This paper addresses the changes in the tectonic regime in the Peruvian and Bolivian Andes that have occurred since the upper Miocene when the present-day elevation of the Cordillera above sea level has been almost reached. The stress patterns are deduced essentially from a field study of fault kinematics and a numerical inversion of the slip vector data measured on the fault planes. The Cuzco fault system in southern Peru is chosen as an exmaple to illustrate the methodology used. In this region, striations on both active and Holocene faults are in agreement with a N-S extension. But faults affecting early Pleistocene deposits exhibit two families of striations. The younger results from the previous N-S extension: the older, involving reverse motions, results from either an E-W or a N-S compression. Faults affecting Pliocene formations often show an oldest family of striations resulting from a NE-SW or an E-W trending extension. Thus three tectonic regimes are demonstrated which are also supported by regional unconformities and sedimentological data: (1) a Pliocene extensional regime, (2) a lower Pleistocene compressional regime, and (3) a mid-Pleistocene-present-day extensional regime. Similar analyses conducted in the Pacific and sub-Andean lowlands allow sketching of the successive Pliocene-Pleistocene stress patterns in the Central Andes.

The Quaternary and present-day stress pattern is characterized by a N-S extension in the High Andes and in the Pacific lowlands and by an E-W compression in the sub-Andean lowlands and at the contact between the Nazca and South American plates. This stress pattern is interpreted at a large wavelength (>100 km) as an effect of compensated topography. This model supposes that the vertical lithospheric stress, &sgr;ZZ, increases with the topography, the crustal thickness, and the low-density mantle beneath and that the lithospheric maximum (compressional) horizontal stress &sgr;Hmax, trending E-W roughly parallel to the convergence, is fairly constant. On both edges of the Andes, the tectonics being compressional, &sgr;ZZ is &sgr;2 and &sgr;Hmax is &sgr;1. In the High Andes, &sgr;ZZ becomes &sgr;1, then the E-W trending &sgr;Hmax is &sgr;3 and &sgr;Hmin trending N-S is &sgr;3, allowing extension to occur in this direction. The Pliocene stress pattern was characterized by a NE-SW or an E-W trending extension in the High Andes, in the Pacific lowlands, and possibly in the sub-Andean lowlands. This stress pattern was clearly different from the present-day one because the E-W trending stress was &sgr;Hmin. This required a weak push or, eventually, tractional boundary forces acting on the South America lithosphere. It is suggested that this might result from a strong slab pull due to a long, steeply dipping slab which decreased the value of the &sgr;XX stress transmitted to the overriding plate. The early Pleistocene state of stress was compressional.

Since the elevation of the Andes had not markedly decreased during this period, this required an increase of the E-W trending stress value. This resulted from a strong coupling between the two lithospheric plates, possibly due to a rupture or a long slab under its own weight. Other spatial changes in the stress pattern are related to the particular situation of the forearc, to the subduction of the buoyant Nazca ridge, and to the different dips of the slab. Extension in the High Andes is of small magnitude, of the order of 1% during the last 1--2 m.y.; in a few basins, it may have attained 40% during the Pliocene (≈5--3 m.y.). ¿ American Geophysical Union 1992