Neogene sedimentary deformation in the Chilean forearc and implications for Andean basin development, seismicity and uplift

Neogene sediments in the northern Chilean forearc display a wide range of near syndepositional structures. Analysis of the origin and distribution of these structures in space and time offers new insights into the development of the forearc basins. The structures are described in detail and show many features associated with soft-sediment deformation, pseudo-diapirism and slope failure. Synsedimentary deformation reached a peak in the Late Miocene to Early Pliocene while the sediments were saturated in a largely plastic state, and many of the structures were probably triggered by seismic shock. Late-stage tilting of the forearc generated shear stresses in the sediments leading to slumping and sliding. Base-level revision and drainage incision led to sediment bypass and cessation of lacustrine sedimentation that was not necessarily linked to climate change. Compaction and dewatering of the basins caused transition of the sediments from a plastic to a brittle state. The age and distribution of structures associated with seismicity appears to correlate with increasing subduction erosion and westward drift of South America but not with basin subsidence, shortening rates or plate convergence. This suggests that upper crustal deformation is at least partly decoupled from plate movement.

The uplift of the Andes as a result of Cenozoic subduction generated a series of longitudinal, terrestrial forearc basins lying between the Coastal Range and the Western Cordillera of the Andes. The timing of this uplift and the developmental history of these basins have important implications for deciphering palaeoclimate and palaeohydrology as well as the tectonic processes of the western Andean slope and Atacama Desert.
Widespread marine sediments in the central Andes show that the area was at sea level during much of the Cretaceous and earliest uplift began with the proto-Precordillera during the Late Cretaceous. Gregory-Wodzicki (2000) and Kennan (2000) suggested that the Western Cordillera had reached 1000 m elevation by 30-40 Ma and 2000 m by 10-20 Ma. More recent work by Victor et al. (2004) and Elger et al. (2005) suggested that significant uplift (of the order of 2000 m) and shortening occurred during the Late Oligocene to Miocene (c. 25-10 Ma), with further uplift and tilting c. 10-5 Ma as a result of westward ductile flow of the lower crust (Farías et al. 2005) or removal of dense lower crust and mantle lithosphere (Garzione et al. 2006). Victor et al. (2004) pointed out that uplift was diachronous from north to south, and Garzione et al. (2006) suggested that at 188S the main uplift (2500-3500 m) occurred between 10.3 and 6.8 Ma.
The Neogene sediments of the northern Chilean forearc between 21 and 238S are host to a rich variety of near syndepositional deformational structures, many of which may be attributed to 'soft-sediment' deformation. These structures can be observed over a wide area and through a thick sequence of sediments as a result of the exceptional preservation and exposure in the region. Coupled with increasingly detailed stratigraphic control, an analysis of the deformation structures and their distribution in space and time offers new insights into the development of forearc basins, their compaction, dewatering and seismic history, as well as Andean uplift and climate change.
Previous geological work in the area included the surveys and mapping by Ferraris (1978), Skarmeta & Marinović (1981), and Marinović & Lahsen (1984). Naranjo & Paskoff (1981) defined the El Loa and Chiu Chiu Formations; the former is now given Group status and comprises the Opache and Jalquinche Formations. They considered the domes and hollows located in the Calama Basin to be due to differential compaction of unconsolidated sediments in a subsiding basin and not related to regional tectonism. Naranjo & Paskoff (1982) suggested that the folds at Puente Posada, and other deformational structures below Calama, had a gravitational origin. Jensen (1992), Bao et al. (1999), and Sáez et al. (1999) provided a detailed study of the sedimentology and Neogene palaeoenvironmental history of the Quillagua-Llamara Basin, whereas Pueyo et al. (2001) specifically discussed the evaporitic sequences. May (1997), May et al. (1999May et al. ( , 2005 and Blanco et al. (2003) undertook a detailed analysis of the sedimentology, palaeoenvironmental conditions, and stratigraphy of the Calama Basin, leading to a revision of the stratigraphic terminology as used in this paper.
The current study is based on extensive mapping, drilling, geophysics, and hydrological sampling and testing carried out by Nazca S.A. as part of a regional groundwater resource investigation during 1999-2001.

