Abstract. Soil dust is a major driver of ice nucleation in clouds leading to precipitation. It consists largely of mineral particles with a small fraction of organic matter constituted mainly of remains of micro-organisms that participated in degrading plant debris before their own decay. Some micro-organisms have been shown to be much better ice nuclei than the most efficient soil mineral. Yet, current aerosol schemes in global climate models do not consider a difference between soil dust and mineral dust in terms of ice nucleation activity. Here, we show that particles from the clay and silt size fraction of four different soils naturally associated with 0.7 to 11.8 % organic carbon (w/w) can have up to four orders of magnitude more ice nucleation sites per unit mass active in the immersion freezing mode at −12 • C than montmorillonite, the nucleation properties of which are often used to represent those of mineral dusts in modelling studies. Most of this activity was lost after heat treatment. Removal of biological residues reduced ice nucleation activity to, or below that of montmorillonite. Desert soils, inherently low in organic content, are a large natural source of dust in the atmosphere. In contrast, agricultural land use is concentrated on fertile soils with much larger organic matter contents than found in deserts. It is currently estimated that the contribution of agricultural soils to the global dust burden is less than 20 %. Yet, these disturbed soils can contribute ice nuclei to the atmosphere of a very different and much more potent kind than mineral dusts.

The interaction between soil organic carbon pools and climate change is an important determinant of future atmospheric CO2 concentrations. Much effort has so far been allocated to manipulative process studies and predictive modelling exercises. Here, we examine the potential for directly detecting predicted changes through repeated soil sampling. Two contrasting benchmark plots were selected in the steppe at the Russian–Mongolian border, where soil organic carbon losses are predicted to be around 10% over the first 50 years of climate change. In both plots, 50 samples were taken to 20 and 30 cm depths. The estimated time intervals before re‐sampling by the same method that were likely to prove significant soil organic carbon losses (α=0.05; statistical power=0.90) were 43 and 26 years.

Soil dust is a major driver of ice nucleation in clouds leading to precipitation. It consists largely of mineral particles with a small fraction of organic matter constituted mainly of remains of micro-organisms that participated in degrading plant debris before their own decay. Some micro-organisms have been shown to be much better ice nuclei than the most efficient soil mineral. Yet, current aerosol schemes in global climate models do not consider a difference between soil dust and mineral dust in terms of ice nucleation activity. Here, we show that particles from the clay and silt size fraction of four different soils naturally associated with 0.7 to 11.8 % organic carbon (w/w) can have up to four orders of magnitude more ice nuclei per unit mass active in the immersion freezing mode at −12 °C than montmorillonite, the most efficient pure clay mineral. Most of this activity was lost after heat treatment. Removal of biological residues reduced ice nucleation activity to, or below that of montmorillonite. Desert soils, inherently low in organic content, are a large natural source of dust in the atmosphere. In contrast, agricultural land use is concentrated on fertile soils with much larger organic matter contents than found in deserts. It is currently estimated that the contribution of agricultural soils to the global dust burden is less than 20 %. Yet, these disturbed soils can contribute ice nuclei to the atmosphere of a very different and much more potent kind than mineral dusts

Retreating glaciers successively expose mineral substrate which is colonised within a few years by soil biota, lichens and plants. Two such successions which cover the first five decades after exposure were studied in the Swiss Alps. Sites differed in parent material and precipitation. At each site, five strata of different age were dated by distance from the current glacier gate and records of annual retreat. The proportion of fine material (< 63 mm) differed by a factor of 2 between sites and a factor of 3 within sites. From the youngest to the oldest stratum, plant cover increased from < 5 to > 85% and organic carbon content (C org ) in the bulk material increased by 6 times at the wetter site and by 16 times at the drier site. Even so, partitioning of C org between particle size fractions 2000 to 63 mm and < 63 mm was the same at both sites (6 : 4), and so was the proportion of microbial C (C mic ) in C org (5%). Coefficients of variation across all 10 strata studied were only 22% for C org partitioning and 21% for the proportion of C mic in C org . Soil carbon models frequently apply prescribed partitioning factors in simulating organic matter input into soil. Our findings lend support to such an approach.

The δ(15)N natural abundance values, when related to the C:N ratios, may readily indicate perturbation of soil N cycling prior to other, more demanding investigations into related processes and extend the current use of IRMS in ecosystem research.