Soil carbon sequestration arises from the interplay of carbon input and stabilization, which vary in space and time. Assessing the resulting microscale carbon distribution in an intact pore space, however, has so far eluded methodological accessibility. Here, we explore the role of soil moisture regimes in shaping microscale carbon gradients by a novel mapping protocol for particulate organic matter and carbon in the soil matrix based on a combination of Osmium staining, X-ray computed tomography, and machine learning. With three different soil types we show that the moisture regime governs C losses from particulate organic matter and the microscale carbon redistribution and stabilization patterns in the soil matrix. Carbon depletion around pores (aperture > 10 µm) occurs in a much larger soil volume (19–74%) than carbon enrichment around particulate organic matter (1%). Thus, interacting microscale processes shaped by the moisture regime are a decisive factor for overall soil carbon persistence.

Natural quartz single crystals were experimentally deformed in two orientations: (1) ⊥ to one prism plane and (2) in O⁺ orientation at 900 and 1000°C, 1.0 and 1.5 GPa, and strain rates of ~1 × 10⁻⁶ s⁻¹. In addition, hydrostatic and annealing experiments were performed. The starting material was milky quartz, which consisted of dry quartz with a large number of fluid inclusions of variable size up to several 100 µm. During pressurization fluid inclusions decrepitated producing much smaller fluid inclusions. Deformation on the sample scale is anisotropic due to dislocation glide on selected slip systems and inhomogeneous due to an inhomogeneous distribution of fluid inclusions. Dislocation glide is accompanied by minor dynamic recovery. Strongly deformed regions show a pointed broad absorption band in the ~3400 cm⁻¹ region consisting of a superposition of bands of molecular H₂O and three discrete absorption bands (at 3367, 3400, and 3434 cm⁻¹). In addition, there is a discrete absorption band at 3585 cm⁻¹, which only occurs in deformed regions and reduces or disappears after annealing, so that this band appears to be associated with dislocations. H₂O weakening in inclusion‐bearing natural quartz crystals is assigned to the H₂O‐assisted dislocation generation and multiplication. Processes in these crystals represent recycling of H₂O between fluid inclusions, cracking and crack healing, incorporation of structurally bound H in dislocations, release of H₂O from dislocations during recovery, and dislocation generation at very small fluid inclusions. The H₂O weakening by this process is of disequilibrium nature because it depends on the amount of H₂O available.

The ordering strength of crystallographic preferred orientation data can be assessed in different ways, however, the appropriateness of the methodology used to do so can depend on the geometry of the distributions. Eigenvector-based ordering evaluation methods, for example, may not be appropriate for data that comprise multiple, variably oriented distributions such as those commonly found in quartz c-axis crystallographic preferred orientations. Examination of artificial data that represent a variety of different potential c-axis distributions shows a significant correlation between the relative orientations of those distributions (i.e., the opening angle in a cross-girdled quartz c-axis pole figure) and the strengths calculated using eigen-vector based evaluation methods; larger pole figure opening angles correlate with decreasing distribution ordering strength. The same correlation does not exist when strength is evaluated using the l2 - norm of the estimated probability density function (JPF) of the same data. The direct correlations between pole figure c-axis opening angles and ordering strength noted in the artificial distributions are also demonstrated in the evaluation of real-world data, though significant complications related to heterogeneous nature, and/or deformation, of the natural specimens can partially obfuscate the relationship. Regardless, given the potential effect of geometry on eigenvector-based evaluation methods we recommend that the ordering strength of pole figure data be evaluated using JPF.