Geological and tectonic setting
The forearc of northern Chile (Fig. 1) consists of several Tertiary basins including the Pampa Tamarugal, which extends from 19 to 248S within the Longitudinal Valley at an average elevation of 1000 m, and the Calama and Turi Basins at 238S, at 2300 m and 3100 m, respectively (Fig. 2). The Turi Basin, over 300 m deep (Houston 2004), is a fault-controlled extension to the Calama Basin, which is at least 700 m deep (May et al. 2005), and may be up to 1.5 km deep (Pananont et al. 2004). This in turn is linked to the Pampa Tamarugal by the valley of the Río Loa, which is incised through the Precordillera (the Calama Gap). The Quillagua-Llamara Basin at the southern end of the Pampa Tamarugal has been proved to be over 900 m deep and may be as much as 2 km (Digbert et al. 2003).
The central Andean region has experienced a compressional regime since at least 50 Ma as a result of subduction of the Nazca plate below South America. Convergence increased to a maximum of 15 cm a À1 during the Late Oligocene-Early Miocene (26-20 Ma), steadily decreasing thereafter to 8 cm a À1 by the Pliocene (Somoza 1998). Periods of increased convergence have been linked with uplift and deformation, whereas slow convergence rates have been associated with extensional tectonics (Sébrier et al. 1988;Scheuber et al. 1994). Separating the components of convergence, Silver et al. (1998) suggested that Andean deformation and uplift are specifically the result of the increasing westward velocity of the South American plate since 30 Ma.
Within this compressional regime the widespread occurrence of arc-parallel extensional basins, faulting, and jointing across the forearc and Precordillera including the Pampa Tamarugal and Calama Basins has been reported (Jensen et al. 1995;Pananont et al. 2004). These extensional features have shown the importance of subduction erosion processes that lead to trenchward gravitational collapse, forearc subsidence, and extension in a thin-skinned zone that may not necessarily be directly linked to convergence rates (Delouis et al. 1998;Hartley et al. 2000;von Huene & Ranero 2003).
The Cenozoic stratigraphic successions in the Pampa Tamarugal, Calama and Turi Basins are similar (Figs 2 and 3), with a conglomerate-red bed-fluvio-lacustrine-evaporite sequence that is typical throughout the central Andes from northern Chile to southern Peru and the Bolivian Altiplano (Allmendinger et al. 1997;Lamb et al. 1997).
Widespread Eocene to Oligocene basal conglomerate and sandstone of the Calama, Sichal and other unnamed formations reach thicknesses of 400 m in local depocentres. The conglomerates are generally clast-supported, and consist of locally derived material with sheet-like geometries. Early to Middle Miocene sedimentation of the Hilaricos, Batea, Jalquinche, and Lasana Formations reached thicknesses of .200 m. They consist of a range of gravel, sand, silt and clay deposits with pervasive evaporitic content. The deposits show a series of fining-upward sequences with predominantly sheet-like geometries, although channel-fill deposits are more common around the basin margins.
Late Miocene to Pliocene deposits occur in the Turi and Calama Basins, extending through the Calama Gap to the Pampa Tamarugal. The Toconce Formation in the Turi Basin consists of multiple andesitic air-fall and fluvially reworked tuffs, grading into sheet sandstone and single-to multi-storey channel-fill conglomerates of the Chiquinaputo Formation downgradient. The centre of the Calama Basin and the upper part of the Calama Gap are largely composed of intraclast peloidal limestone, packstone and wackestone of the Opache Formation. The carbonates grade into calcite-cemented sandstone and conglomerate with continuous broad channel geometries west of the Calama Gap. Clastic carbonate, marl, intraclast and brecciated limestone occur in the Puente Posada area in sheet-like and channel-fill geometries. The Quillagua Formation in the Quillagua-Llamara Basin consists of laminated marl, silt and sandstone with frequent diatomaceous beds, and channelized conglomerate.
In the southern Pampa Tamarugal and Calama Basin, extensive areas of sediments of Late Pliocene to Pleistocene age overlie Miocene-Pliocene formations. The Chiu Chiu and Soledad Formations are formally defined units of this age occurring in the Calama and Quillagua-Llamara Basins, respectively. The Chiu Chiu Formation is largely composed of clay, marl, and laminated diatomite with occasional channel-fill conglomerate and reworked volcanic ash. The Soledad Formation largely consists of evaporites with fine-grained clastic sediments. Volcanic activity from strato-volcanoes and calderas in the Western Cordillera has produced widespread interbedded andesitic to dacitic tuff and ignimbrite in the Pampa Tamarugal, Calama and Turi Basins (de Silva 1989;Wörner et al. 2000) throughout the Cenozoic. However, there is a well-recognized concentration of lava and pyroclastic deposits as a result of the Late Miocene 'flare-up' in the Western Cordillera, followed by renewed activity in the Plio-Pleistocene.

Stylolites, piping and syneresis cracks
Well-developed cuspate stylolites with wavelengths in the centimetre-decimetre range and amplitudes of a few centimetres occur frequently in the evaporitic mudstones of the Hilaricos, Batea and Jalquinche Formations (Fig. 8b) often associated with fenestrae. Stylolites associated with fenestrae appear to be less well developed in the Opache and Quillagua Formation limestones. Stylolites are formed by pressure solution (Collinson & Thompson 1982). Piping is common in the coarser-grained facies of the Chiquinaputo, Toconce, and Lasana Formations of the Calama and Turi Basins (Houston 2004;Fig. 11). Piping is caused by dissolution, elluviation, seepage face erosion and tunnel scour (Parker & Higgins 1990).
Syneresis cracks (Fig. 6c) are typically 2-5 cm in length and ,2 cm in depth. They are restricted to the lacustrine and palustrine carbonates of the Opache Formation in the Calama Basin and at Puente Posada. Syneresis cracks have been attributed to subaqueous shrinkage as a result of pore-fluid escape (Plummer & Gostin 1981).

