Surface soil carbon (C) in the form of organic matter is crucial for various ecosystem services, including climate regulation, plant production, nutrient cycling, and water storage. Global data reveals higher soil C content in colder and mesic ecosystems than in warmer and xeric ones. Soil C typically accumulates during early ecosystem development and decreases during later stages. The interaction between soil organic matter and minerals largely governs soil C sequestration. The long-term preservation of soil C depends on its occlusion within aggregates and formation of organo-mineral associations, reducing microbial decomposition. Mineral-associated C fractions are generally considered less vulnerable to microbial decomposition and warming-induced respiration increases than unprotected carbon fractions. Previous research has emphasized the mineral control of soil C accumulation and loss across large temporal and spatial gradients, often focusing on individual chronosequences. However, a comprehensive understanding of how the mineral protection of C changes during soil development across different ecosystems remains limited. This study addresses this gap by exploring the major drivers of soil organic C fractionation during soil development across 16 globally distributed soil chronosequences from contrasting ecosystems, aiming to determine how and why the amount and proportion of surface (0–10 cm) soil C stored in different organic matter fractions change during soil development across biomes. The selected chronosequences encompass diverse parent materials (volcanic material, sedimentary rocks, sand dunes) and ecosystems (from deserts to tropical forests and croplands), each with four to six stages ranging from hundreds to millions of years, sharing similar parent material and climate. A density-based soil organic matter fractionation method separated and quantified soil organic C content into three fractions: free light (unprotected), occluded light (protected by occlusion), and mineral-associated (protected by sorption). The focus was on surface soil (top 10 cm) due to its high biological activity and exposure to environmental factors. This study builds upon previous single-chronosequence studies by offering a broader, global perspective on the factors influencing soil carbon dynamics.

Existing literature highlights the importance of soil organic carbon (SOC) in regulating various ecosystem services. Studies have established the relationship between SOC and climate, showing that cold and mesic ecosystems generally store more SOC than warm and xeric ones. The role of soil age in SOC accumulation has also been studied, with findings suggesting a hump-shaped relationship, where SOC initially accumulates and subsequently declines over millennia. The mechanisms of SOC stabilization and destabilization, largely governed by interactions between organic matter and minerals, have been extensively investigated. The concept of mineral protection of SOC, where minerals enhance the stability of SOC by reducing its accessibility to microbial decomposition, has been a central focus. Studies on single chronosequences have provided valuable insights into the relationship between SOC and mineral reactivity, revealing the influence of mineral weathering on SOC storage and turnover. However, much of this research has been conducted on individual chronosequences, making it challenging to generalize findings across diverse ecosystems. The lack of a comprehensive understanding of how mineral protection of C changes over centuries to millennia across different ecosystems necessitates this study which aims to bridge this knowledge gap.

This research utilized sixteen globally distributed long-term soil chronosequences from six continents, encompassing diverse parent materials, climates, vegetation types, and age gradients. Each chronosequence included four to ten sites representing increasing temporal stages of ecosystem development. Climatic types were classified using the Koppen-Geiger system. Mean annual temperature and precipitation data were obtained from the WorldClim database. Vegetation surveys were conducted in 50 x 50 m plots using the line-intercept method to determine total plant cover. The Normalized Difference Vegetation Index (NDVI) from MODIS satellite data (2008-2017) served as a proxy for net primary productivity (NPP). Soil samples (0-10 cm depth) were collected from each site, excluding litter. Bulk density was calculated, and samples were sieved (2 mm). Total organic C was analyzed by colorimetry after oxidation with K2Cr2O7 and H2SO4. A density-based fractionation method, slightly modified from existing protocols, separated soil organic C into three fractions: free light, occluded light, and mineral-associated. The proportion of C in each fraction was calculated relative to total organic C. Linear mixed-effects models assessed the effects of ecosystem, productivity, soil age, and their interactions on C fractions, accounting for spatial autocorrelation. Structural equation modeling (SEM) investigated environmental drivers of mineral protection of surface soil C, including NPP (NDVI), soil texture, mean annual temperature (MAT), mean annual precipitation (MAP), and spatial distance. Soil basal heterotrophic respiration was measured to assess soil biological activity. The temperature sensitivity of soil respiration (Q10) was determined for a subset of chronosequences to evaluate the vulnerability of different C fractions to warming. Statistical analyses were performed using IBM SPSS, R, and various R packages, including ‘ggplot2’, ‘vegan’, ‘ggstatsplot’, ‘lme4’, ‘lmerTest’, ‘patchwork’, ‘readxl’, and ‘agricolae.

