Fig. S2

Fig. S2 Temperature increases variability in specific cell types and stages of development, related to Figure 2 (A) Scatterplot comparing whole-embryo staging determined by either cell composition (x axis) or transcriptional data (y axis); inset bar plots show CV for each metric in groups of embryos sampled at various time points (time points in hpf indicated in gray text above each bar). Loess fit lines are shown for embryos raised at each temperature: 28°C (gray), 32°C (pink), and 34°C (red). (B) Stacked bar chart showing how the number of cell types with significant abundance changes increases with temperature; this phenomenon held for all pseudostage bins (in color) except bin 4, which lacked sufficient control embryos to analyze statistically. (C) Scatterplot showing positive relationship of log2 fold change in abundances in both elevated temperatures, colored as in (A). (D) Histogram showing changes in cell composition resulting from 34°C treatment, as predicted by multinomial model. (E) Scatterplots showing raw data (black points) and model outputs of mean (solid line) and variability (ribbons with dotted line) for three temperature-sensitive cell types. (F) Heatmap of each cell type showing relative increases in observed variability compared with variability expected during wild-type development through developmental time (pseudostage bins); shades of red indicate increased variability, while white to blue shades show neutral or decreased variability. The upper panel shows the average CVs for each temperature across all stages, highlighting pseudostages with increased variability at higher temperatures. (G) Bar plot showing the log2 fold change in abundance for all cell types in 34°C embryos with severe phenotypes relative to those that were phenotypically normal