” medium provided 0.0550 g C
L-1 and 0.0470 g N L-1, a C:N ratio of 1.17. The “Low-N” medium provided 1.80 g C L-1 and
0.00290 g N L-1, a C:N ratio of 621. The dual-limiting or “optimum” C:N ratio for the P1 isolate
(see Results), was 16.1. The C:N ratio in the “Low-N” environment was thus 38.5 times the
optimum and in the “Low-C” environment it was 0.076 times optimum. Phosphorus and YNB
w/o amino acids concentrations were held at constant, non-limiting levels. We ran sixteen
chemostat adaptation cultures – eight “Low-N” (i.e., ‘high C:N’) and eight “Low-C” (i.e., ‘LowC:N’) – for 79 days (404 generations at 4.69 h turnover). Culture samples were aseptically
extracted at several checkpoints and cryo-stored. We checked the density of each terminal
culture and selected the one from each medium that exhibited the highest density for
characterization and comparison with our growth response model. These were designated as
“Low-N” and “Low-C” isolates.
Table S1. The eight carbon (C) and nitrogen (N) concentrations under which growth
rates were measured for each fungal population and used to estimate growth trait
parameters and corresponding niche spaces.
C, g L-1 N, g L-1
0.094
0.188
0.064
0.375
0.750
1.500
0.032
0.016
0.008
0.004
Table S2. Initial round of competitive pairings with 8 replicates each and an exact enumeration of CFU counts. [C]: carbon resource
concentration, g L-1; [N]: nitrogen resource concentration, g L-1; Pop. Y: non-antibiotic-resistant population; Pop. Z: antibioticresistant population; RY: predicted realized growth rate of population Y; RZ: predicted realized growth rate of population Z; ΔR:
difference of predicted realized growth rates (RZ-RY); Prop. Z: mean proportion of antibiotic-resistant CFUs across the 8 replicates
(calculated as ‘CFUs on antibiotic plate / Total CFUs’); S.E.: standard error of the proportion; 95% C.I.: 95% confidence interval of
the proportion; P: p-value testing the null hypothesis that the proportion is equal to 0.5; Pred. Win: predicted winner based on ΔR;
Win by Prop: winning strain based on relative proportion of CFUs on antibiotic plates relative to standard plates; Win by CI: winner
based on confidence interval being less than, greater than, or crossing 0.5. The symbol ‘?’ indicates stochastic dominance (Predicted
Winner column) or that no clear winner was observed across the replicates (Actual Winner).
Actual
Winner
0.069
0.64
(0.52, 0.73)
0.025
P2R
P2R
0.709
-0.051
0.40
(0.35, 0.46)
0.004
P2
P2
0.642
0.641
-0.001
0.43
(0.31, 0.54)
0.179
?
?
0.370
0.372
0.002
0.48
(0.31, 0.64)
0.756
?
?
0.710
0.778
P2
P1R
0.759
P1
P1
R
P2
R
R
P1
1.2
0.08
0.01
Predicted
Winner
P2R
0.08
0.2
P
ΔR
1.2
0.08
95% C.I.
RZ
[N]
0.7
Prop. Z
RY
[C]
Pop. Y Pop.Z
P2
0.3
0.04
Low-N
P1
0.542
0.485
-0.057
0.09
(0.00, 0.24)
<0.001
Low-N
Low-N
0.6
0.02
Low-N
P1R
0.534
0.478
-0.056
0.26
(0.18, 0.33)
<0.001
Low-N
Low-N
Low-N
P1
R
0.690
0.641
-0.049
0.16
(0.00, 0.35)
0.005
Low-N
Low-N
P1
R
0.842
0.814
-0.028
0.21
(0.17, 0.26)
<0.001
Low-N
Low-N
R
0.7
6.0
0.05
0.40
Low-N
0.3
0.04
Low-C
P1
0.540
0.485
-0.055
0.30
(0.25, 0.34)
<0.001
Low-C
Low-C
0.3
0.4
Low-C
P1R
0.540
0.485
-0.055
0.30
(0.21, 0.39)
0.002
Low-C
Low-C
Low-C
P1
R
0.604
0.559
-0.045
0.27
(0.21, 0.34)
<0.001
Low-C
Low-C
P1
R
0.840
0.811
-0.029
0.25
(0.16, 0.33)
<0.001
Low-C
Low-C
0.6
5.6
0.03
0.36
Low-C
Table S3. Second round of competitive pairings with 2 replicates each and a visual estimation of CFU proportions. [C]: carbon
resource concentration, g L-1; [N]: nitrogen resource concentration, g L-1; Pop. Y: non-antibiotic-resistant population; RY: predicted
realized growth rate of population Y; Pop. Z: antibiotic-resistant population; RZ: predicted realized growth rate of population Z; ΔR:
difference of predicted realized growth rates (RZ-RY); Pred. Win: predicted winner based on ΔR; Win by Prop.: winning strain based
on relative proportion of CFUs on antibiotic plates relative to standard plates. The symbol ‘?’ indicates that either stochastic
dominance was predicted or that no clear winner was observed across the replicates.
[C]
[N]
Pop. Y
Pop. Z
RY
RZ
ΔR
Pred.
Win
Win by
Prop.
0.2
0.010
P1
P1R
0.363
0.333
-0.030
P1
P1
P1
P1
R
0.363
0.333
-0.030
P1
P1
P1
R
0.407
0.400
-0.007
?
?
P1
R
0.710
0.713
0.003
?
R
0.491
0.524
1.2
0.2
1.2
0.010
0.100
0.100
P1
P1
?
0.033
P2
R
P2R
0.3
0.025
P1
P2
1.2
0.025
P1
P2R
0.549
0.581
0.032
P2R
P2R
0.3
0.100
P1
P2R
0.491
0.524
0.033
P2R
P2R
1.2
0.100
P1
P2R
0.710
0.778
0.069
P2R
P2R
0.3
0.025
P2
P1R
0.521
0.485
-0.036
P2
P2
R
1.2
0.025
P2
P1
0.571
0.523
-0.049
P2
P2
0.3
0.100
P2
P1R
0.521
0.485
-0.036
P2
P2
1.2
0.100
P2
P1R
0.759
0.713
-0.047
P2
P2
P2
P2
R
0.370
0.372
0.001
?
P2
P2
R
0.370
0.372
0.001
?
?
R
0.431
0.430
-0.001
?
P2
0.759
0.778
0.019
P2R
P2R
0.2
1.2
0.010
0.010
P2
0.2
0.100
P2
P2
1.2
0.100
P2
P2R
References
1.
Goddard, M.R. & Bradford, M.A. (2003). The adaptive response of a natural microbial population to
carbon- and nitrogen-limitation. Ecol. Lett., 6, 594–598.
2.
Janke, C., Magiera, M.M., Rathfelder, N., Taxis, C., Reber, S., Maekawa, H., et al. (2004). A versatile
toolbox for PCR-based tagging of yeast genes: New fluorescent proteins, more markers and promoter
substitution cassettes. Yeast, 21, 947–962.