Leaf CO2 response curve fitting method matters: cross-tool assumptions spread photosynthetic parameters and bias leaf-canopy photosynthesis.

An-Ci curves are used to infer Vcmax,25 and J25 for terrestrial biosphere models, but fitting-tool assumptions can alter both parameters; although Vcmax,25 varies more, J25 differences can also shift inferred limitation regimes and thereby affect propagation from leaf to canopy simulations. We compared 11 widely used An-Ci fitting configurations and one optimal model configuration across four C3 crops (soybean, perennial ryegrass, red clover, and winter wheat), ran two complementary sensitivities (one-at-a-time toggles relative to a baseline and variance-based decomposition), and used the resulting parameters in the Soil-Canopy Observation, Photochemistry and Energy (SCOPE) model to simulate An and gross primary productivity (GPP). Vcmax,25 differed by up to fourfold across tools, with consistent fingerprints across crops. Pooled sensitivities identified mesophyll conductance (gm) and limiting-rate selection as the dominant drivers of between-tool spread. In SCOPE, using the same assumptions in simulation as in parameter estimation kept leaf An errors below 9%, while mismatched assumptions produced An deviations -68% to +81% across crops and soybean GPP shifts -12% to +20%. Vcmax,25 is transportable only with its assumptions. Users should align gm, limiting-rate, and kinetic settings with intended simulations or treat catalogued values as priors unless recomputed from raw An-Ci data.