Mollisols are of major importance for food security worldwide but are increasingly degraded by soil erosion. Mollisols in Northeast China have been converted into agricultural use only recently, but gullies are widely distributed and gully erosion history, rates and causes remained unclear. We chose a study typical village to estimate initiation years and development rates of the gully systems from 1968 to 2018 by using aerial and satellite imagery. The outlet fan deposits of a large gully system were dated by Caesium-137 (137Cs) and artefacts. To verify the results, we collected information from local farmers. Gully volumes were measured by structure-from-motion technique using photos taken from an unmanned aerial vehicle. Our results showed that gully systems had already appeared on the steep slopes and along unpaved roads in 1968 and had become more complex by 2018 despite terracing and afforestation. Based on gully morphology and 137Cs, gully erosion was estimated to have started in the 1950s to 1960s when the original grassland and forest were completely converted into arable land. From 1968 to 2018, the gully density increased from 1.2 to 2.3 km km-2. The gully heads retreated at speeds from 1.5 to 2.5 m yr-1, and the soil loss from gully erosion ranged from 25.7 to 44.7 Mg yr-1 ha-1. These data demonstrate the severity of gully erosion in study region and underline the importance of appropriate countermeasures, such as maintenance of abandoned terraces under reforested land and better design and construction of roads within the arable land.

Soil organic matter (SOM) is formed through partial decomposition and transformation of plant litter inputs. Thus, litter chemistry is generally regarded as the primary control over the formation efficiency of litter-derived SOM through selective preservation of recalcitrant litter fractions. Here we used model soils and showed that the SOM formation efficiency was, instead, controlled by discriminative protection of litter- and microbially-derived residues by different clay minerals. The SOM formation efficiency was higher for vermiculite than for kaolinite and illite because vermiculite protected more labile litter- and fungal residues through surface adsorption than did kaolinite or illite through pore entrapment. We developed a novel model to quantify mineral-protection strength following litter decomposition, and demonstrated that the mineral-protection strength explained well the variation in the SOM formation efficiency among the model soils, and could be predicted for a natural soil material from those of its compositional clay mineral types and their relative abundances in the soil. Our results provide solid evidence that soil clay mineralogy plays a critical role in SOM formation as known in long-term SOM stabilization, with important implications for model improvement to predict SOM dynamics for different soil types.