Vegetation cover and topography rather than human disturbance control gully density and sediment production on the Chinese Loess Plateau

• The high erosion rates on the Chinese Loess Plateau (CLP) are mainly due to non-topsoil erosion.
• Rates of non-topsoil erosion are strongly related to gully density.
• Gully density is related to river slope, suggesting that river incision also drives gully erosion.
• Humans have strongly accelerated topsoil erosion on the CLP but had far less impact on non-topsoil erosion.
• Therefore humans had a moderate impact on overall sediment production on the CLP.

The factors controlling topsoil erosion rates on the Chinese Loess Plateau (CLP) are currently relatively well understood and topsoil erosion rates can now be relatively accurately estimated. This is, however, not the case for non-topsoil erosion (sediment production by gullying and landslides): while it is well known that these processes produce significant amounts of sediment, the factors controlling their intensity and spatial distribution on the CLP are less well understood.In this study we quantified the contribution of non-topsoil erosion to total sediment production on the CLP and investigated which factors control spatial and temporal variations in non-topsoil erosion. We estimated non-topsoil erosion rates (ENT) by comparing the measured average sediment yields of 46 gauged catchments for the 1950–1970 period, when soil conservation measures were nearly absent in the area, with predicted topsoil erosion rates (using a recently developed empirical model). In addition, gully density was estimated in each catchment using Google Earth data. Our results showed that the area-weighted average catchment erosion rate (E) and ENT were 58.60 ± 51.80 and 48.68 ± 49.78 t ha− 1 yr− 1 respectively for the studied catchments. The sediment contribution of non-topsoil erosion to total sediment production ranged between ca. 0 and 97% with a mean of 70 ± 25%. Both E and ENT were significantly correlated to longitudinal river slope, land use, NDVI, and gully density. However, gully density was the only variable explaining a major part of the variance in both E (60%) and ENT (57%). Gully density itself was significantly related to topography and vegetation cover but not to rainfall erosivity. Importantly, gully density was not only related to overall slope steepness, but also to the longitudinal slope of the river network and the hypsometric integral, suggesting that not only land cover disturbance but also tectonic uplift controls gully density and erosion rates. The absence of a clear climate signal, both with respect to the variation in gully density and in E, can be explained by the overwhelming effect that climate has on vegetation cover.Our research showed that non-topsoil erosion processes are the dominant sediment sources on the CLP and are strongly controlled by natural factors. The effect of human disturbance on non-topsoil erosion processes is far less important than its effect on topsoil erosion. Given the dominance of non-topsoil erosion processes on the CLP, this suggests that the high sediment production of the CLP is mainly attributable to natural factors.