The tectonic setting of the convergent margin of South America is controlled by subduction of the oceanic Nazca plate beneath the South American continental plate. The convergence rate varies between 66 mm yr-1 in the north (8∘ S latitude) and 74 mm yr-1 in the south (27∘ S latitude) (Fig. 1). The convergence azimuth changes slightly from 81.7∘ N toward 77.5∘ N from north to south (DeMets et al., 2010). The South American subduction zone is divided into four major segments with variable subduction angles inferred from the spatial distribution of Benioff seismicity (Barazangi and Isacks, 1976; Jordan et al., 1983) (Fig. 1). The segments beneath northern and central Peru (2–15∘ S) and beneath central Chile (27–33∘ S) are characterized by a gentle dip of the subducting plate between 5 and 10∘ at depths of ∼ 100 km (Hayes et al., 2018), whereas the segments beneath southern Peru and northern Chile (15–27∘ S) and beneath southern Chile (33–45∘ S) have steeper dips of 25–30∘. Spatial distributions of earthquakes furthermore indicate a steep-slab subduction segment in Ecuador and southern Colombia (2∘ S to 5∘ N) and a flat-slab segment in NW Colombia (north of 5∘ N) (Pilger, 1981; Cahill and Isacks, 1992; Gutscher et al., 2000; Ramos and Folguera, 2009). Processes that have been inferred to be responsible for the shallowing of the subduction slab include the subduction of large buoyant ridges or plateaus (Espurt et al., 2008), as well as the combination of the trenchward motion of thick, buoyant continental lithosphere accompanied by trench retreat (Sobolev and Babeyko, 2005; Manea et al., 2012). Volcanic activity, as well as the forearc architecture and distribution of upper-plate deformation, further emphasizes the location of flat-slab subduction segments (Jordan et al., 1983; Kay et al., 1987; Ramos and Folguera, 2009).

Several high bathymetric features have been recognized on the subducting Nazca plate. The two most prominent bathymetric features being subducted beneath South America are the Carnegie and Nazca aseismic ridges at 0 and 15∘ S, respectively; they consist of seamounts related to hot-spot volcanism (Gutscher et al., 1999; Hampel, 2002). The 300 km wide and ∼ 2 km high Carnegie Ridge subducts roughly parallel to the convergence direction, and its geometry should have remained relatively stable beneath the continental plate (Angermann et al., 1999; Gutscher et al., 1999; DeMets et al., 2010; Martinod et al., 2016a). In contrast, the obliquity of the 200 km wide and 1.5 km high Nazca Ridge with respect to the convergence direction resulted in 500 km SE-directed migration of its locus of ridge subduction during the last 10 Myr (Hampel, 2002; Saillard et al., 2011; Martinod et al., 2016a). Similarly, smaller aseismic ridges such as the Juan Fernández Ridge and the Iquique Ridge subduct beneath the South American continent at 32 and 21∘ S, respectively. The intercepts between these bathymetric anomalies and the upper plate are thought to influence the characteristics of interplate coupling and seismic rupture (Bilek et al., 2003; Wang and Bilek, 2011; Geersen et al., 2015; Collot et al., 2017) and mark the boundaries between flat and steep subduction segments and changes between subduction erosion and accretion (Jordan et al., 1983; von Huene et al., 1997; Ramos and Folguera, 2009) (Fig. 1).

In addition to bathymetric anomalies, several studies have shown that variations in the volume of sediments in the trench may control the subduction regime from an erosional mode to an accretionary mode (von Huene and Scholl, 1991; Bangs and Cande, 1997). In addition, the volume of sediment in the trench has also been hypothesized to influence the style of interplate seismicity (Lamb and Davis, 2003). At the southern Chilean margin, thick trench-sediment sequences and a steeper subduction angle correlate primarily with subduction accretion, although the area of the intercept of the continental plate with the Chile Rise spreading center locally exhibits the opposite case (von Huene and Scholl, 1991; Bangs and Cande, 1997). Subduction erosion characterizes the region north of the southern volcanic zone from central and northern Chile to southern Peru (33–15∘ S) due to decreasing sediment supply to the trench, especially within the flat-slab subduction segments (Stern, 1991; von Huene and Scholl, 1991; Bangs and Cande, 1997; Clift and Vannucchi, 2004). Clift and Hartley (2007) and Lohrmann et al. (2003) argued for an alternate style of slow tectonic erosion leading to underplating of subducted material below the base of the crustal forearc, which is synchronous with tectonic erosion beneath the trenchward part of the forearc. For the northern Andes, several authors also classify the subduction zone as an erosional type (Clift and Vannucchi, 2004; Scholl and Huene, 2007; Marcaillou et al., 2016).

