Plants and microbes are the primary drivers in affecting the formation and accrual of soil organic carbon (SOC) for natural ecosystems. However, experimental evidence elucidating their underlying mechanisms for SOC accumulation remains elusive. Here, we quantified plant and microbial contributions to SOC accrual in successional subtropical forests by measuring leaf-, root-, and microbial biomarkers, root and leaf litter inputs, and microbial C decomposition. The long-term monitoring results showed that SOC accumulated rapidly at the early-successional stage, but changed little at the mid- and late-successional stages. SOC accrual rate was positively correlated with fine-root production and microbial C turnover, but negatively with annual litterfall. Biomarker data exhibited that the rapid SOC accumulation was jointly driven by root- and microbe-derived C inputs from the early- to mid-successional stages. In contrast, aboveground litterfall considerably contributed to soil C accrual from the mid- to late-successional stages compared to belowground processes, although SOC accumulation is low. Our study revealed the importance of root production and microbial anabolism in SOC accrual at the early stages of forest succession. Incorporating these effects of belowground C inputs on SOC formation and accumulation into earth system models might improve model performance and projection of long-term soil C dynamics.

As urban expansion continues, the intensifying land surface temperature (LST) underscores the critical need for accurate predictions of future thermal environments. However, no study has investigated which method can most effectively and consistently predict the future LST. To address these gaps, our study employed four methods—the multiple linear regression (MLR), geographically weighted regression (GWR), random forest (RF), and artificial neural network (ANN) approach—to establish relationships between land use/cover and LST. Subsequently, we utilized these relationships established in 2006 to predict the LST for the years 2012 and 2018, validating these predictions against the observed data. Our results indicate that, in terms of fitting performance (R² and RMSE), the methods rank as follows: RF > GWR > ANN > MLR. However, in terms of temporal stability, we observed a significant variation in predictive accuracy, with MLR > GWR > RF > ANN for the years 2012 and 2018. The predictions using MLR indicate that the future LST in 2050, under the SSP2 and SSP5 scenarios, is expected to increase by 1.8 ± 1.4 K and 2.1 ± 1.6 K, respectively, compared to 2018. This study emphasizes the importance of the MLR method in predicting the future LST and provides potential instructions for future heat mitigation.

Understanding the sensitivity of ecosystem respiration (ER) to increasing temperature is crucial to predict how the terrestrial carbon sink responds to a warming climate. The temperature sensitivity of ER may vary on a diurnal basis but is poorly understood due to the paucity of observational sites documenting real ER during daytime at a global scale. Here, we used an improved flux partitioning approach to estimate the apparent temperature sensitivity of ER during the daytime (E_0,day) and nighttime (E_0,night) derived from multiyear observations of 189 FLUXNET sites. Our results demonstrated that E_0,night is significantly higher than E_0,day across all biomes, with significant seasonal variations in the day–night discrepancy in the temperature sensitivity of ER (ΔE₀ = E_0,night/E_0,day) except for evergreen broadleaf forest and savannas. Such seasonal variations in ΔE₀ mainly result from the effect of temperature and the seasonal amplitude of NDVI. We predict that future warming will decrease ΔE₀ due to the reduced E_0,night by the end of the century in most regions. Moreover, we further find that disregarding the ΔE₀ leads to an overestimation of annual ER by 10~80% globally. Thus, our study highlights that the divergent temperature dependencies between day- and nighttime ER should be incorporated into Earth system models to improve predictions of carbon–climate change feedback under future warming scenarios.