Load structures
Based on the classification of Owen (2003), the full range of load structures can be seen throughout the Miocene sequences in rocks ranging from micritic wackestones through evaporitic mudstones to sandstones, from the Turi Basin to the Quillagua-Llamara Basin. On the other hand they are virtually absent from the overlying Upper Pliocene to Pleistocene strata. Simple and pendulous load casts (Fig. 4a) range in size from 0.5 to 10 cm, associated with flame structures (Figs 4a and 7d). Pseudonodules (Figs 4b and 8c) also occur with a similar range of size, occasionally internally zoned and apparently transitional with both disrupted bedding and brecciation (Fig. 4b). Ball and pillow structures (Figs 5g and 8c) in the decimetre-scale range are less frequent but still common, and some show signs of rotation ( Fig.  5g).
Load structures as described are associated with a reverse density gradient undergoing Rayleigh-Taylor instability at the interface and/or with uneven loading during temporary liquefaction (Owen 2003).

Dish and pillar structures
Dish structures (Fig. 4b) occur occasionally in micritic wackestones of the Opache Formation in the Calama Basin. They are typically a few centimetres across and c. 1 cm deep, and may be associated with small (millimetres diameter, centimetres tall) pillars. The dish itself and pillar are composed of clay particles. They are frequently associated with small-scale brecciation and fenestrae. The pop-up structure identified in Figure 4b, which ruptures a thin layer of homogeneous carbonate overlying the disturbed zone, appears to be genetically linked to the brecciation and underlying dish structures. Such structures have been shown to be the result of fluidization during the upward escape of pore water (Lowe & LoPiccolo 1974;Lowe 1975).
Vertically oriented pillars of larger dimensions (centimetres diameter, up to 20 cm high) are found in low-permeability facies of the Jalquinche and Opache Formations in the Calama Basin. They may be rooted in an obvious source bed, and may terminate in overlying sill-like bodies with embedded breccias (Fig. 4c). These pillars are also interpreted to be due to water escaping along fluidized channels.

Breccias
Breccias occur frequently in playa-delta-lacustrine facies associations of the Soledad, Chiu Chiu, Quillagua, Opache, Hilaricos, Batea and Jalquinche Formations. They do not occur in fluvial or pyroclastic facies associations of the Toconce and Lasana Formations. Based on the classification of Morrow (1982), a wide range of breccia types occur, ranging from stratiform, mosaic floatbreccia through packbreccia (Fig. 4c) and both lateral ( Fig. 5d and e)  and vertical rubble ( Fig. 5b and c) to cemented crackle breccia ( Fig. 6c and d). Clast sizes are widely variable from millimetre to decimetre scale, but are generally monomict, clasts having travelled only short distances. Clasts are composed of wackestones, packstones and grainstones in carbonates, and siltstones and sandstones in clastic rocks, with matrices of marl or mudstone and siltstone.
Stratiform breccias are more frequent and extensive (up to .100 m laterally) within the sheet evaporitic mudstones of the Hilaricos, Batea and Jalquinche Formations, but also occur in the Quillagua and Opache Formations. The breccia units frequently have a relatively sharp lower contact with insoluble residual material present, and a more irregular, fractured upper boundary (Fig. 4c), and often show a lateral transition into undisturbed or convolute bedding (Fig. 5d). These features are consistent with a solution collapse origin as described by Stanton (1966) and Warren (1999).
At Puente Posada the cemented crackle breccia ( Fig. 6b-d) contains remnants of the original intraclast limestone with fenestrae, syneresis cracks, vugs and multiple vein infill, indicating a complex polyphase diagenesis as a result of circulating groundwaters in alternating phreatic and vadose environments. The breccia is interpreted to be contemporaneous in origin with the slumping that is common at this horizon (see Supplementary Publication available online at http://www.geolsoc.org.uk/ SUP18284.

Clastic dykes
Clastic dykes (Fig. 5a-c) are found in carbonate-cemented sandstones, clastic carbonates and travertines of the Opache Formation in the Calama Basin and Gap. Their distribution suggests that they are more common in the vicinity of faults (the Precordillera fault zone, West Fissure, Angostura Fault and Ayquina Horst). The dykes have an irregular shape, with widths commonly in the decimetre range and lengths up to 2 m, usually having an identifiable source bed and frequently manifesting silllike projections. Some show evidence of more than one phase of formation (Fig. 5a). Internally, the dykes consist of matrixsupported clasts, whose origin is the underlying source bed and dyke walls. Larger clasts may be displaced downwards after spalling from the dyke walls. The matrix frequently shows subvertical lamination suggesting vertical shear.

Disturbed and convolute laminates and tepee structures
Disturbed and convolute bedding is widespread in medium-to fine-grained clastic rocks, tuffs, limestones and especially traver- tines of the Chiu Chiu, Opache and Quillagua Formations throughout the Turi, Calama and Quillagua-Llamara Basins. These structures are typically limited to a particular bed within a sequence, ranging from centimetre to metre scales. A wide range of forms are exhibited, from disrupted convolute bedding (Fig.  5d), which may be closely associated with brecciation, through typical convolute bedding occurring discretely, to tepee structures that show gentle synclines and sharp anticlines, which may be overturned (Fig. 5f) or display anticlinal axial faulting (Fig. 5e). Convolutions tend to increase in intensity upwards within the bed ( Fig. 5e and f) and may affect the base of overlying beds (Fig. 5f), die out (Fig. 5d) or be truncated (Fig. 5e).
Convolute lamination may occur where sediments are underconsolidated, and plastic deformation allows the escape of pore water from partially liquefied sediments (Collinson 1994). Tepee structures indicate a strandplain environment with fluctuating saturation and salinity.

