Understanding the dynamics of organic matter (OM) at global and local scales is one of the challenges in the environmental sciences and i.e. terrestrial biogeochemistry. The accurate identification of OM is an essential element to achieve this goal. In our study, a novel application for quantitatively apportioning sources of eroded sedimentary OM from an eco-geomorphologic perspective was shown successfully via a coupled molecular n-alkane biomarkers and stable isotopic signatures (¹³C and ¹⁵N) along with elemental compositions (TOC and TN) using a Bayesian mixing model (SIAR). Soil source samples were collected from different land use types (i.e., forests, grassland, cropland, and fallow) and gully, which were probably transported downstream along the steep terrain. Meanwhile, three soil profiles with a total of 90 sediment samples were also sampled in check dam. The results indicated that cropland was the main sedimentary OM source in this catchment, contributing 29.5%, whereas the forests, grassland, fallow and gully contributed 12.17%, 15.39%, 21.53% and 21.85%, respectively. Although the molecular biomarker as a tracer was not valid solely, the combined approaches of n-alkanes biomarker and bulk parameters were efficient complements in tracing OM source in a hilly-gully region on the Loess Plateau of China.