The South American convergent margin comprises several fault systems with different kinematics whose presence is closely linked to oblique subduction and the motion and deformation of forearc slivers. Here we summarize the main structures that affect the Pacific coastal areas. North of the Talara bend (5∘ S), active thrusting and dextral strike-slip faulting dominate the coastal lowlands of Ecuador (e.g., Mache, Bahía, Jipijapa faults), although normal faulting also occurs at Punta Galera (Cumilínche fault) and the Manta Peninsula (Río Salado fault) (Fig. 1). Farther south, normal faulting is active in the Gulf of Guayaquil (Posorja fault), and dextral strike-slip faulting occurs at the Santa Elena Peninsula (La Cruz fault) (Veloza et al., 2012; Costa et al., 2020). The most prominent dextral fault in this region is the 2000 km long, NE-striking Dolores–Guayaquil megashear (DGM) which starts in the Gulf of Guayaquil and terminates in the Colombian hinterland east of the range-bounding thrust faults of the Colombian Andes (Veloza et al., 2012; Villegas-Lanza et al., 2016; Costa et al., 2020) (Fig. 1). Normal faults have been described along the coast of Peru at the Illescas Peninsula in the north (6∘ S), within the El Huevo–Lomas fault system in the San Juan de Marcona area (14.5–16∘ S), and within the Incapuquio fault system farther south (17–18∘ S) (Veloza et al., 2012; Villegas-Lanza et al., 2016; Costa et al., 2020). The main fault zones of the Chilean convergent margin comprise the Atacama fault system (AFS) in the Coastal Cordillera extending from Iquique to La Serena (29.75∘ S; Fig. 1) with predominantly north–south-striking normal faults, which result in the relative uplift of their western side (e.g., Mejillones fault, Salar del Carmen fault) (Naranjo, 1987; González and Carrizo, 2003; Cembrano et al., 2007). Coastal fault systems farther south are located in the Altos de Talinay area (30.5∘ S; Puerto Aldea fault), near Valparaíso (33∘ S; Quintay and Valparaíso faults), near the Arauco Peninsula (36–39∘ S; Santa María and Lanalhue faults), and in between these areas (Topocalma, Pichilemu, Carranza, and Pelluhue faults) (Ota et al., 1995; Melnick et al., 2009, 2020; Santibáñez et al., 2019; Maldonado et al., 2021) (Fig. 1). However, there is still limited knowledge regarding Quaternary slip rates and kinematics and, most importantly, the location of active faults along the forearc region of South America (Jara-Muñoz et al., 2018; Melnick et al., 2019).

The 8000 km long Andean orogen is a major, hemisphere-scale feature that can be divided into segments with distinctive geomorphic and tectonic characteristics. The principal segments comprise the NNE–SSW-trending Colombian–Ecuadorian segment (12∘ N–5∘ S), the NW–SE-oriented Peruvian segment (5–18∘ S), and the north–south-trending Chilean segment (18–56∘ S) (Jaillard et al., 2000) (Fig. 1). Two major breaks separate these segments; these are the Huancabamba bend in northern Peru and the Arica bend at the Peru–Chile border. The distance of the trench from the WSAC coastline averages 118 km and ranges between 44 and 217 km. The depth of the trench varies between 2920 and 8177 m (GEBCO Bathymetric Compilation Group, 2020), and the continental shelf has an average width of 28 km (Paris et al., 2016).

