The river that never was: Fluvial taphonomy at Olduvai Bed I and II sites and its bearing on early human behavior.

Olduvai site integrity has been questioned through interpretations of fluvial inputs on most assemblages from Beds I and II that rest on a contradictory use of taphonomic variables, lack of geological support, and lack of adequate experimentally-supported referential frameworks. Most variables used to address fluvial impact in faunal assemblages have been independently used in other contexts and all of them are subjected to variable degrees of equifinality. Their resolution when applied to archaeological assemblages is low. A more recent approach founded on experimentation simulating actual archaeological sites shows that it is the intrinsic properties of each bone specimen (i.e., composition and shape) that best determine the resistance of each bone to be moved by water, regardless of size. The application of this approach, which combines the relative representation of specimens according to their composition (i.e., texture: dense or spongy-trabecular or mixed) and shape (flat, tube, or cube-polyhedral), shows not only that fluvial inputs on bone assemblages can be well detected, but also that the scale and magnitude of such inputs can be efficiently assessed. Thus, site formation can be more accurately interpreted with the aid of these new experimental analogical frameworks. A selection of some of the most relevant Olduvai sites from Bed I and Bed II have been used to show that all represent autochthonous concentrations of fossil materials and that their potential fluvially-caused distortion was marginal in most sites and moderate (in the form of semi-lag assemblages) in just a couple of them. This reinforces the causality of biotic agents (i.e., hominins and carnivores) in the formation of these sites.