Extended Data Fig. 2: Modeled relative photorespiration rate (V_O/V_C) vs. monthly daytime growing temperature, grouped by [CO₂]_atm bin.

Daytime temperatures were predicted from a downscaling of a time-evolving earth system model (see Methods) for every month in a typical year for every sample at the time of growth. Relative photorespiration rates predicted from [CO₂]_atm and daytime temperature along with other required climate variables (see Supplementary Information Fig. 2) using P-model v.1.0⁵¹. Lines are 2^nd-order polynomial fits to the data and associated ± 95% prediction intervals where data are binned by [CO₂]. The strong correlations (r² = 0.98 for all bins) justify the use of [CO₂] bins to explain our data and suggest that other P-model inputs that vary naturally between sample sites (for example, vapor pressure deficit, light intensity, soil moisture) do not meaningfully impact relative photorespiration rates.