In the 50 to 180 km wide coastal area of the Ecuadorian Andes, where the Western Cordillera is flanked by a structural depression, relief is relatively low (< 300 m a.s.l.). The Gulf of Guayaquil (3∘ S) and the Dolores–Guayaquil megashear separate the northern from the southern forearc units. The coast-trench distance along the Huancabamba bend is quite small (∼ 55–90 km), except for the Gulf of Guayaquil, and the trench east of the Carnegie Ridge is at a relatively shallow depth of ∼ 3.5 km. Farther south, the Peruvian forearc comprises the up to 160 km wide Coastal Plains in the north and the narrow, 3000 m high Western Cordillera. While the Coastal Plains in north-central Peru are relatively narrow (< 40 km), they widen in southern Peru, and the elevation of the Western Cordillera increases to more than 5000 m (Suárez et al., 1983; Jaillard et al., 2000). The region between the coast and the trench in central Peru (up to 220 km) narrows toward the San Juan de Marcona area (∼ 75 km) near the intercept with the Nazca Ridge, and the relatively deep trench (∼ 6.5 km) becomes shallower (< 5 km) (GEBCO Bathymetric Compilation Group, 2020). Between 18 and 28∘ S, the Chilean forearc comprises the 50 km wide and up to 2700 m high Coastal Cordillera, which is separated from the Precordillera by the Central Depression. In the flat-slab subduction segment between 27 and 33∘ S there is neither a morphotectonic region characterized by a central depression nor active volcanism in the high Andean cordillera (Fig. 1) (Jordan et al., 1983). The Chilean forearc comprises the Coastal Cordillera, which varies in altitude from up to 2000 m at 33∘ S to 500 m at 46∘ S, and the Central Depression that separates the forearc from the Main Cordillera. From the Arica bend, where the coast-trench distance is up to 170 km and the trench ∼ 8 km deep, a slight increase in coast-trench distance can be observed in Chile toward the south (∼ 80–130 km), as can a decrease in trench depth to ∼ 4.5 km.

Wave erosion generates wave-cut terrace levels, while the accumulation of shallow marine sediments during sea-level highstands forms wave-built terraces. Another type of terrace is known as “rasa” and refers to wide shore platforms formed under slow-uplift conditions (< 0.2 m kyr-1) and the repeated reoccupation of this surface by high sea levels (Regard et al., 2010; Rodríguez et al., 2013; Melnick, 2016). Other studies indicate a stronger influence of climate and rock resistance to erosion compared to marine wave action (Prémaillon et al., 2018). Typically, the formation of Pleistocene marine terraces in the study area occurred during interglacial and interstadial relative sea-level highstands that were superposed on the uplifting coastal areas; according to the Quaternary oxygen-isotope curve defining warm and cold periods, high Quaternary sea levels have been correlated with warm periods and are denoted with the odd-numbered Marine Isotope Stages (MISs) (Lajoie, 1986; Shackleton et al., 2003).

Along the WSAC, staircase-like sequences of multiple marine terraces are preserved nearly continuously along the coast. These terraces comprise primarily wave-cut surfaces that are frequently covered by beach ridges of siliciclastic sediments and local accumulations of carbonate bioclastic materials (Ota et al., 1995; Saillard et al., 2009; Rodríguez et al., 2013; Martinod et al., 2016b). Rasa surfaces exist in the regions of southern Peru and northern Chile (Regard et al., 2010; Rodríguez et al., 2013; Melnick, 2016). Particularly the well-preserved MIS 5e terrace level has been largely used as a strain marker in the correlation of uplifted coastal sectors due to its lateral continuity and high potential for preservation. Global observations of sea-level fluctuations during MIS 5 allow us to differentiate between three second-order highstands at 80 ka (5a), 105 ka (5c), and 128 to 116 ka (5e) with paleo sea levels of -20 m for both of the younger and +3±3 m for the oldest highstand (Stirling et al., 1998; Siddall et al., 2006; Hearty et al., 2007; Rohling et al., 2009; Pedoja et al., 2011), although glacio-isostatic adjustments (GIAs) can cause local differences of up to 30 m (Simms et al., 2016; Creveling et al., 2017). The database generated in this study is based exclusively on the last interglacial marine terraces exposed along the WSAC between Ecuador and southern Chile (1–40∘ S). In the following section we present a brief review of previously studied marine terrace sites in this area.

