The community background alters the evolution of thermal performance
Abstract
Microbes are key drivers of global biogeochemical cycles and their functional roles are heavily dependent on temperature. Large population sizes and rapid turnover rates mean that the predominant response of microbes to environmental warming is likely to be evolutionary, yet our understanding of evolutionary responses to temperature change in microbial systems is rudimentary. Natural microbial communities are diverse assemblages of interacting taxa. However, most studies investigating the evolutionary response of bacteria to temperature change are focused on monocultures. Here we utilise high throughput experimental evolution of bacteria in both monoculture and community contexts along a thermal gradient to determine how interspecific interactions influence the thermal adaptation of community members. We found that community-evolved isolates tended towards higher maximum growth rates across the temperature gradient compared to their monoculture-evolved counterparts. We also saw little evidence of systematic evolutionary change in the shapes of bacterial thermal tolerance curves along the thermal gradient. However, the effect of community background and selection temperature on the evolution of thermal tolerance curves was variable and highly taxon-specific – with some taxa exhibiting pronounced changes in thermal tolerance, while others were less impacted. We also found that temperature acted as a strong environmental filter, resulting in the local extinction of taxa along the thermal gradient, implying that temperature-driven ecological change was a key factor shaping the community background upon which evolutionary selection can operate. These findings offer novel insight into how the community background impacts thermal adaptation.
README: The community background alters the evolution of thermal performance
GENERAL INFORMATION
Corresponding author
- Name: Joseph Westley
- Institution: University of Exeter
- Email: jw1235@exeter.ac.uk
Principal Investigator
- Name: Prof. Gabriel Yvon-Durocher
- Institution: University of Exeter
- Email: G.Yvon-Durocher@exeter.ac.uk
Co-author 1
- Name: Dr. Francisca C. García
- Institution: King Abdullah University of Science and Technology (KAUST)
- Email: paquigrcgrc@gmail.com
Co-author 2
- Name: Ruth Warfield
- Institution: University of Exeter
- Email: R.Warfield@exeter.ac.uk
Date of data collection: 2020
SHARING/ACCESS INFORMATION
Recommended citation for this dataset: Westley, J., García, F. C., Warfield, R., Yvon-Durocher, G. (2024). Data from: The community background alters the evolution of thermal performance. Dryad Digital Repository. doi.org/10.5061/dryad.vq83bk41b
Publication associated with this dataset: Westley, J., García, F. C., Warfield, R., Yvon-Durocher, G. (2024). The community background alters the evolution of thermal performance. Evolution Letters.
Data description and file structure
Contained within the directory "datafiles" are both the raw data files, partially processed data files, and a single processed data file used in all analyses. Raw and partially processed data files for ancestral and monoculture-evolved isolates are combined and are found in the directory "datafiles/monraw". Raw and partially processed data files for community-evolved isolates are located in "datafiles/comraw". The fully processed data file, "final_data.csv", used in all statistical analyses is in the "datafiles" directory (not within a sub-directory).
Processed data
- "final_data.csv"
- Description: A single file containing all growth rate data for monoculture, community, and ancestral isolates
- Number of columns/variables: 11
- Number of rows/observations: 1421
- Variable List:
- r: maximum growth rate per hour r(h−1)
- K: maximum optical density reached at a wavelength of 600nm (OD600)
- T0.biom: The OD600 at the point at which cultures were inoculated
- AIC: The Akaike information criterion for the fit of the logistic growth curve
- Id: A variable containing the ID of the isolate. For example, in "OTU2-T1-15-2", "OTU2" means the taxonomic identity is OTU2 (Pseudomonas spp.), "T1" means it was evolved in treatment 1 (monoculture), "15" means it was evolved at 15°C, and "2" means this is biological replicate 2
- temp.c: This is the growth temperature in Celsius
- temp: This is the growth temperature in kelvin
- ID.a: A variable containing the ID but without distinguishing between biological replicates of the same experimental unit. For example, in "OTU2-T1-15", "OTU2" means the taxonomic identity is OTU2 (Pseudomonas spp.), "T1" means it was evolved in treatment 1 (monoculture), and "15" means it was evolved at 15°C.
- evotemp: The temperature an isolate was evolved at in Celsius
- OTU: The taxonomic identity of the isolate
- Treatment: Whether the isolate is ancestral, monoculture-evolved, or community evolved
- Specific abbreviations:
- T0 = ancestral isolate, T1 = monoculture evolved isolate, T2 = community evolved isolate
- OTU2 = Pseudomonas spp., OTU15 = Serratia spp., OTU18 = Aeromonas spp., OTU20 = Herbaspirillum spp., OTU23 = Janthinobacterium spp.
