LONG-TERM SOIL CARBON FEEDBACKS TO CLIMATE CHANGE AND LAND MANAGEMENT

Abstract

Soil serves as a major carbon (C) reservoir within terrestrial ecosystems, playing a crucial role in global C cycling. Since the Industrial Era, rising atmospheric CO2 levels—primarily driven by fossil fuel combustion and land-use change—have contributed to global warming and altered precipitation patterns. Elevated CO2 and warming also enhance plant photosynthesis, potentially increasing plant productivity and promoting greater C sequestration in soils, which may counteract atmospheric CO2 accumulation. Conversely, climate and land-use changes can accelerate microbial activity, leading to increased soil CO2 emissions. As a result, the net effect of soil feedback on atmospheric CO2 remains uncertain, depending on the magnitude and persistence of these opposing processes. This uncertainty arises from the complex interactions among plants, soils, and microbial communities, which exhibit dynamic and interdependent responses to climate conditions and soil properties over time. While numerous studies have explored these interactions, relatively few have addressed the predictability of their temporal patterns. Incorporating microbial processes into ecosystem models has been recognized as a critical step in improving soil C cycle simulations. However, parameterizing these increasingly complex models remains a challenge. Long-term field experiments, which capture temporal patterns and provide diverse measurements, offer a valuable benchmark for calibrating microbial-ecological models. Additionally, experimental manipulations simulating climate and land-use changes provide insights into how environmental factors influence model parameters—parameters often assumed to be constant in global and century scale. Advancements in DNA-based analyses and isotope tracing have uncovered novel mechanisms that were previously difficult to integrate into microbial models. However, it remains unclear whether incorporating this new information can enhance model parameterization and improve extrapolation. Therefore, this dissertation focuses on understanding and predicting soil C feedback to climate change and land management practices while evaluating the role of long-term field data in improving model predictions.First, using a 20-year field experiment involving warming and biomass harvesting, we investigated and modeled the treatments effects on soil respiration over time. Soil heterotrophic respiration (Rh) is a fundamental driver of terrestrial carbon cycling, yet its long-term responses to climate warming and grassland management, such as biomass harvesting, remain uncertain. While warming is expected to enhance Rh, uncertainties persist regarding its temporal trajectory and the interactive influence of harvesting. Using a 20-year field experiment in a temperate grassland, we show that warming persistently increased Rh, with its effects nearly doubling in magnitude over the second decade. This intensified response was accompanied by a stronger relationship between microbial functional genes and Rh. In contrast, harvesting moderated this increasing trend, likely due to reductions in soil moisture, plant productivity, microbial biomass, and functional gene abundance. Integrating the empirical findings into a microbial-ecological model, we demonstrated that decadal-scale Rh variation can be predicted when accounting for complex biotic and climate interactions. However, long-term projections indicate that both the amplified warming effects and harvesting’s mitigation influence on Rh may decline over a century. Our study reveals the evolving nature of warming-harvesting interactions on Rh, and emphasizes the importance of understanding and predicting decadal carbon dynamics trajectory in sustainable grassland management. While the first study primarily revealed C exchange between the soil and atmosphere at the ecosystem scale, understanding and predicting the internal fluxes among plants, soils, and microbes that drive C dynamics remain more challenging. Ecosystem C dynamics, driven by the balance between C assimilation and decomposition, remain highly uncertain under altered precipitation levels. While field 13CO2 isotope pulse labeling has provided insights into C allocation and turnover of newly assimilated C, the efforts of using the information in predicting ecosystem soil C responses remains limited. To address the limitations, we integrate a three-year precipitation manipulation experiment with 13CO2 pulse labeling and microbial-ecological modeling to assess ecosystem C dynamics under reduced precipitation. We find that half precipitation increases shoot biomass while reducing 13C enrichment in plant biomass, suggesting shifts in carbon allocation. In contrast, 13C enrichment in the free-light fraction of soil organic matter increases, indicating a slowdown in turnover of newly fixed C. Incorporating isotope data into microbial-ecological models significantly improves predictions of responses in ecosystem respiration, plant biomass, and soil 13C dynamics, reducing both parameter and simulation uncertainty. Model simulations informed by isotope data reveal that reduced precipitation suppresses soil C turnover and microbial respiration. These results underscore the importance of internal C cycling processes in shaping ecosystem responses to climate change and highlight the power of isotope-informed modeling for improving predictions of terrestrial C dynamics. Changes in soil C fluxes, as shown in the first two studies, will ultimately influence soil C stock, which determines the overall soil–atmosphere C exchange balance. The third study focuses on deep soil C changes, which have been rarely explored in previous field and modeling studies, in response to climate warming and harvesting. Deep soils are considered favorable for long-term C storage due to their biogeological constraints on organic matter decomposition. However, the stability of deep soil C under warming and land management, such as biomass harvesting, remains uncertain. We present findings from a decade-long temperate grassland experiment with +3 °C warming and annual plant biomass harvesting to assess the responses of deep soil C, N, plant and microbial communities. Results showed combined warming and harvesting reduced soil C and N stock more than either treatment alone, with stronger effects observed below 0.2 m. Harvesting enhanced plant shoot and root biomass, while both warming and harvesting increased the soil respiration and its autotrophic component. Microbial community shifts indicative of altered C acquisition strategies were also observed, with warming reducing bacterial species positively associated with soil C, while harvesting increased species negatively associated with soil C, particularly in deeper soils. A vertically resolved biogeochemical model calibrated with the field data accurately simulated soil C, N loss, and respiration dynamics, suggesting that deep soil C and N losses were driven by reduced litter input and enhanced microbial activity under combined warming and harvesting. Our findings highlight the vulnerability of deep soil C storage to warming and land management, with important implications for managing grassland C storage under changing climate and rising agriculture product demand. Overall, this work deepens our understanding of the effects of climate change and grassland management on soil C fluxes, internal C allocation, and C stock in both surface and deep soils, capturing dynamics from monthly to decadal scales. We also explored the drivers of soil C changes, considering the influence of climate, biotic factors, and soil properties and their responses to environmental changes. In particular, we highlighted the microbial role in mediating soil C by linking the ecological and functional traits of microbes to soil C changes. By integrating observations with microbial-explicit ecosystem models, we improved model performance and better constrained internal fluxes, underscoring the feasibility and importance of incorporating microbial responses into soil C models. Building on empirical findings and model applications, our work lays a foundation for extending the observed environmental change effects to larger spatial scales and longer timeframes, while considering the significant complexity and variability of soil C responses to future climate change and land management.