Paleo-shoreline elevations of the last interglacial (MIS 5e) in Ecuador are found at elevations of around 45 ± 2 m a.s.l. in Punta Galera (Esmeraldas area), 43–57 ± 2 m on the Manta Peninsula and La Plata Island, and 15±5 m a.s.l. on the Santa Elena Peninsula (Pedoja et al., 2006b, a). In northern Peru, MIS 5e terraces have been described at elevations of 18–31 m a.s.l. for the Tablazo Lobitos (Cancas and Mancora areas), at 25 ± 5 m a.s.l. on the Paita Peninsula, and at 18 ± 3 m a.s.l. on the Illescas Peninsula and the Bay of Bayovar (Pedoja et al., 2006b). Farther south, MIS 5e terraces are exceptionally high in the San Juan de Marcona area immediately south of the subducting Nazca Ridge, with maximum elevations of 80 m at the Cerro Trés Hermanas and 105 m at the Cerro El Huevo (Hsu et al., 1989; Ortlieb and Macharé, 1990; Saillard et al., 2011). The Pampa del Palo region in southern Peru exhibits relatively thick vertical stacks of shallow marine terrace deposits related to MIS 7, 5e (∼ 20 m), and 5c that may indicate a different geodynamic behavior compared to adjacent regions (Ortlieb et al., 1996b). In central and northern Chile, the terrace levels of the last interglacial occur at 250–400, 150–240, 80–130, and 30–40 m and in southern Chile at 170–200, 70, 20–38, and 8–10 m (Fuenzalida et al., 1965). Specifically, between 24 and 32∘ S, paleo-shoreline elevations of the last interglacial (MIS 5e) range between 25 and 45 m (Ota et al., 1995; Saillard et al., 2009; Martinod et al., 2016b). Shore platforms are higher in the Altos de Talinay area (30.3–31.3∘ S) but are small, poorly preserved, and terminate at a high coastal scarp between 26.75 and 24∘ S (Martinod et al., 2016b). Shoreline-angle elevations between 34 and 38∘ S (along the Maule seismotectonic segment) vary from high altitudes in the Arauco and Topocalma areas (200 m) to moderate elevations near Carranza (110 m) and to very low elevations in between (15 m) (Melnick et al., 2009; Jara-Muñoz et al., 2015).

Coastal uplift-rate estimates along the WSAC mainly comprise calculations for the Talara Arc, the San Juan de Marcona area, the Mejillones Peninsula, the Altos de Talinay area, and several regions in south-central Chile. Along the Talara Arc (6.5∘ S–1∘ N), marine terraces of the Manta Peninsula and La Plata Island in central Ecuador indicate the most pronounced uplift rates of 0.31–0.42 m kyr-1 since MIS 5e, while similar uplift rates are documented to the north in the Esmeraldas area (0.34 m kyr-1) and lower ones to the south at the Santa Elena Peninsula (0.1 m kyr-1). In northern Peru, last interglacial uplift rates are relative low, ranging from 0.17–0.21 m kyr-1 for the Tablazo Lobitos and 0.16 m kyr-1 for the Paita Peninsula to 0.12 m kyr-1 for the Bay of Bayovar and the Illescas Peninsula (Pedoja et al., 2006b, a). Marine terraces on the continental plate above the subducting Nazca Ridge (13.5–15.6∘ S) record variations in uplift rate where the coastal forearc above the northern flank of the ridge is either stable or has undergone net subsidence (Macharé and Ortlieb, 1992). The coast above the ridge crest has been rising at about 0.3 m kyr-1 and the coast above the southern flank (San Juan de Marcona) has been uplifting at a rate of 0.5 m kyr-1 (Hsu, 1992) or even 0.7 m kyr-1 (Ortlieb and Macharé, 1990) for at least the last 125 kyr. Saillard et al. (2011) state that long-term regional uplift in the San Juan de Marcona area has increased since about 800 ka related to the southward migration of the Nazca Ridge, and it ranges from 0.44 to 0.87 m kyr-1. The Pampa del Palo area in southern Peru rose more slowly or was even down-faulted and had subsided with respect to the adjacent coastal regions (Ortlieb et al., 1996b). These movements ceased after the highstand during the MIS 5e, and slow-uplift rates of approximately 0.16 m kyr-1 have characterized the region since 100 ka (Ortlieb et al., 1996b). In northern Chile (24–32∘ S), uplift rates for the Late Pleistocene average around 0.28 ± 0.15 m kyr-1 (Martinod et al., 2016b), except for the Altos de Talinay area where pulses of rapid uplift occurred during the Middle Pleistocene (Ota et al., 1995; Saillard et al., 2009; Martinod et al., 2016b). The Central Andean rasa (15–33∘ S) and Lower to Middle Pleistocene shore platforms – which are also generally wider – indicate a period of tectonic stability or subsidence followed by accelerated and spatially continuous uplift after ∼ 400 ka (MIS 11) (Regard et al., 2010; Rodríguez et al., 2013; Martinod et al., 2016b). However, according to Melnick (2016), the central Andean rasa has experienced slow and steady long-term uplift with a rate of 0.13 ± 0.04 m kyr-1 during the Quaternary, predominantly accumulating strain through deep earthquakes at the crust–mantle boundary (Moho) below the locked portion of the plate interface. The lowest uplift rates occur at the Arica bend and increase gradually southward; the highest values are attained along geomorphically distinct peninsulas (Melnick, 2016). In the Maule segment (34–38∘ S), the mean uplift rate for the MIS 5 terrace level is 0.5 m kyr-1, exceeded only in the areas of Topocalma, Carranza, and Arauco, where it amounts to 1.6 m kyr-1 (Melnick et al., 2009; Jara-Muñoz et al., 2015). Although there are several studies of marine terraces along the WSAC, these are isolated and based on different methodological approaches, mapping and leveling resolutions, and dating techniques, which make regional comparisons and correlations difficult in the context of the data presented here.