- survival.csv
- Taxon: The taxonomic identity of the isolate
- Evolution_temperature: The temperature in Celsius at which the isolate was evolved
- Replicates_survived: The count of biological replicates for which the respective isolate survived to the end of the community evolution experiment, out of a total of three
Raw and partially processed data
The following is an explanation of the structure of "datafiles/monraw", but the same file structure is used in "datafiles/comraw". Within "datafiles/monraw" there are the following files:
- 192 Raw OD600 measurement files, following the naming format of "T0_15_P1.csv"
- Description: These are raw OD600 files output by the Thermo Scientific Multiskan Sky Microplate Spectrophotometer recording at a wavelength of 600 nm. There is a file for all combinations of time points of the growth assay, temperature of the growth assay, and plate identity (plate 1 or plate 2). In the example file name, "T0_15_P1.csv", "T0" refers to timepoint 0 (when the culture was inoculated, not to be confused with the treatment factor level "T0", which denotes ancestral isolates), "15" denotes that the plate was grown at 15°C, and P1 denotes that the data is for plate one.
- Dataframe structure: These files do not follow a typical "tidy" or "long form" data structure. Cells are populated by values in the shape of a 96-well plate, where the columns are numbered 1-12, and rows contain sequential letters A-H. For example, the value in row A, column 1, denotes the OD600 for well A1 of the plate being measured.
- "Data_mon.csv"
- Description: This file contains all data from all 192 raw data files described above collated into an R object in a "tidy" or "long" format.
- Number of columns/variables: 12
- Number of rows/observations: 11520
- Variable List:
- Replicate: A number designating the biological replicate for the respective experimental unit (note: in the analogous community datafile "Data_com.csv", the replicate variable is named community instead of Replicate)
- od_cor: The OD600 measure was corrected to remove the absorbance of the culture media
- OTU: The taxonomic identity of the isolate
- Treatment: The treatment group that the isolate was evolved in. For example, in "T1-15", "T1" denotes that the isolate evolved in a monoculture, and "15" denotes that the isolates evolved at 15°C.
- Timepoint: An integer value specifying the timepoint that the measure was taken, e.g., "0" means at inoculation, "1" is the first measure post-inoculation, etc.
- growthtemp: The temperature the plate was grown at in Celsius.
- timestampcode: A variable that contains a combination of the growth temperature and the timepoint, e.g., "0-15" denotes timepoint 0 and a growth temperature of 15°C.
- Hours: The exact time in hours since culture inoculation that the OD600 reading was recorded
- Specific abbreviations:
- T0 = ancestral isolate, T1 = monoculture evolved isolate, T2 = community evolved isolate
- OTU2 = Pseudomonas spp., OTU15 = Serratia spp., OTU18 = Aeromonas spp., OTU20 = Herbaspirillum spp., OTU23 = Janthinobacterium spp.
- "mon_out.csv"
- Description: A file containing all specific OD600 measurements to be removed from the "Data_mon.csv" data frame prior to construction of logistic growth curves (these are typically measurements occurring after carrying capacity has been reached)
- Number of columns/variables: 3
- Number of rows/observations: 630
- Variable List:
- t: The time in hours that the datapoint occurs
- LOG10N: The OD600 measure of the datapoint
- pa: A variable containing the ID of the isolate. For example, in "OTU2-T1-15-2", "OTU2" means the taxonomic identity is OTU2 (Pseudomonas spp.), "T1" means it was evolved in treatment 1 (monoculture), "15" means it was evolved at 15°C, and "2" means this is biological replicate 2
- Specific abbreviations:
- T0 = ancestral isolate, T1 = monoculture evolved isolate, T2 = community evolved isolate
- OTU2 = Pseudomonas spp., OTU15 = Serratia spp., OTU18 = Aeromonas spp., OTU20 = Herbaspirillum spp., OTU23 = Janthinobacterium spp.
