Overview of data. (a) Workflow for data analysis. Metabolomics data was collected at 4 different temperatures (4, 13, 22, and 37 degrees Celsius) across multiple time points. The concentrations were plotted against time, and the rates of change were estimated. For measurements that were fit as either linear or nonlinear profiles, log₂(rate) was plotted against temperature in order to calculate the temperature coefficient. (b) Histograms show the distribution of temperature coefficients for the linear and nonlinear measurements separately; vertical dashed lines are at Q₁₀ = 2 and 3, the typical estimated range for biological systems. The inset pie chart shows the percentage of measurements that used a linear and nonlinear fit; the remaining portion (26%) represents measurements that were either noisy or did not show a clear trend. (c) PCA plots show the same but accelerated trend at the four different temperatures; the Day 7 measurement is colored the same (cyan) in each plot for reference. + indicates the mean of three replicates at a particular time point.