Constant and variable warming diffeentially shape bacterial coexistence through phage-mediated interactions

Here, we experimentally tested how constant and variable warming (both +4°C above ambient, but negligible versus high thermal variance) affect the coexistence of two Pseudomonas species in the presence of their lytic phage. We used soil temperatures from Bern, Switzerland for a 9 day coexistence experiment involving these bacteria and their phages. This data set includes flow cytometry data, growth curves of bacteria at different temperatures, and phage growth at different temperatures from this experiment. The methods are as follows for our experiment: Species and culture conditions We used two bacterial species in our experiments, Pseudomonas protegens CHAO and Pseudomonas putida MM1. We used two phage species, φGP100 and Psp1, that are specialized predators infecting P. protegens or P. putida respectively, but are unable to actively infect the other species** . All bacteria were grown in M9 minimal media made using 100 ml of 5x M9 minimal salts with 4 ml of 20% Glucose (W/V), 6 ml of 20% Arabinose (W/V), 1 ml of MgSO47H2O (1M), 50 µl of CaCl2 (1M) and then filled to 500 ml with milliQ water. This formulation, hereafter M9 60:40, was used in all experiments because differential arabinose utilization by P. putida (but not P. protegens) supports their stable coexistence (Supplementary Fig. 1). We used Sodium Magnesium (SMG) buffer for our dilutions with 5.8 g NaCl, 2 g MgSO4·7H2O, 50 mL 1M Tris HCl, (pH 7.5), 5 ml of 2% gelatine solution (filter sterilized) and milliQ water to 1 L total volume. Populations of both P. protegens and P. putida were grown up as overnight cultures at 28°C, shaking at 220 RPM in 4 ml of M9 media 60:40. These overnight cultures were then re-inoculated at 28°C in 20 mL of M9 60:40. Experimental design We established a two-species bacterial community composed of the inferior competitor P. putida and the superior competitor P. protegens. These bacteria were exposed to specialized phage predators, Psp1 (for P. putida) and φGP100 (for P. protegens). We conducted a 9-day serial transfer experiment under three temperature regimes: (1) ambient, reflecting local summer conditions (details below); (2) constant warming (ambient + 4°C); and (3) variable warming (ambient + 4°C with stochastic fluctuations generated using a sinusoidal model to simulate diurnal temperature variability). The ambient temperature profiles were derived from MeteoSwiss records for Bern (46.993° N, 7.467° E) , Switzerland (2013–2023 averages from June, July and August). The constant and variable warming treatments shared the same mean temperature but differed in temporal structure: constant warming applied a uniform +4°C offset, while variable warming introduced fluctuations around the daily mean, varying the magnitude of warming across each day (Supplementary Fig. 2). The +4°C offset reflects end-of-century warming projected for Switzerland under high-emission scenarios . These temperature regimes were programmed into Percival incubators (Model E-36L2, Percival Scientific Inc., Perry, IA, USA) using the built-in ramping function to simulate diurnal thermal fluctuations. These temperatures were also measured in real time using a HOBO Pendant MX2201 Bluetooth data logger (Onset Computer, Bourne, Massachusetts, USA) to record the temperature every five minutes. Each combination of bacteria and phages was tested in a full factorial design with six replicates across all biotic and temperature treatments, including blanks that only contain media to monitor for bacterial cross-contamination. Populations were cultivated in 96 deep-well plates (2 ml) and transferred after 24 hours of growth into a new 96 deep-well plate with 1ml of fresh media per each well (1:100). We used 100 µl of culture mixed with 100 µl of 50% glycerol to make freezer stocks at -80° C each day. Bacterial abundance was quantified by flow cytometry (Invitrogen Attune CytPix with CytKick Autosampler; Thermo Fisher Scientific, USA) on Days 1, 3, 5, 7, and 9 post-inoculations. The flow cytometry outputs were first size gated with the SSC-A (Side Scatter Area) and the SSC-H (Side Scatter Height) to filter debris and doublets followed by gating using the FSC-A (Forward Scatter Area) and fluorescent detector channel BL1-A (488 nm). We measured the cell counts of both bacterial species using the flowCore package version 2.20 in R statistical software . Thermal performance curves To understand the mechanisms underlying phage-mediated bacterial interactions under different warming regimes, we characterized thermal performance curves for all species used in the serial transfer experiment. The temperature gradient used to generate performance curves spanned the range experienced across the three experimental