Slumps, slides, thrusts and collapse structures
Large-scale (metres to kilometres) slumping appears to be largely restricted to the Opache and Quillagua Formation clastic carbonates of the Calama Gap and Quillagua-Llamara Basin. Smallscale (centimetres) examples of slumping are commonly found in travertines and diatomites throughout the area and to a lesser extent in the Chiu Chiu Formation.
An example of small-scale thrusting in travertine from the Opache Formation close to the Ayquina Horst is shown in Figure  7d. Whereas the upper marked bed appears to be disrupted by a micro-thrust and water escape pipes, the lower marked bed is still contiguous and has developed a flame structure into the micro-thrust zone. A micro-fold with incipient thrusting in laminated travertine is shown in Figure 7c. Disrupted laminations and bedding are extremely common in travertines in the Turi, Calama and Quillagua Basins, and suggest that deformation took place whilst the travertine was in a gell-like state.
Laminated travertine deposits are well developed within the Quillagua and Opache Formations in the Quillagua-Llamara and Calama Basins, and show signs of both inorganic and organic (cyanobacterial and charophytic) origin. Invariably they display a range of deformation, including brecciation, disturbed lamination, and small-scale slump folding and thrusting. The travertines are frequently found associated with fault zones, and in these environments, are considered to be the result of CO 2 degassing of emergent groundwater rising along the fault (Chafetz & Folk 1984;Hancock et al. 1999).
The largest and best-defined slumps occur at Puente Posada, where a wide spectrum of deformational features occurs, over an area of 15 km 2 . Slump folds (Fig. 6a) affect multiple beds from 1 to 10 m thick and are interbedded between undisturbed sediments. Upper surfaces are sometimes eroded. Slumped pods of brecciated and contorted beds are common at specific horizons that can be traced up to several hundred metres (Fig. 6b). Slump folds are open, asymmetric, and gently non-cylindrical with a wide range of axial-strike from SSW to NNE, and exhibit gentle plunges (Fig. 6, inset). This style of folding is consistent with plastic deformation of partially lithified sediments undergoing gravitational collapse on a shallow slope oriented towards the west and north (Lajoie 1972).
Rotational slumps found in the Quillagua area ( Fig. 7f) have dimensions of 5-10 m and show basal shear surfaces that tend to be strike-parallel to valley axes. There is relatively little internal distortion. Their geometry and occurrence suggest that they formed on shallow slopes as brittle failures, probably associated with the release of high pore pressures, during valley incision, and are therefore relatively late-stage structures.
Associated with the rotational slumping, mesoscale thrusts are found in the Quillagua Formation between Puente Posada and Quillagua. They show multiple bifurcations and a ramp-flat geometry ( Fig. 7e) with dimensions of several metres. Displacements are usually not more than 1 m. Even though it is difficult to establish their orientation, they appear to indicate a movement generally towards the north. Their orientation and association with rotational slumping point to late-stage formation.
Collapsed blocks of the Quillagua and Opache Formations, from metre to tens of metres size, frequently rotated, are commonly found along the Ríos Loa, San Salvador and Salado (Fig. 8d). They overlie fine-grained siltstones and mudstones of the Hilaricos, Batea and Jalquinche Formations. Foundering of the overlying strata is a result of valley incision and bank slope dewatering.

Fold domes
A series of folds in the centre of the Calama Basin largely affecting the Opache and Jalquinche Formations (Fig. 8a) are related to other structures in the Calama Basin (Figs 9 and 10). They have a complex, domal shape with no preferred orientation (dip directions are random at p ,0.01, based on a Rayleigh test; Fisher et al. 1987). The domes change into a weakly developed rim-syncline that surrounds them both. In the core of the domes the Opache Formation is eroded and the underlying Jalquinche Formation is exposed, overlain by hummocky reworked evaporitic clays, suggestive of diffuse pore-fluid escape from deeper liquefied strata. The Opache Formation forms relatively steeply upturned strata (c. 78) around the cores of the domes, and contains widespread brecciation and disturbed bedding. Dips decrease away from the dome cores to only a few degrees around the basin margin. Deposition of the Chiu Chiu Formation is constrained by the prior initiation of the domes, and warping of the Chiu Chiu Formation around the lower slopes of the domes indicates continued movement into the Plio-Pleistocene.

Monoclines and faults
The central part of the Calama Basin is partially surrounded, on the NE, NW and SW, by a series of inward-dipping monoclines (Figs 7b and 9) affecting the Opache Formation, underlying ignimbrites and Jalquinche and Lasana Formations. They can be seen to be genetically linked to the fold domes and rim syncline of the basin centre. At some localities (e.g. Angostura, Ayquina Horst), these monoclines are associated with complex synsedimentary normal growth faults (Fig. 7a) that appear to be listric although no basal décollement is exposed, and that have variable throws of up to several tens of metres towards the basin centre. At localities where the monoclines are not obviously associated with faults, it is speculated that they may be underlain by blindfaults. Both the monoclines and faults are interpreted to be due to synsedimentary subsidence of the central part of the Calama Basin, which generated basin boundary collapse features.
On the SE side of the basin, subsidence is constrained by the limits of competent ignimbrite flows that occur between the Opache and Jalquinche Formations, inhibiting deformation of the latter. By contrast with the deformed basin centre, the tablelike, undisturbed nature of the Opache Formation limestones between Angostura and Calama suggest underlying, stable bedrock.
The progressive displacement of sedimentary units from the Opache to Jalquinche Formations on the growth faults, coupled with fault-controlled restrictions on the extent of the Chiu Chiu Formation sedimentation in the basin, shows that these faults have been active from at least the Middle Miocene to the latest Pliocene or later.
A series of small faults occur below Calama in the Calama Gap, apparently restricted to the Opache Formation. They seem to occur as conjugate sets with orientations of 000-0208 and 290-3108, with lengths less than 5 km, and displacements of less