Apart from latitudinal temperature changes, the present-day morphotectonic provinces along the South American margin have a pronounced impact on the precipitation gradients on the west coast of South America. Since mountain ranges are oriented approximately perpendicular to moisture-bearing winds, they affect both flanks of the orogen (Strecker et al., 2007). The regional-scale pattern of wind circulation is dominated by westerly winds at subtropical and extratropical latitudes primarily up to about 27∘ S (Garreaud, 2009). However, anticyclones over the South Pacific result in winds blowing from the south along the coast between 35∘ S and 10∘ S (Garreaud, 2009). The moisture in the equatorial Andes (Ecuador and Colombia) and in the areas farther south (27∘ S) is fed by winds from the Amazon basin and the Gulf of Panama, resulting in rainfall mainly on the eastern flanks of the mountain range (Bendix et al., 2006; Bookhagen and Strecker, 2008; Garreaud, 2009). The Andes of southern Ecuador, Peru, and northern Chile are dominated by a rain-shadow effect that causes aridity within the Andean Plateau (Altiplano-Puna), the Western Cordillera, and the coastal region (Houston and Hartley, 2003; Strecker et al., 2007; Garreaud, 2009). Furthermore, the aridity is exacerbated by the effects of the cold Humboldt current which prevents humidity from the Pacific from penetrating inland (Houston and Hartley, 2003; Garreaud, 2009; Coudurier-Curveur et al., 2015). The precipitation gradient reverses between 27 and 35∘ S where the Southern Hemisphere westerlies cause abundant rainfall on the western flanks of the Coastal and Main cordilleras (Garreaud, 2009). Martinod et al. (2016b) proposed that latitudinal differences in climate largely influence coastal morphology, specifically the formation of high coastal scarps that prevent the development of extensive marine terrace sequences. However, the details of this relationship have not been conclusively studied along the full extent of the Pacific coast of South America.