- "mon_curves_outrem.csv"
- Description: A file containing maximum growth rate data for all monoculture evolved isolates across all growth and evolution temperatures "mon_curves_outrem.csv". Each curve is for a single biological replicate and is produced by fitting a logistic growth model to the OD600 measurements
- Number of columns/variables: 6
- Number of rows/observations: 960
- Variable List:
- pa: A variable containing the ID of the isolate. For example, in "OTU2-T1-15-2", "OTU2" means the taxonomic identity is OTU2 (Pseudomonas spp.), "T1" means it was evolved in treatment 1 (monoculture), "15" means it was evolved at 15°C, and "2" means this is biological replicate 2
- r: maximum growth rate per hour r(h−1)
- k: maximum OD600 reached
- T0.biom: The OD600 at the point at which cultures were inoculated
- AIC: The Akaike information criterion for the fit of the logistic growth curve
- quasi_r2: A "quasi" or "pseudo" r squared value for the fit of the logistic growth curve
- "Montimestamps.csv"
- Description: A file containing the actual length of time in hours since inoculation (timepoint 0) that each OD600 measurement was taken, e.g., OD600 measurements at timepoint 1 for plates grown at 15°C were taken 4.05 hours after inoculation
- Number of columns/variables: 3
- Number of rows/observations: 96
- Variable List:
- Timepoint: An integer value specifying the timepoint that the measure was taken, e.g., "0" means at inoculation, "1" is the first measure post-inoculation, etc.
- growthtemp: The temperature the plate was grown at in Celsius.
- Hours: The exact time in hours since culture inoculation that the OD600 reading was recorded
Code
In the scripts directory, all data processing steps are numbered sequentially. In summary, these scripts perform the following:
Step 1: Collating raw OD600 data
GRE_mon_step1.R and GRE_com_step1.R each collate the raw 96 well plate OD600 datafiles for all time points (e.g., files following the naming format of "T0_15_P1.csv") into an R object.
Step 2: Converting data to 'tidy' format
GRE_mon_step2.R and GRE_com_step2.R take the outputs from step 1 and convert them to a "tidy" format, producing the files "Data_mon.csv" and "Data_com.csv".
Step 3: Creating logistic growth curves
GRE_mon_logcurves_step3.R and GRE_com_logcurves_step3.R take the outputs from step2 ("Data_mon.csv" and "Data_com.csv" respectively) and produce individual logistic growth curves for each isolate ("mon_curves_outrem.csv" and "com_curves_outrem.csv" respectively). "mon_out.csv" and "com_out.csv" are also produced at this stage, and include OD measurements to be removed, prior to growth curve estimation (for example, OD measurements showing a decline occurring after carrying capacity is reached would be removed). Plots of these growth curves are also written to the 'plots_growth_curves' directory as mon_curves_outrem.pdf and com_curves_outrem.pdf (not included in this repository).
Step 4: Consolidating monoculture-evolved, ancestral, and community-evolved data into a single .csv file
GRE_data_consolidation_step4.R collates the monoculture and community growth curve data ("mon_curves_outrem.csv" and "com_curves_outrem.csv") into the single file, "final_data.csv". Curves where carrying capacity is not reached or there is no growth are removed at this stage.
Step 5: Conducting growth rate analyses and creating Figures 1 and 2
GAMM_r_analysis_step5.R takes "final_data.csv" as input and produces generalised additive mixed-effects models (GAMMs), conducts model comparison via AICc to select the best models, and produces Figure 1 and Figure 2 for the manuscript, based on the predictions of these best models. Additionally, post-hoc analysis is conducted using R package emmeans to get effect sizes and the significance of pairwise differences. See the section 'Statistical analysis' within 'Materials and Methods' of the manuscript for more detailed methods.
Step 6: Conducting survival analysis and creating Figure 3
Survival_analysis_step6.R also takes "final_data.csv" as input and creates a heatmap showing the number of replicates of each community member that survive to the end of the community experiment, at each evolution temperature. Additionally, a binomial model is used here to test if survival to the end of the community experiment depends on taxonomic identity and evolution temperature.
Methods
Study taxa
Study taxa were derived from biofilm samples collected in May 2016- May 2017 from rock surfaces in several freshwater streams in Hvergerdi Valley, Iceland (64.02, −21.18). These samples were frozen in a 17% glycerol solution after collection and were stored at -20°C. The freshwater streams from which they originated ranged in temperature from 7°C - 38°C, due to variation in the levels of geothermal warming at the site (O’Gorman et al., 2014). On return to the laboratory, samples were thawed at 20°C. The solution they were transported in was then diluted consecutively, and 10 µL of solution was spread onto agar plates and incubated for 10 days at 20°C. Samples were taken from a random selection of the resulting colonies and were placed into 200 µL of lysogeny broth and incubated for 48 hours. This inoculated lysogeny broth was then centrifuged, and the supernatant was discarded. The pellet of bacterial cells was then placed into a lysogeny broth containing 17% glycerol and was frozen at -80°C.