Total surface soil C stocks were generally highest in grasslands and forests, particularly in temperate and tropical regions, exhibiting a hump-shaped relationship with soil age. The free light and mineral-associated C fractions consistently dominated soil C stocks, with the occluded light fraction comprising a minor portion. Analysis focused on the proportion of C in each fraction rather than absolute stocks due to variations in soil bulk density. Linear mixed-effects modeling revealed a significant influence of ecosystem productivity on the proportion of organic C in the free light and mineral-associated fractions. However, soil age class did not significantly affect these proportions. Warm and wet tropical and temperate ecosystems (often forests) consistently had a larger proportion of free light C compared to mineral-associated C, irrespective of soil age. These ecosystems also showed relatively high soil organic C content. Conversely, arid and cold ecosystems with low primary production showed a predominance of mineral-associated C. The proportion of surface soil C in the free light fraction positively correlated with microbial respiration rates, while the opposite was true for the mineral-associated fraction. An experiment across a subset of chronosequences showed a positive relationship between the proportion of free light C and the temperature sensitivity of soil respiration (Q10). Structural equation modeling indicated that contemporary ecosystem productivity (NDVI) and soil fine texture were the primary drivers of differences in C fraction proportions. Higher precipitation, temperature, and forest vegetation led to increased productivity, with soil C in productive ecosystems predominantly stored in the free light fraction. In contrast, less productive environments showed higher mineral-associated C, positively linked to fine texture and indirectly to soil age. Exceptions to the general pattern were observed in specific chronosequences. For example, a highly productive continental forest showed a predominance of free light organic matter, while a subtropical cropland chronosequence had lower C content and mineral-associated C dominance.

The findings highlight the importance of ecosystem productivity in determining the proportion of protected versus unprotected soil carbon. The dominance of free light C in productive ecosystems underscores their vulnerability to changes in productivity and warming, potentially leading to substantial C losses. The greater stability of mineral-associated C in less productive environments offers some resilience to warming impacts. These results have implications for climate change modeling and land management practices. Preserving ecosystem productivity, particularly in tropical and temperate forests, is essential for maintaining soil C storage. Further research is needed to fully integrate these findings into predictive models, accounting for spatial variability and the complexity of interactions between soil biogeochemistry and climate change.

This study demonstrates that ecosystem type, specifically productivity, is a more significant driver of surface soil carbon fraction proportions than soil age. Productive ecosystems store more unprotected carbon, vulnerable to warming and reduced productivity, while less productive ecosystems have more stable, mineral-associated carbon. Conserving ecosystem productivity is crucial for maintaining soil carbon storage and mitigating climate change impacts. Future research should focus on extending these findings to deeper soil layers and integrating them into more comprehensive models of soil carbon dynamics.

The study's findings are limited to the surface soil layer (0-10 cm) and may not fully represent total soil C stocks. While established chronosequences were used, the impact of regressive processes like erosion cannot be entirely ruled out. Broad climatic and ecosystem drivers may mask site-specific soil geochemistry effects. The interaction between soil depth and soil age in influencing soil C stabilization was not explicitly addressed, representing a potential area for future investigation. Extrapolation of results to global changes in soil C storage requires caution.