The MIS 5e terrace levels in Ecuador and northern Peru (sites Ec1 to Ec4 and Pe1) are discontinuously preserved along the coast (Fig. 6). They often occur at low elevations (between 12 and 30 m) and show abrupt local changes in elevation, reaching a maximum at the Manta Peninsula. Punta Galera in northern Ecuador displays relatively broad and well-preserved marine terraces ranging between 40 and 45 m elevation that rapidly decrease eastward to about 30 m a.s.l. across the Cumilínche fault (Ec1). Farther south, between Pedernales and Canoa (Ec1), narrow terraces occur at lower altitudes of 22–34 m a.s.l. A long-wavelength (∼ 120 km) pattern in terrace-elevation change can be observed across the Manta Peninsula with the highest MIS 5e terraces peaking at ∼ 100 m a.s.l. at its southern coast (Ec2). This terrace level is hardly visible in its highest areas with platform widths smaller than 100 m due to deeply incised and narrowly spaced river valleys. We observe lower and variable elevations between 30 and 50 m across the Rio Salado fault in the San Mateo paleo-gulf in the north, while the terrace elevations increase gradually from ∼ 40 m in the Pile paleo-gulf in the south (Ec3) toward the center of the peninsula (El Aromo dome) and the Montecristi fault (Ec3). A lower terrace level correlated to MIS 5c displays similar elevation patterns as MIS 5e within the Pile paleo-gulf and areas to the north. Near the Gulf of Guayaquil and the Dolores–Guayaquil megashear, the lowest terrace elevations occur at the Santa Elena Peninsula ranging between 17 and 24 m a.s.l., even lower altitudes in its southern part, and between 11 and 16 m a.s.l. on the Puna Island (Ec4). In northern Peru (Pe1), we observe dismembered MIS 5e terraces in the coastal area between Cancas and Talara below the prominent Mancora Tablazo. “Tablazo” is a local descriptive name used in northern Peru (∼ 3.5–6.5∘ S) for marine terraces that cover a particularly wide surface area (Pedoja et al., 2006b). South of Cancas, MIS 5e terrace elevations range between 17 and 20 m a.s.l., reaching 32 m near Organos, and vary between 20 and 29 m in the vicinity of Talara. In the southward continuation of the Talara harbor, the Talara Tablazo widens, with a lower marine terrace at about 23 m a.s.l. immediately north of Paita Peninsula reaching 30 m a.s.l. in the northern part of the peninsula. The last occurrence of well-preserved MIS 5e terraces in this sector exists at the Illescas Peninsula, where terrace elevations decrease from around 30 to 17 m a.s.l. southward.

This segment comprises marine terraces at relatively low and constant elevations but which are rather discontinuous (sites Pe2 to Pe10), except in the San Juan de Marcona area where the terraces increase in elevation drastically (Fig. 7). The coast in north-central Peru exhibits poor records of MIS 5e marine terraces characterized by mostly narrow and discontinuous remnants that are sparsely distributed along the margin with limited age constraints. Marine terraces increase in elevation from 11 to 35 m a.s.l. south of Chiclayo (Pe2) and decrease to 17 m a.s.l. near Cercado de Lima (Pe3, Pe4), forming a long-wavelength (∼ 600 km), small amplitude (∼ 20 m) upwarped structure. The MIS 5e terrace levels are better expressed in the south-central and southern part of Peru at elevations between 35 and 47 m a.s.l. in San Vicente de Cañete, decreasing to approximately 30 m a.s.l. in the vicinity of Pisco (Pe5). South of Pisco, the coastal area becomes narrow with terrace elevations ranging between 25 and 34 m a.s.l. (Pe6) and increasing abruptly to 74–79 m near Puerto Caballas and the Río Grande delta. MIS 5e terrace elevations are highest within the San Juan de Marcona area, reaching 93–109 m at Cerro Huevo and 56–87 m at Cerro Trés Hermanas (Pe7). These higher terrace elevations coincide with a wider coastal area, a better-preserved terrace sequence, and several crustal faults, such as the San Juan and El Huevo faults.

Terrace heights west of Yauca indicate a further decrease to 50–58 m before a renewed increase to 70–72 m can be observed in the Chala embayment (Pe8). We observe a similar trend in elevation changes for the shoreline angles attributed to the MIS 5c interglacial within the previously described high-elevation area: 31–39 m near the Río Grande delta, 58–62 m below the Cerro Huevo peak, 27–64 m below the Cerro Trés Hermanas peak (Pe7), 36–40 m near Yauca, and 34–40 m within the Chala embayment (Pe8). Besides various changes in between, terrace elevations decrease slowly from 54 m south of the Chala region to 38 m near Atico (Pe8). The overall decrease south of the San Juan de Marcona area therefore contrasts strikingly with the sharper decrease to the north. These high-elevation marine terraces, which extend ∼ 250 km along the coast from north of the San Juan de Marcona area to south of Chala Bay, constitute one of the longest wavelength structures of the WSAC. SE of Atico, less well-preserved marine terraces appear again in the form of small remnants in a narrower coastal area. Starting with elevations as low as 24 m, MIS 5e terrace altitudes increase southeastward to up to 40 m near Mollendo (Pe9) before they slightly decrease again. The broader and quite well-preserved terraces of the adjacent Ilo area resulted in a smooth increase from values greater than 25 to 33 m and a sudden decrease to as low as 22 m across the Chololo fault (Pe9). North of the Arica bend, shoreline-angle measurements yielded estimates of 24–29 m in altitude (Pe10).

