Terrestrial vegetation currently absorbs approximately a third of anthropogenic CO 2 emissions, mitigating the rise of atmospheric CO 2 . However, terrestrial net primary production is highly sensitive to atmospheric CO 2 levels and associated climatic changes. In C 3 plants, which dominate terrestrial vegetation, net photosynthesis depends on the ratio between photorespiration and gross photosynthesis. This metabolic flux ratio depends strongly on CO 2 levels, but changes in this ratio over the past CO 2 rise have not been analyzed experimentally. Combining CO 2 manipulation experiments and deuterium NMR, we first establish that the intramolecular deuterium distribution (deuterium isotopomers) of photosynthetic C 3 glucose contains a signal of the photorespiration/photosynthesis ratio. By tracing this isotopomer signal in herbarium samples of natural C 3 vascular plant species, crops, and a Sphagnum moss species, we detect a consistent reduction in the photorespiration/photosynthesis ratio in response to the ∼100-ppm CO 2 increase between ∼1900 and 2013. No difference was detected in the isotopomer trends between beet sugar samples covering the 20th century and CO 2 manipulation experiments, suggesting that photosynthetic metabolism in sugar beet has not acclimated to increasing CO 2 over >100 y. This provides observational evidence that the reduction of the photorespiration/photosynthesis ratio was ca. 25%. The Sphagnum results are consistent with the observed positive correlations between peat accumulation rates and photosynthetic rates over the Northern Hemisphere. Our results establish that isotopomers of plant archives contain metabolic information covering centuries. Our data provide direct quantitative information on the "CO 2 fertilization" effect over decades, thus addressing a major uncertainty in Earth system models.A tmospheric CO 2 levels have increased from ∼200 ppm during the last ice age to currently 400 ppm, and they may, according to pessimistic scenarios, exceed 1,000 ppm in the year 2100 (1). Understanding plant responses to increasing CO 2 is currently hampered by two fundamental limitations: First, it is unknown how well manipulation experiments represent responses to the gradual CO 2 increase over decades and centuries. In Free-Air CO 2 Enrichment (FACE) experiments, which most closely mimic natural conditions, increases in [CO 2 ] generally increase plant growth, but this "CO 2 fertilization" effect often declines after a few years of enrichment (2). Such transient responses may be related to the step increases in [CO 2 ] used in the experiments, their limited duration (2), or factors other than CO 2 becoming limiting (3). Second, in response to the [CO 2 ] increase since industrialization, genetic (4) and phenotypic plant responses (5-7) have been observed. Although century-scale changes have been detected in carbon isotopes (δ 13 C) and attributed to [CO 2 ], these responses are tied to differences in intercellular substrate concentrations that reflect several metabolic fluxes and dif...