Soft-sediment deformation
The term 'soft-sediment' is here used in relation to deformational structures that occur in largely unlithified sediments and are not part of the depositional fabric. Nevertheless, any aspect of deposition that leaves the sediment in an unstable condition will facilitate later deformation (namely, inversely graded bedding, small-scale density inversion, and lateral variations in bed thickness), as well as rapid deposition leading to underconsolidated beds. Such variations in sedimentation will also be reflected by their basic hydrophysical properties, porosity and permeability, which will control many subsequent processes.
The widespread occurrence of various types of load structures throughout the Miocene in all environments indicates the weakening of sediments as a result of slow pore-fluid dissipation that does not keep pace with lithostatic loading (Maltman & Bolton 2003). The absence of load structures from suitable host sediments in the Plio-Pleistocene Chiu Chiu and Soledad Formations indicates either insufficient overburden pressure or saturation-density for their generation in the most recent sedimentary deposits.
Increased loading leads to liquefaction of the sediments as the load becomes entirely borne by the pore-fluid, such that the material loses all strength, and is exemplified by disrupted convolute bedding throughout the Upper Miocene to Pliocene sequences, and dish and pillars, which are restricted to Upper Miocene Opache Formation of the Calama Basin. Once the porefluid starts to move, fluidization occurs, frequently as a result of an external trigger. The clastic dykes and their enclosed breccias found in the Upper Miocene units from Puente Posada to the Calama Basin are considered to be the result of fluidized injection through a ruptured caprock. In some cases, gravitational shear stresses are implied by overturning and rotation ( Fig. 5f and g).
The shallow burial of these sediments points to high pore pressure rather than overburden load (,100 kN m À2 at 50 m depth) as a cause for liquefaction and fluidization. Such processes allow water escape through zones of (temporary) higher permeability and transition to early diagenetic processes including dissolution and piping, leading to stylolites and collapse breccias. The latter features are common throughout the Lower-Middle Miocene sequences of the Calama and Turi Basins, and inferred to be the result of a combination of groundwater flushing and the high frequency of soluble evaporitic minerals.
Late-stage depositional to early diagenetic structures are abundant. Breccias are the most widely distributed, occurring throughout all areas from the Lower Miocene to Pliocene units, and less frequently in the Pleistocene units, and are the result of varied processes including syneresis, dissolution, fluidization and synsedimentary folding and faulting.

Pseudo-diapirism
The appearance of the domes in the central Calama Basin is suggestive of diapirism, showing features associated with the upward intrusion of a low-density, viscous mass. Such classical diapirs are defined by their discordant contacts with host rocks (Warren 1999) and occur in both extensional (Jackson & Vendeville 1994) and contractional regimes (Brun & Fort 2004). Their initiation is usually due to Rayleigh-Taylor instability at the surface of a density inversion between evaporite and sediment overburden, at a depth theoretically equivalent to c. 0.4 times the distance between adjacent diapirs (Turcotte & Schubert 2002). Inter-domal distances in the Calama Basin vary from 1 to 4 km, which would suggest overburden depths between c. 390 and 1560 m. However, drilling evidence indicates an average overburden (Jalquinche plus Opache Formation) thickness of c. 200 m (maximum 290 m), insufficient to conform to the theoretical relationship. The underlying coarse-grained Oligocene-Eocene clastic sediments, and bedrock, are considered unlikely to participate in diapirism. Furthermore, drilling does not indicate any discrete evaporite bed at depth, and geophysical logging does not indicate any significant density inversions. Also, the domes are not accompanied by any radial or arcuate faulting, nor have any discordant boundary contacts been observed. The domes in the Calama Basin are not, therefore, diapirs sensu stricto.
The presence of a weakly developed rim syncline, on the other hand, suggests that differential loading has occurred, and in the core of the domes there is evidence for diffuse pore-fluid escape from deeper liquefied strata. The Jalquinche Formation is, today, overpressured (seen, for example, as squeezing clays during drilling) as a result of the low permeability of the mudstones, and artesian pressure from the underlying Yalqui and Calama Formation aquifers (Houston 2004). Geothermal heating, as evidenced by deep (.250 m) groundwater with temperatures up to 31 8C and geothermal gradients from 0.02 to 0.08 8C m À1 , may also contribute to overpressurization. It is probable that overpressurization was present during burial, and under these conditions, lateral and vertical flow of sediment took place, enhanced particularly by dissolution of disseminated evaporite minerals, liquefaction and differential loading. The domes are thus considered to represent pseudo-diapirs, somewhat analogous to the mudlumps of the Mississippi delta (Morgan et al. 1968), or the mobile shales and muds described by Morley & Guerin (1996), at an early stage of development, now largely frozen as a result of dewatering and conversion from a ductile to a brittle state.