16S PCR was performed for these samples, and the resulting rRNA was sequenced using Sanger sequencing, and taxonomy was assigned by comparing these sequences with existing databases (see (García et al., 2018)). The specific methodology is as follows: A master-mix solution was created and consisted of 7.2 μl of DNA-free water, 0.4 μl of 27 forward primer, 0.4 μl of 1492 reverse primer and 10 μl of Taq polymerase, per sample. A template solution was prepared by adding 2 μl of the sample diluted 100 x in DNA free water, to 18 μl of master-mix solution. These samples were then placed in a thermal cycler (Applied Biosystems Veriti Thermal Cycler). The cycling protocol consisted of 1 cycle at 94°C for 4 minutes, 35 cycles at 94, 48 and 72°C for 1 minute, 30s, and 2 minutes, respectively, and finally, 1 cycle at 72°C for 8 minutes. The final product of the PCR was cleaned using Exonuclease I and Antarctic Phosphatase. Sanger sequencing was conducted on high-quality samples using the 27F, 1492R primers (Core Genomic Facility, University of Sheffield). Geneious (version 6.1.8, (Kearse et al., 2012) was used to trim the sequences, removing the bp from the 5' end and trimming the 3' end to a maximum length of 1000bp. Sequences longer than 974bp were then aligned to the Silva.Bacteria.Fasta database using Mothur version 1.39.5 (Schloss et al., 2009) and the RDP trainset 9 032012 was used as a reference database to assign taxonomy to the isolates. A total of 36 different taxa were identified, and from these five were chosen for use in this study. These five taxa were chosen as they differed in their thermal traits, and in their colony morphologies, the latter requirement being to facilitate visual identification when cultures consisting of more than one taxon were grown on agar. The five taxa chosen for this study and the Genbank accession number were: Pseudomonas spp. (w_Ic161A, MZ506751), Serratia spp. (h_Ic174, MZ506746), Aeromonas spp. (n_Ic167, MZ506748), Herbaspirillum spp. (j_Ic165, MZ506747), and Janthinobacterium spp. (h_Ic161A, MZ506745).
Evolution of bacteria in monocultures and communities
Bacterial communities comprising all five taxa, as well as monocultures of each taxon, were evolved at temperatures ranging from 15°C - 42°C for ~110 generations. We used 110 generations as past research suggests this would be ample time for the communities to reach an equilibrium. In a previous community evolution experiment conducted at 20°C, it was observed that the majority of communities reached stability after approximately 50 generations (García et al., 2023). Earlier investigations passaging natural communities indicated that around 60 generations were needed for most communities to achieve population equilibria in various instances (Goldford et al., 2018). In the current study, we collected ‘initial’ growth rate data following 2-3 transfers (~10 generations) to allow communities to acclimate to the temperature (mainly to avoid acute stress responses). We then subsequently gathered data at approximately 100 generations later (~110 generations total). The time to reach this number of generations was calculated for the colder evolution temperature groups, to ensure all treatment groups reached a minimum of ~100 generations.
The specific methodology follows: An initial stock solution for each taxon was created from a single colony clone, using lysogeny broth, which was then incubated overnight at 20°C. These were then standardised to a common optical density with R2 media, and then a community stock solution was constructed by combining 100 µL of each of the five taxa. 40 µL of stock solution was then used to inoculate 5000 µL of R2 media. Three replicates of these inoculated media were then incubated at each of the following temperatures: 15°C, 20°C, 23°C, 27°C, 30°C, 33°C, 37°C, and 42°C. This was then repeated, but instead of inoculation with community stock solution, monoculture stock solution was used, ensuring the same starting biomass of each taxon for each treatment group. Every 48 hours during incubation, 40 µL was removed from each culture and was used to inoculate a fresh 5000 µL of R2 media, to prevent resource limitation from occurring. This was done 18 times, equating to ~110 generations. At the end of the experiment, serial dilutions of the resulting cultures were then grown on agar, and samples of individual taxa were isolated and frozen at -80°C in 17% glycerol. For the community cultures, individual taxa were identified based on colony morphology.