treatments (12–40°C). For the measurements of phage thermal performance curves, we used the ancestral bacterial species for the bacterial lawns and grew them on LB agar plates with LB overlay agar. LB agar being made with 20 g of LB Medium (Lennox) from Carl Roth (Art. No. X964.2) and 15g of Agar-Agar, Kobe I Carl Roth (Art. No. 5210.2) for 1 L of LB agar. LB overlay agar is made with 20 g of LB Medium (Lennox) from Carl Roth (Art. No. X964.2) and 7g of Agar-Agar, Kobe I Carl Roth (Art. No. 5210.2) for 1 L of LB overlay agar. Phage stocks of Psp1 and φGP100 were standardized to 108 PFUs/mL and the overnight P. protegens and P. putida were re-inoculated and grown at 220 RPM and 28°C until reaching an OD600~ of ~1 (~107 CFUs/mL) and are in exponential growth phase. Eppendorf tubes are used with 1 ml of bacterial culture (~107 CFUs/ml) and 10 µl of phage stock (~106 PFUs/mL) with three replicates of each bacteria-phage combination at each temperature. We measured seven different temperatures for each set of bacteria and phages to include the entire range of thermal conditions of our experiment where the lowest recorded temperature was 15.96° C and the highest was 35.95°C. These temperatures were 12°, 16°, 20°, 24°, 28°, 32°, 36°, and 40° C. Phage-bacteria culture combinations were incubated in Percival E-36L2 incubators for four hours at their set temperature. After incubation, 100 µl of chloroform was added to each tube, three tubes of the initial phage-bacteria inoculum were also chloroformed. These tubes were vortexed and centrifuged at 10,000 x g for 5 minutes separating the aqueous phase (containing phages) from the chloroform and cell debris. 20 µl from the aqueous phase was taken from each tube to make a dilution series. Phages were plated on overlay agar plates and incubated at 28° C overnight. We then counted the plaques of the initial phages and phages grown over four hours of inoculation calculating PFUs as Plaques × Dilution Factor. The growth rates of our bacteriophages were calculated using equation (1) where N4 is the PFUs after 4 hours and N0 is the initial PFUs. (1) For bacterial thermal performance curves, each species was grown in overnight cultures at 28°C with shaking at 220 RPM that was then re-inoculated into 96 microwell plates with a dilution of 1:100. We used bacterial cultures from ancestral bacteria as well as from day 5 and day nine clones to evaluate the bacterial population’s acclimatization to heat via their thermal performance curves. We also examined each bacterial population with and without their specialized phage to measure the effects of phages on population growth. We read the OD600 values of our microplates over 24 hours at 20°, 24°, 28°, 32°, 36° and 40° C using a Biotek Synergy H1 microplate reader (Agilent, Santa Clara, California, United States) every 10 minutes. We then analyse these growth curves in R to get the growth rate (r) and area under the curve (AUC) as measures of bacterial growth that were then used in our analysis . Data analysis Serial transfer experiment All analysis was done in R (v 4.5.0) . Mixed effect models were implemented in R using glmmTMB package (v 1.1.13) , Effsize DHARMa, and emmeans. Our fixed effects were warming regimes with three levels (ambient, + 4°C (constant), + 4°C (variable)), predation with four levels (no predator present, specialized predator, predator of other bacteria, both predators), competition with two levels (competitor, no competitor), and experimental days with two levels (three versus nine). We chose these two days as comparisons as day three marked a sufficient acclimatization time and further showed a local maximum for population sizes for some treatments (time series data shown in Supplementary Fig.3), whereas day nine represented the end of the experiment (Supplementary Fig. 3). The well position, i.e., where each population was located within the plate, was treated as a random factor within our model. Cohen’s d effect sizes were calculated for the population response of both bacterial species across all treatment combinations using the Effsize package. Linear model assumptions (normality of residuals and homogeneity of variance) were assessed visually using diagnostic plots with DHARMa, and response variables were log-transformed when these assumptions were violated. Thermal performance curves Thermal performance curves were analyzed using the growthrates package in R (v 0.8.5) for growth measures such as growth rates, and the Area Under the Curve (AUC). The thermal performance curves for bacteria and phages were calculated with the rTPC package (v 1.0.4) and the nls.multstart package (v 2.0.0) using growth rates calculated from the growthrates package for bacteria and the manually calculated growth rate for phages following the methods explained in Padfield et. al. (2021).