Slope failure
Slope failure in the form of slumps and slides occurs at different scales and in different environments. Meso-scale slumps and slides are relatively common in the Plio-Pleistocene Chiu Chiu Formation and the Upper Miocene-Pliocene Quillagua and Opache Formations, frequently associated with small-scale folding, convoluted laminations and brecciation. The larger slumps and slides are seen to occur in an intra-formational setting in the Quillagua and Opache Formations of the Quillagua-Llamara Basin and Puente Posada area, as well as associated with sidewall failure of the major incised rivers. Intra-formational slope failure is essentially synsedimentary and occurs on very shallow slopes whilst the sediment is unlithified and saturated. On the other hand, slope failure associated with river incision, which can be constrained to 3.3-2.5 Ma, is strictly post-depositional.
Based on assumed properties typical for such materials (see Supplementary Publication; see p. 296), calculations show that the intra-formational slumps observed at Puente Posada have a static factor of safety less than unity for any degree of saturation prior to diagenesis. They therefore probably occurred as a result of gravitational shear failure whilst pore pressures were greater than zero. In contrast, the rotational slump at Quillagua, as well as the collapsed blocks that occur throughout the Río Loa incised valleys (especially at Ojos de Opache), all have static factors of safety above 1.2 even under saturated conditions, indicating that a trigger was required.
Under drained conditions these slumps all have high factors of safety, including the intra-formational slumps once water levels fall below the toe of the slump. This signifies that slope failure was an active process only during the early stages of valley incision before any significant dewatering took place, and is not active today.

Compaction and dewatering
Data from the Calama Basin provide an insight into its subsidence history and allow estimates of compaction and dewatering to be made (see Supplementary Publication; see p. 296). The geometry of the Neogene basin fill has been determined from field mapping and c. 3600 m of well and geophysical logging, together with c. 1800 m of outcrop section. The resulting isopach maps allow formation volume, area, and mean and maximum thickness to be estimated (Table 1), where mean thickness is simply volume divided by area.
Composite neutron-density logs averaged from three wells penetrating the Opache and Jalquinche Formations show smallscale variations as a result of lithological changes, overpressured zones, fissuring and calcrete cementation associated with palaeosol horizons. Nevertheless, porosity follows a well-defined logarithmic stress-strain relationship (Jackson et al. 2004) that results in a loss of porosity, surface subsidence and a consequent release of water with increasing burial. The effects of burial can be deconstructed using a backstripping technique, allowing effective subsidence rates and dewatering to be estimated (Table 1).
The mean subsidence rate (uncorrected for loading or regional uplift) was at a maximum during the Miocene, decreasing significantly post-Late Pliocene, so that additional basin fill (Chiu Chiu Formation) was largely accommodated by compaction rather than subsidence.
The volume of water liberated as a result of compaction slowly decreased from over 8 km 3 Ma À1 to 2 km 3 Ma À1 (equivalent to 0.008-0.002 Mm 3 a À1 ) as the basin became filled with sediment. These rates are insignificant in comparison with current drainage losses from the Río Loa Basin, which are estimated to be around Fig. 10. Structural cross-sections through the NE (above) and SW (below) domes based on surface traverses and borehole logs. Dashed lines show bedding structure deduced from surface dips. Density log (r, scale 1-3 g cm À3 ) and neutron-porosity log (Ł, scale 0-1) from well E1 are superimposed. The location of the sections and symbols are shown in Figure 9. Vertical exaggeration 43. 50 Mm 3 a À1 , or evaporation losses from the Calama Basin, which are estimated to be over 150 Mm 3 a À1 (Houston 2006a, b).

