Vertical gradients and seasonal variations in the stem CO2 efflux of Larix principis-rupprechtii Mayr

Abstract Stem CO 2 efflux (E S ) plays an essential role in the carbon balance of forest ecosystems. Therefore, it is necessary to study the vertical and seasonal variations in E S in forests with different ages, especially in response to factors associated with temperature, nutrients and wood structure. In this study, we investigated E S and its association with temperature factors using a carbon flux system (Li-8100A) at monthly intervals during the growing season (May to September) and the non-growing season (October) from 2013 to 2015. In addition, we collected data on nutrients and wood structure in August 2016 from 16- (young), 25- (immature), and 41-year-old (mature) Larix principis-rupprechtii Mayr stands in North China. Our analysis showed that the E S values at stem positions near the crown and the root system were generally higher than those in the middle and that the minimum E S values in July occurred at 2.0 m (3.61 μmol m −2 s −1 ), 3.0 m (2.42 μmol m −2 s −1 ), and 5.0 m (4.03 μmol m −2 s −1 ) in the young, immature, and mature forests, respectively. Air temperature and wood temperature influenced the vertical variation in E S , as did stem nitrogen concentration and sapwood width. Compared to the method that uses the vertical E S gradient to determine tree-scale CO 2 effluxes for 16-, 25- and 41-year-old L. principis-rupprechtii forests, the method that uses the E S value at the 1.3-m stem position underestimated the CO 2 effluxes by 5%, 6%, and 24%, respectively. The E S and Q 10 models for each stem position for the three forest ages in this study can be used to accurately estimate tree-scale CO 2 effluxes. These results not only clarified the relationships between the vertical variation in E S and temperature, nutrient content and wood structure, but also revealed the likely response mechanisms of E S to these factors. Methodologically, incorporating the vertical variation in E S and the associated drivers into conventional models would improve the accuracy of annual E S estimates.