The dynamics of soil organic carbon (SOC) play a critical role in the global carbon cycle. In context of global warming, numerous experimental studies have reported temperature-sensitive responses of SOC. However, the limited temporal frequency and spatial density of repeated sampling and whole-profile SOC observations have hindered the understanding of large-scale, long-term spatiotemporal patterns of SOC and their responses to environment changes under global warming, thereby constraining the ability to accurately predict the global carbon cycle.
Motivated by these challenges, Professor Lin Yang, Academician Chenghu Zhou and Dr. Feixue Shen from Nanjing University, in collaboration with Professor A-Xing Zhu from University of Wisconsin-Madison, Academician Shilong Piao from Peking University, and Professor Yiqi Luo from Cornell University, analyzed 10,639 soil profiles from forests and croplands across the contiguous United States to reconstruct the dynamics of soil organic carbon (SOC) over a 45-year period (1970–2014) in both topsoil (0–30 cm) and subsoil (30–100 cm). They further examined the temporal relationships between SOC changes at different depths and key environmental drivers such as climate, vegetation, soil properties, and nitrogen deposition. This study is the first to reveal large‑scale, long-term connections between temperature fluctuations and SOC dynamics across diverse ecosystems, as well as the depth-dependent nature of environmental controls. The findings provide a critical scientific basis for improving Earth system models and refining carbon management strategies in the context of global warming.
Highlights:
Two-stage SOC changes in the CONUS from 1970 to 2014 were strongly associated with changes in warming rates.
Rising temperatures predominantly coincided with reduced topsoil SOC stock.
Soil water content emerged as the strongest negative relationship with subsoil SOC dynamics.
The loss of SOC induced by warming may exhibit a threshold effect controlled by the rate of temperature increase.

Core content:
This study integrated SOC measurements, time-series statistical analysis, and machine learning techniques, revealing a two-phase trajectory of SOC change in forest and cropland ecosystems across the contiguous United States. From the 1970s to the 1990s, SOC showed a non-linear loss trend. Since the 1990s, however, SOC loss has stagnated. Furthermore, starting from the late 1990s, while SOC changes in the cropland 0–30 cm layer remained stagnant, both the cropland 30–100 cm layer and the entire soil profile in forest systems began to show an increasing trend.
The study found that the two-phase SOC change patterns in both forest and cropland ecosystems are closely related to the rate of warming. Overall, the rate of temperature change was negatively correlated with the rate of SOC change. Notably, SOC loss commenced only when the rate of temperature increase exceeded a certain threshold.
Partial correlation analysis indicated that air temperature is the primary factor driving SOC loss in surface layers, showing a significant negative correlation in both forests and croplands. In contrast, changes in subsurface SOC were mainly dominated by soil water content, also showing a negative correlation. Other environmental factors—such as increased net primary productivity (NPP) in forest surface layers and nitrogen fertilizer application in croplands—also demonstrated the potential to enhance carbon storage through targeted management.
Using machine learning, this paper mapped SOC stock changes at a ~2 km resolution. Nationwide, SOC stocks in the top 1 meter increased by 1.41% (from 19.05 Pg to 19.32 Pg) in forests and 1.14% (from 13.10 Pg to 13.25 Pg) in croplands.
Outlook: The study highlights a strong connection between the rate of warming and changes in SOC stocks from 1970 to 2014. Given the recorded temperature increases since 2015, the recently accelerated warming may reverse the current trend of increasing SOC stocks in forest and cropland soils. Therefore, extending monitoring beyond 2014 is crucial to validate the observed change patterns. Concurrently, mechanistic research on how warming rates drive SOC changes across different ecosystems and soil depths should be pursued, with particular attention to deep soil moisture dynamics. These insights are essential for improving the parameterization of SOC responses to temperature and precipitation changes in Earth system models.
DOI: 10.1016/j.geosus.2025.100359