Seismicity
Many of the soft-sediment deformational structures, together with slope failures and faulting, require a trigger mechanism for their initiation. Such triggers are likely to involve near instantaneous excess loading and the most likely candidate for such a shock is seismic activity. A wide variety of structures have been interpreted to be initiated by seismic shock: these include load structures, including ball and pillow structures in carbonate sediments (Neuwerth et al. 2006); water escape-and liquefaction-induced features such as dish and pillar and clastic dykes (Obermeier 1996); convolute bedding and brecciation (Bowman et al. 2004); and even tepee structures and syneresis (Pratt 1998(Pratt , 2002 have been suggested to have a seismic origin. Moretti et al. (1999) reproduced load structures, dish and pillars, convolute bedding, sand dykes and volcanoes through the liquefaction and fluidization of unlithified materials in the laboratory using a shaking table to simulate seismic shock.
In the northern Chilean Miocene forearc all these criteria may be applied to the observed soft-sediment structures in general, although not all such structures possess all criteria. Sudden formation is implied in the case of clastic dykes, slumps, faults and some breccias. Synchroneity is to some extent scale dependent, thus the Upper Miocene to Lower Pliocene sequences host many more structures than the underlying Oligocene to Middle Miocene rocks. Within the narrow time horizon implied by a series of earthquakes, however, insufficient evidence is available to demonstrate discrete synchroneity, except for the slumps and associated breccias at Puente Posada. Detailed mapping of single structures has not been attempted and thus evidence for a zoned distribution is limited to a subjective judgement that clastic dykes, disturbed laminated travertine and some breccias become more frequent adjacent to the known major fault zones (Fig. 11). All the structures herein described as soft-sediment deformation are within size limits displayed by similar features reported in the literature to be conclusively formed by seismic activity. Northern Chile has been the site of subduction since the Cretaceous and as such is a site with the most intense seismic activity. The sediments that host the deformation structures were largely saturated at the time of their formation and the ideal sediments for liquefaction appear to be well-sorted sands with median diameters in the range 0.02-2.0 mm (Tsuchida & Hayashi 1971), which is typical for some beds in the Jalquinche Formation (see Supplementary Publication; see p. 296). The results of liquefaction have now also been identified in sandy gravels and fine-grained sediments (Obermeier 1996) and carbonates (Onasch & Kahle 2002) similar to the Opache Formation. Seismicity may lead to increased surface water flows and rising groundwater levels (e.g. Montgomery & Manga 2003), and may have led to pulsed dewatering with significantly higher flow rates than the long-term averages calculated above.
In addition to soft-sediment deformation initiated by seismic shock, slopes with factors of safety greater than 1.5 require a seismic trigger for failure (Jibson & Keefer 1993), thus suggesting that the majority of slope failures along the incised rivers were triggered by seismic activity. Neogene activity on all the major fault systems in the Chilean forearc has been previously reported; for example, extensional normal displacements on the Atacama Fault Zone (Dewey & Lamb 1992;González et al. 2006) and the West Fissure system (Tomlinson et al. 2001). Contractional reverse displacements have taken place along the Precordillera Fault Zone (Audin et al. 2003;Victor et al. 2004). All the deformation structures reported are within a few tens of kilometres of these major fault systems and would thus be capable of being triggered by earthquakes on these faults with moment magnitudes (M w ) of 5-7 and greater (Ambraseys 1988;Obermeier 1996).

