The maximum carboxylation rate (Vcmax) is a key parameter in determining the plant photosynthesis rate per unit leaf area. However, most terrestrial biosphere models currently treat Vcmax as constants changing only with plant functional types, leading to large uncertainties in modelled carbon fluxes. Vcmax is tightly linked with Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco). Here we investigated the relationship between leaf chlorophyll content and Rubisco (Chl-Rub) within a winter wheat paddock. With chlorophyll as a proxy of Rubisco, a semi-mechanistic model was developed to model Vcmax_25. The Chl-Rub relationship was validated using measurements in a temperate forest in Canada. The results showed that Rubisco was strongly correlated with chlorophyll (R² = 0.96, p < 0.001) for winter wheat, due to the absorption of light energy by chlorophyll and the amount of CO₂ catalysed by Rubisco are tightly coupled. Incorporating the Chl-Rub relationship into the semi-mechanistic model, the root mean square error of modelled Vcmax₂₅ was the lowest among all estimation models. The slopes of Chl-Rub relations were almost identical for the winter wheat and temperate forest, demonstrating the potential for using leaf chlorophyll content as a proxy of leaf Rubisco in modelled Vcmax₂₅ at large spatial scales. We anticipate that improving Vcmax₂₅ estimates over time and space will reduce uncertainties in terrestrial biosphere models and improve the estimates of global carbon budgets.

Location----Field campaigns were conducted at a farmland ecosystem station in Shangqiu, Henan Province, China (34.27°N, 115.4°E).

Samples----Unfolded leaves at the top of the canopy were randomly selected for destructive sampling approximately once every 10 days, from a 30×30 m subset area within the larger field. On each sampling date, mean values of leaf chlorophyll, nitrogen, and Rubisco were calculated from 3 samples.

Leaf chlorophyll and nitrogen----Leaf chlorophyll and nitrogen content was measured from the same leaves due to the small sampling mass required for laboratory analysis. The leaves sampled for chlorophyll measurements were packaged in plastic sample bags and transported into an ice box immediately after collection. About half an hour later, these samples were sent to the laboratory for leaf chlorophyll measurement. In the laboratory, leaf chlorophyll was extracted into a solution with absolute ethyl alcohol and formaldehyde mixed in a volume ratio of 1:1. The absorbance of the dissolving solution was measured at 663.8, 646.8 and 480 nm using a Shimadzu UV-1700 spectrophotometer (Wellburn, 1994). The remaining samples for measuring nitrogen were dried in an oven at 80 ℃ for approximately 48 hours until a constant weight and then determined total leaf nitrogen by the Kjeldahl method (Wolf, 1982). 

Rubisco----A whole leaf was needed to measure the leaf Rubisco content, as the analysis requires a sample exceeding 0.5g fresh weight. A portable chlorophyll meter SPAD (Konica Minolta Inc., Osaka, Japan) was used to observe the SPAD values of leaves, for which Rubisco were measured. From tial to tip of one leaf, 5 SPAD values were measured evenly and averaged for representing the SPAD value of the whole leaf. SPAD values were converted into leaf chlorophyll content using an empirical model fitted using measured SPAD values and laboratory analysed chlorophyll content in the same field. The Rubisco content of leaves was determined by the method of Makino et al. (1985, 1986).

Vcmax25-----Carbon dioxide response curves (A/Ci) were measured using an LI-6800 portable infrared gas analyser (LI-COR, Lincoln, NE, USA). To measure more leaves on one day, 3 leaves were measured using the A/Ci response curve method, and additional 3 measurements were conducted using the rapid response curve method (Stinziano et al., 2017). For the A/Ci response curve, PPFD was set at 1800 μmol m^-2 s^-1, and the stepwise CO₂ concentrations were 400, 200, 100, 50, 400, 600, 800, 1000, 1200, 1500, and 1800 ppm, respectively. The function fitaci in ‘plantecophys’ R package was used to estimate Vcmax₂₅ for both the A/Ci response curve and fast response curve methods.