Landslides are purportedly the dominant agents of erosion in humid and tectonically active
mountain ranges. The rate of landsliding is also thought to compensate for effects of rock
uplift on mountainous landscapes by maintaining threshold hillslopes and limiting growth of
topographic relief. Still few data exist to corroborate this widely held notion. Here I review
rates of landslide erosion from around the world, and show that they vary by at least six
orders of magnitude, i.e. from 10-3 to 102 mm/yr. Part of this variability is systematically
linked to the timescales of interpolation, such that mean rates are hardly meaningful let alone
representative. The highest rates are invariably tied to tectonically active mountain belts that
experience frequent earthquake shaking, high precipitation, and sustain dense forest
vegetation. Landslide erosion rates derived for short-term periods (100–102 yr) rely
on higher data density, but are prone to bias by regional triggering events such as
earthquakes and rainstorms, anthropogenic changes to land cover, and the occurrence of
large, infrequent landslides. In contrast, the considerably lower long-term landslide
erosion rates (103–104 yr) suffer from incomplete records because of erosional
censoring, and thus remain underestimates. Despite this shortcoming, there is strong
evidence that the larger landslides are dominating volumetric sediment production in
tectonically active mountain belts. At a global scale, soil erosion appears to be the
only other process capable of matching the rates of landslide erosion. Yet, data on
specific yields of mountain rivers and glacier-dominated drainage basins indicate that
landslide-produced sediment may be routed more or less at capacity in many settings. |