Discussion
A schematic representation of the distribution of Neogene deformational structures in the northern Chilean forearc compared with basin subsidence history, shortening, volcanic activity, subduction erosion, and plate movement is given in Figure 11. A quantitative analysis of the structures described in space and time is difficult, as they vary across several orders of magnitude in size, intensity and host lithology. Nevertheless, the exceptional exposure and preservation potential in all Neogene formations allow generalizations to be made. The abundance of structures increased through the Miocene, reached a peak in the Late Miocene-Pliocene, and declined somewhat thereafter. The wide range and disposition of near syndepositional sedimentary deformation within Neogene strata is associated with saturated sediments in a plastic state, undergoing slow dewatering as a result of compaction and uplift. Many of the deformation structures may have been initiated by seismic shock. Spatially, the structures appear to concentrate within the subsiding basins and adjacent to major faults. Genetically, the higher frequency of load-induced deformation in the Lower Miocene-Pliocene rocks has already been noted. Dissolution features are most common in the Lower-Middle Miocene sequences, probably because of the longevity of groundwater circulation that continues today. Dewatering structures are especially abundant in the Upper Miocene-Pliocene sequences, with evidence of liquefaction occurring up to several kilometres from known faults. Pseudo-diapirism and basin boundary collapse features achieve maximum development in the Upper Miocene-Pliocene units of the Calama Basin, whereas shear-induced intra-formational slumps, slips and rotations are more frequent in the Calama Gap and especially in the Quillagua-Llamara Basin.
These relationships allow us to draw a number of conclusions about the development of forearc basins in this part of the Andes. Groundwater was widespread and abundant, with near-surface water levels as recently as 3.3 Ma. We speculate that this groundwater derived from Amazonian sourced precipitation over the Western Cordillera, as it does today (Houston 2006a), albeit at possibly higher rates as a result of either a wetter climate or lower topography, or both. At 3.3 Ma or soon after, uplift generated base-level revision. This caused incision of the main rivers and groundwater drainage of strata near ground surface and adjacent to the developing river valleys. Hence, many tributary systems were left 'hanging' and flow in them led to basin margin infiltration rather than discharge to the surface water network. This hydraulic system still operates today (Houston 2006b). As dewatering proceeded, the sediments underwent progressive consolidation and lithification, converting from a plastic to a brittle state, which prohibited subsequent 'softsediment' deformation. The amount of water expelled as a result of compaction was not enough to generate significant river flow, nor maintain palustrine-lacustrine conditions. Furthermore, the base-level revision that caused main river incision and groundwater drainage led to the cessation of basinal lacustrine deposition and sedimentary bypass. Hence, this scenario does not specifically require Pliocene climate change to be invoked as a cause for the cessation of sedimentary deposition (Hartley & Chong 2001). Seismicity, as manifested by deformation that is likely to require a sudden shock for its formation (some load structures, water escape structures, some breccias and disrupted, convolute bedding), appears to become more common, or more effective with time. On the other hand, the comparative lack of manifested seismicity in the Pleistocene could be explained as a result of the sediments being in a less susceptible (dewatered, brittle, and partially lithified) state. Recent deformation associated with seismic activity has been widely documented (Delouis et al. 1998;Audemard et al. 2005), but apart from faulting, there is no previous record of ancient seismic activity in the forearc.  (Table 1), forearc shortening (based on an analysis of reverse faulting in the Precordillera by Victor et al. 2004), volcanic activity (Trumbull et al. 2006), subduction erosion (Kukowski & Oncken 2006), recalculated for 21-238S (see Supplementary Publication; see p. 296), plate convergence at 228S (Somoza 1998), and South American drift (Silver et al. 1998).
Present-day earthquake frequency in the area between 21 and 238S, encompassing the Loa Basin, is 0.7 a À1 for M s .7, and 0.1 a À1 for M s .8 (Servicio Sismológico, Universidad de Chile; http://ssn.dgf.uchile.cl/home/terrem.html), and may generate ground motions up to 0.3g (Keefer & Moseley 2004), which would provide ample opportunity for soft-sediment and other related structures to develop.
Overwhelmingly, most earthquakes in the central Andes originate from the Wadati-Benioff Zone (ANCORP Working Group 2003), where subduction erosion takes place. The changing rate of subduction erosion through time might therefore correlate with seismic activity. The increasing incidence of sedimentary deformation structures is mirrored by the increasing subduction erosion throughout the Neogene (Fig. 11), except for the Pleistocene, where the lack of structures has been suggested to be due to the lack of sediment susceptibility. If this correlation is in fact a causal mechanism then the current level of seismic activity may be higher than during the Neogene.
The recurrence of deformational structures in time also correlates well with volcanic frequency in the Western Cordillera. The structures, however, show no tendency to migrate towards the Cordillera over time, and earthquakes associated with volcanism generally result in lower magnitudes and less extensive ground shaking than those associated with subduction (Zorbin 2003). Therefore, volcanic seismicity is not considered likely to be a major initiator for the structures described.
The increasing quantity of all structures does not correlate with basin subsidence, shortening or plate convergence rates. Significant early basin subsidence in the Middle Miocene might be a reflection of higher shortening rates that could have given rise to basin boundary uplift rather than basin subsidence as a reason for the increased accommodation space. In this respect, Neogene reverse faulting on the Precordillera Fault Zone has been confirmed for the Quillagua-Llamara Basin (Victor et al. 2004), but not for the Calama Basin, where most faulting is generally normal (or transtensional) and extensionally generated (Hartley et al. 2000;Pananont et al. 2004). Decreasing subsidence and shortening rates since the Miocene may also correlate with decreasing plate convergence rates resulting in reduced basin boundary uplift and accommodation space with time. Thus basin development can be linked to plate convergence and uplift but the deformational structures do not appear to be linked to basin development and therefore neither to plate convergence nor uplift. This reinforces the need for other factors such as seismicity to be involved in their generation and hence the possibility of subduction erosion as an ultimate cause.
A distinct group of medium-to large-scale deformation structures require extension coupled with shear stress for their formation: pseudo-diapirism and basin boundary collapse in Calama Basin, and intra-formational slumps and slides in the Calama Gap and Quillagua-Llamara Basin. The origin of this shear stress is likely to be found associated with gentle tilting across the forearc associated with uplift of the Western Cordillera, including localized areas of extension such as the Calama Basin and flexural stress generated by movement on major faults. The concentration of this group of structures in the Upper Miocene to Pliocene units is probably due to far-field tilting of the western flank of the Altiplano, estimated to be 2.48 between 10 and 7 Ma and 1.68 since 7 Ma (Nester et al. 2006). The series of normal faults affecting the Opache Formation in the Calama Gap may also be related to this late-stage flexural stress. Restoration of the pre-incision middle Pliocene (c. 3.3 Ma) surface of the Río Loa valley (see Supplementary Publication; see p. 296) compared with present-day slopes provides an estimate of post 3.3 Ma tilting that amounts to less than 18, suggesting that such rotation had largely ceased by then. Consistent with this is the fact that slumps in the Pleistocene sediments are not oriented down-dip of the regional slope but are largely associated with side-wall collapse of incised river valleys.

Conclusion
Deformational structures within the Neogene sediments provide evidence of both proximate and ultimate causal mechanisms that may be integrated over time to provide new perspectives on basin development and tectonics. Previously defined tectonic episodes are generally based on unconformities, a cessation of sedimentation and/or structural deformation. However, the occurrence of deformation within the sedimentary record points to greater continuity of stress and strain. It is therefore postulated that seismic activity as a result of subduction processes is the mechanism controlling many synsedimentary structures. Basin subsidence rates are not correlated with syndeformational structures and therefore are not considered to be a major controlling feature, although compaction and dewatering were clearly involved in many structures. Tilting across the full width of the forearc during the Late Miocene to Pliocene generated shear stresses in saturated sediments that contributed to slumping and sliding. River incision as a result of base-level revision during the Pliocene facilitated groundwater drainage and resulted in sedimentary bypass and the cessation of deposition that is therefore not necessarily due to climate change. Dewatering also led to a transition in the character of deformation from plastic to brittle. The structures imply both contractional and extensional regimes across the forearc through the Neogene, although almost all the larger structures are extensional in origin, supporting the concept that upper crustal tectonic activity is at least partially decoupled from plate movement.
Funding for this study was provided by Nazca S.A. The authors are grateful for a review of an early draft of this paper by A. Hartley, and his helpful comments and suggestions that greatly improved its contents. T. Debacker and G. May also provided valuable reviews that helped to clarify the ideas expressed.