Growth assay of evolved isolates
From every evolution experiment a single clone was isolated. These isolates, as well as the original ancestral samples, were then grown at temperatures ranging from 15°C - 42°C. Maximum growth rates (r(h-1)) were calculated at each temperature. The specific methodology is as follows:
Every evolved isolate, as well as the original ancestral taxa, were thawed in R2 growth media at 20°C for 24 hours. These cultures were then diluted with more R2 media until all cultures were at an optical density (OD600) of 0.05, measured using a Themo ScientificTM Multiskan Sky Microplate Spectrophotometer, at a wavelength of 600nm. 200 µL of each culture was then transferred into 96 well plates. Control ‘blank’ wells were filled with only R2 medium. The plate was then incubated at 15 °C until carrying capacity was reached (~54 hours), and OD600 measurements were taken every ~4 hours. This process was repeated for all isolates, at incubation temperatures of 20°C, 23°C, 27°C, 30°C, 33°C, 37°C, and 42°C. Due to handling time, there was some variation in measurement intervals, but in all analyses exact intervals calculated from timestamps are used. The mean OD600 value for blank wells in a plate was subtracted from all OD600 measurements.
Fitting growth curves
All modelling of growth curves was conducted in R version 4.0.2 (R Core Team, 2021), using the nlsLoop package (Padfield, 2016/2020). The maximum growth rates (r (h−1), hereafter simply r, or maximum growth rate) for each incubated culture (see section 2.3) were calculated by fitting the logistic growth equation to the OD600 measurements, using non-linear least squares regression.
Nt = K/(1 +〖Ae〗^(-rt) ); A = (K - N_0)/N0 (1)
In equation 1, Nt is the biomass at time, t, K is the carrying capacity, N0 is the starting biomass and r is the exponential population growth rate (r(h−1)). For some cultures, after reaching carrying capacity there would be a slow decline in cell density. As the above model cannot estimate this decline, these datapoints demonstrating a post-asymptote decline were removed.
Statistical analysis
All statistical analyses were conducted in R version 4.0.2 (R Core Team, 2021), and all plots were created using ggplot2, and other tidyverse (Wickham et al., 2019) packages were used for data handling. For all analyses, monoculture-evolved isolates were only included if their respective isolate had survived to the end of the community evolution experiment. For example, at an evolution temperature of 33°C, only Aeromonas spp. and Herbaspirillum spp. survived to the end of the evolution experiment, therefore only Aeromonas spp. and Herbaspirillum spp. isolates that were evolved at 33°C in monoculture were included in the analysis, and Pseudomonas spp., Serratia spp., and Janthinobacterium spp. isolates that were evolved at 33°C in monoculture were excluded from the analysis. This prevented a survivorship bias from confounding the results.
For the within-taxon analysis, separate generalised additive mixed-effect models (GAMMs) were fit for each taxon, using the function uGamm from the R package MuMIn (Bartoń, 2022). The initial full model included maximum growth rate as the response variable, and the following fixed effects and smoothing terms: evolution temperature, treatment (monoculture, ancestral, or community evolved), an interaction between evolution temperature and treatment, and a smoothing term on growth temperature, which was allowed to vary by treatment. A single random effect encompassing taxon and biological replicate (at the level of each evolved lineage) was included in all models. All possible sub-models were created and compared by their sample-corrected Akaike information criterion (AICc) using the function AICc from the package MuMIn, although models without a smoothing term on growth temperature were not considered. The threshold for determining a significant difference between models was when ΔAICc was >2. Where there were one or more models falling within 2 ΔAICc of the lowest AICc model, the more minimal model was selected as the best model. The R package emmeans (Lenth, 2017/2023) was used to conduct post-hoc pairwise comparisons for model estimates across both evolution temperatures and treatment group.
For the ‘all taxa combined’ analysis, model creation and selection was conducted in the same way as the within taxon analysis. Fixed and random effects were the same as in the within taxon analysis. However, due to issues with rank deficiency when trying to incorporate ancestral data into this model, a separate model for the effect of growth temperature on growth rate for the ancestor was constructed. This negates pairwise significance testing of differences between ancestral and evolved lineages but does allow both visual comparison of TPCs by superimposing the ancestral model predictions onto a figure displaying the predictions for the monoculture and community evolved isolates (Figure 1).
For the community survival analysis, binomial Generalised linear models were fit using the base R glm function, and fixed effects taxon identify and evolution temperature. Fixed effects were determined to be significant if the ΔAICc of their removal was >2.