Downslope soil transport in the absence of overland flow has been attributed to numerous mechanisms, including particle-by-particle creep and disturbances associated with biological activity. Process stochasticity and difficulties associated with field measurement have obscured the characterization of relevant long-term soil transport rates and mechanisms. In a series of incised fluvial terraces along the Charwell River, South Island, New Zealand, we documented vertical profiles of tephra concentration and topographic derivatives along a hillslope transect to quantify soil transport processes. Along the undissected hilltop, we observed a thin primary tephra layer (ca. 22.6 ka) within loess deposits ∼80 cm below the landscape surface. In the downslope direction, the depth to the highly concentrated tephra layer decreases, coincident with an increase in hillslope convexity (which is proportional to landscape lowering rate if soil flux varies linearly with hillslope gradient). Exhumation of the tephra layer results from landscape lowering due to disturbance-driven soil transport. Approximately 20 m downslope of the interfluve, the depth to the tephra layer declines to 40–50 cm, peak tephra concentrations decrease by a factor of 4, and tephra is distributed uniformly within the upper 40 cm of soil. The transition from a thin, highly concentrated tephra layer at depth to less concentrated, widely distributed tephra in the upper soil may result from soil mixing and transport by biological disturbances. Along our transect, the depth to this transition is ∼50 cm, coincident with the rooting depth of podocarp and Nothofagus trees that populated the region during much of the Holocene. Our observations can be used to calibrate the linear transport model, but, more important, they suggest that over geomorphic time scales, stochastic bioturbation may generate a well-mixed and mobile soil layer, the depth of which is primarily determined by flora characteristics.