Anna Reetta Salonen

Soil organic carbon (SOC) sustains many of the processes on which agriculture depends, including nutrient cycling, soil structure formation, and water regulation. It also represents one of the largest actively cycling carbon pools on Earth. Understanding how SOC is stored in agricultural soil and what determines its persistence is therefore important both for maintaining crop productivity and for assessing whether soils can contribute to climate change mitigation.

A widely adopted framework divides soil organic matter (SOM) into two pools with contrasting properties. Particulate organic matter (POM) consists mainly of relatively fresh plant residues with a residence time from years to decades. Mineral-associated organic matter (MAOM) comprises organic compounds associated with soil mineral surfaces and is assumed to persist for decades to centuries. Despite growing interest in this framework, SOC stocks and their distribution into POM and MAOM remain poorly characterised in northern agricultural soils, where short growing seasons and cold winters create conditions for SOC cycling that differ fundamentally from the systems where most research has been conducted. Moreover, most knowledge of what controls MAOM storage, including the concept of soil carbon saturation and fractionation protocols, was derived in older, more weathered soils. Finnish agricultural soils, one of the main focuses of this thesis, are relatively young, having been formed around 10 000 years ago from material deposited by glaciers. Where chemical weathering in older soils has transformed primary minerals into more reactive secondary minerals with larger surface areas, Finnish soils have not had sufficient time to undergo this transformation. Instead, the glacial action physically ground primary minerals into clay-sized minerals, with a substantial proportion of the clay fraction remaining composed of these primary minerals with a limited reactive surface area. This raises the question of whether the concepts and methods developed elsewhere are also applicable to these soils.

In this thesis, I examined SOC in Finnish agricultural soils with three main objectives: (1) to characterise carbon stocks and their distribution into POM and MAOM across soil depths and management systems, (2) to assess which soil properties control MAOM storage and whether SOC saturation concepts apply, and (3) to evaluate whether the fractionation methods commonly used to separate POM and MAOM produce fractions that are consistent with the conceptual definitions of these pools. The first two objectives were addressed using Finnish soils (Chapters 2 and 3), while the methodological evaluation (Chapters 4 and 5) drew on a wider set of soils to test assumptions that are relevant across soil types.

To characterise SOC stocks across the soil profile, I studied soils to 70 cm depth in a 24-year field experiment in South-West Finland (Chapter 2). The experiment compared organic and conventional cropping systems with similar crop rotations but differing fertilisation, alongside an unmanaged meadow as reference. MAOM constituted the largest carbon pool throughout, accounting for 83% (topsoil) to 97% (subsoil) of total SOC. Aluminium and iron oxides were positively associated with MAOM across the soil profile, whereas clay content correlated with MAOM only in the topsoil (0–20 cm). While the topsoils of the cropping systems remained below their predicted MAOM saturation capacity, the meadow topsoil exceeded it, suggesting that texture-based models underestimate the carbon storage potential of these soils. In the 10–30 cm layer, the organic system stored significantly more SOC than the conventional system, possibly due to differences in organic matter input. Subsoil SOC stocks below 40 cm were comparable across all three systems despite very different management histories and, in the case of the meadow, higher root biomass at all depths.

Broadening the scope to 93 farms across southern Finland (Chapter 3), I found that MAOM accounted for on average 86% of total SOC in topsoils (0–20 cm), a proportion that remained stable across clay contents from 2% to 68%. Clay content explained only 9% of the variation in MAOM, whereas aluminium and iron oxides were better predictors (21–22% each) but still left most variation unexplained. SOC saturation was assessed with Hassink's equation, clay to SOC ratios, and boundary line analyses, and yielded contradictory results: some soils appeared oversaturated by one metric and undersaturated by another.

Chapters 4 and 5 address whether the two widely used approaches to separate POM and MAOM produce fractions consistent with their conceptual definitions. I compared size-based fractionation (53 μm cut-off) with density-based (1.6 g cm-3 cut-off) fractionation using Rock-Eval thermal analysis. Density fractionation isolated POM and MAOM with clearly different thermal signatures, consistent with the conceptual framework. Size-fractionated POM, by contrast, had properties resembling MAOM rather than plant material, suggesting it contained substantial mineral-associated carbon (Chapter 4). Chapter 5 examined a likely contributing factor: incomplete dispersion of soil aggregates during size fractionation. Across 38 soils, chemical dispersion frequently failed to fully break down aggregates, leaving MAOM in the POM fraction. On average, this led to a 33% overestimation of SOC in POM and an 8% underestimation of SOC in MAOM, with the degree of under-dispersion increasing with SOC content and aluminium and iron oxide concentrations.

Dividing SOC into POM and MAOM may help to better understand SOC dynamics, but this requires fractionation methods that produce distinct pools in line with the conceptual framework. Based on this thesis, density fractionation seems a better tool than size fractionation for better understanding SOC cycling in future studies, despite operational challenges. This thesis furthermore illustrates that models and methods developed in older, more weathered soils do not necessarily transfer to soils with different mineralogy and weathering history. SOC saturation concepts based on soil texture did not perform consistently in the Finnish soils studied here, and local mineralogy may be more relevant than texture for understanding what controls MAOM storage. This highlights that conceptual frameworks, models and methods require validation across soils with contrasting properties before they can be applied as universally valid.