Modifying the stomatal conductance model of Camellia oleifera in the southern hilly region of China

This study aims to determine the optimal response model of stomatal conductance to the Camellia oleifera leaves in hilly areas of south China. A portable photosynthetic apparatus (LCI-SD) was used to observe the leaves of three healthy Camellia oleifera trees in four directions of southeast, northwest every 10 days from June to September in 2020 and 2021. The field test was performed at Hunan National Camellia oleifera Engineering Technology Research Center in Changsha City, Hunan Province, China. A miniature automatic meteorological observation U30 Station (HOBO ware) was utilized to record synchronously the stomatal conductance, net photosynthetic rate, and CO2 concentration in the leaves, together with the meteorological data, such as atmospheric temperature, relative humidity, and photosynthetically active radiation. Then, the stomatal conductance of leaf was simulated using the Jarvis models with nine combinations, the Ball-Woodrow-Berry (BWB) and Ball-Berry-Leuning (BBL) model with two CO2 concentrations of leaf and intercellular. The Jarvis-8 and BBL model were modified to compare the introduced temperature of camellia oil leaves, and the CO2 concentration inside and outside stomata. A nonlinear regression and least square method were used to determine each parameter by the SPSS25.0 software platform. The results show that the influence of soil water or leaf water potential could be ignored when using the Jarvis model in the hilly region of southern China. It infers that the Jarvis model was dependent mainly on the photosynthetically active radiation, saturation vapor pressure deficit and temperature. Therefore, the soil water or leaf water potential was be ignored, when the Jarvis model was used in the hilly region of southern China. A better simulation effect was achieved in the Jarvis-8, BWB, and BBL model calculated by the leaf CO2 concentration. Among them, the BBL model presented the largest determination coefficient (R2=0.763) and the smallest absolute mean error (MAE=0.015), indicating the best effect. By contrast, the BWB model presented the smallest determination coefficient (R2=0.599), and the maximum mean absolute error (MAE=0.020), indicating the worst simulation effect. After two parameters were introduced, the simulation effect of leaf temperature difference was improved in the Jarvis-8 and BBL model, but not significantly. The CO2 concentration difference between inside and outside the stomatal was significantly improved the accuracy of the BBL model, where the determination coefficient increased from 0.763 to 0.950, and the slope of the model was very close to 1 (1.004), indicating the better consistence between the simulation and the measured stomatal conductance value. But, there was no change of the Jarvis-8 model in this case. A better simulation was achieved for the stomatal conductance values of leaves in the key growth periods of Camellia oleifera in 2020 (R2=0.92) and 2021 (R2=0.95), and the variation values of stomatal conductance degree-days under different scenarios. On the whole, the simulation performance was ranked in the descending order of the BBL-C model, BBL-Tmodel, BBL model, Jarvis-T model, Jarvis model, and Jarvis-C model. Therefore, the stomatal conductance response model of Camellia oleifera leaves can be recommended to introduce the BBL model with the CO2 concentration difference between stomatal. These findings can provide a strong reference to select a suitable stomatal conductance model for the large-scale Camellia oleifera in the hilly region of southern China.