Soils record both material and energy fluxes.  We adopted a catena approach to ask the question: Where are differences in soils along hillslopes maximized, and what are ecosystem consequences of this differentiation?  We used a mass balance approach to quantify elemental gains and losses relative to the underlying granite.  For each catena, we calculated a delta hillslope as the downslope mass balance minus the upslope mass balance; a hillslope differentiation index (HDI) was then calculated as delta hillslope divided by the distance between pedons.  Field sites ranged in soil residence times from ~3 ky (San Gabriel Mountains, California) to ~500 ky (Kruger National Park, South Africa).

Results/Conclusions

Hillslope differences--as quantified by elemental mass losses and gains of Si, Al, Fe, Na, and P along different parts of hillslopes--are maximized under the semiarid climate, gentle slopes, and long soil residence times that characterize African catenas.  For the youngest catenas in the San Gabriel Mountains, delta hillslope values for SiO2, Al2O3, Fe2O3, Na2O, and P2O5 were 203, 27, 3.9, 5.4, and 0.104 kg m-2, respectively.  For the Sierra Nevada catenas, corresponding delta hillslope values were 66, 119, 129, -10, and 0.045 kg m-2.  For the Kruger National Park catenas, corresponding delta hillslope values were 341, 140, 39, 16, and 0.431 kg m-2.  HDI for SiO2, Na2O, and P2O5 were greatest for the South African catena.

Although mass losses of Si were generally smaller for downslope than upslope pedons, two distinct processes appeared responsible. In the San Gabriel Mountains, hillslopes responding to rapid tectonic uplift show much lower mass losses downslope because short residence times preclude chemical weathering or illuvial gains of weathering products from upslope.  At older sites, smaller mass losses of Si reflect illuvial mass gains from upslope.  While soil differences were more pronounced along the African catena, the differences we measured on younger catenas suggests considerable hillslope differentiation can occur during the first several millennia of soil development.  We observed little aboveground expression of these soil differences along the younger California catenas.  The extreme differentiation of the African catenas, by contrast, coincided with an abrupt ecotone between nutrient-poor and nutrient-rich savanna.  Ecosystem consequences of hillslope differentiation, therefore, appear to depend on the tempo of catenary dynamics; only where soil residence times exceed some threshold can vegetation and fire, for example, act to reinforce these differences, making poor upslope soils poorer and rich downslope soils richer.