Along the northern Chilean coast, marine terraces of the MIS 5e are characterized by a variable elevation pattern and the occurrence of numerous crustal faults associated with the Atacama fault system, although the changes in terrace elevation are not as pronounced as in the northern segments (Fig. 8) (sites Ch1 to Ch7). The local widening of the coastal area near the Arica bend narrows southward with MIS 5e terraces at elevations of between 24 and 28 m a.s.l. in northernmost Chile (Ch1). Just north of Pisagua, we measured shoreline-angle elevations of well-preserved marine terraces between 19 and 26 m across the Atajana fault (Ch1). An areally limited zigzag pattern starting with shoreline-angle elevation values of 32 m south of Iquique and south of the Zofri and Cavancha faults decreases rapidly to approximately 22 m but increases again to similar altitudes and drops as low as 18 m toward Chanabaya south of the Barranco Alto fault (Ch1). A gentle, steady rise in terrace elevations can be observed south of Tocopilla where altitudes of 25 m are attained. South of Gatico, terrace markers of the MIS 5e highstand increase and continue northward for much of the Mejillones Peninsula within an approximate elevation range of 32–50 m a.s.l. before reaching a maximum of 62 m a.s.l. at the Pampa de Mejillones (Ch2). With its ∼ 100 km latitudinal extent, we consider this terrace-elevation change to be a medium-wavelength structure. Although no MIS 5e terrace levels have been preserved at the Morro Mejillones Horst (Binnie et al., 2016), we measured shoreline-angle elevations at the elevated SW part of the peninsula that decrease sharply from 55 to 17 m a.s.l. in the vicinity of the Mejillones fault system (Ch2). After a short interruption of the MIS 5e terrace level at Pampa Aeropuerto, elevations remain relatively low between 19 and 25 m farther south (Ch2). Along the ∼ 300 km coastal stretch south of Mejillones, marine terraces are scattered along the narrow coastal area ranging between 25 and 37 m a.s.l. (Ch3). South of Puerto Flamenco, MIS 5e terrace elevations range between 40 and 45 m a.s.l. until Caldera and Bahía Inglesa (Ch4). The MIS 5e marine terrace elevations decrease abruptly south of the Caldera fault and the Morro Copiapó (Morro Copiapó fault) to between 25 and 33 m a.s.l., reaching 20 m a.s.l. north of Carrizal Bajo (Ch4). In the southernmost part of the northern Chilean sector, the MIS 5e terraces rise from around 30 m a.s.l. to a maximum of 45 m a.s.l. near the Cabo Leones fault (Ch4) before decreasing in elevation abruptly near Caleta Chañaral and Punta Choros (Ch5, Ch6, Ch7).

Marine terraces along central Chile display variable, high-amplitude terrace-elevation patterns associated with numerous crustal faults and include a broad-scale change in terrace altitudes with the highest MIS 5e marine terrace elevations of the entire South American margin on the Arauco Peninsula (Fig. 9) (sites Ch8 to Ch78). South of Punta Choros, marine terrace elevations decrease from values close to 40 to 22 m a.s.l. north of Punta Teatinos (Ch8, Ch9). A maximum elevation of 40 m is reached by the terraces just south of this area (Ch10), whereas north of La Serena, a sharp decrease leads to values between 20 and 30 m for marine terraces south of Coquimbo Bay and in the Tongoy Bay area (Ch11, Ch12). South of Punta Lengua de Vaca, our measurements of the exceptionally well-preserved staircase morphology of the terraces are within the same elevation range between 20 and 30 m, increasing slowly to 40 m near the Quebrada el Moray (Ch13). Although we could not observe a significant change in terrace elevation across the Puerto Aldea fault, we measured an offset of ∼ 7 m across the Quebrada Palo Cortado fault. MIS 5e terrace levels decrease thereafter and vary between 20 and 30 m in altitude until north of Los Vilos (Ch14–Ch18) where they increase in elevation (Ch19), reaching 60 m near the Quilimari River (Ch20). The marine terraces become wider in this area and are associated with scattered sea stacks. Decreasing farther south to only 20 m a.s.l. (Ch21–Ch25), the coastal area narrows and has terrace heights of up to 64 m near Valparaíso in an area that is cut by numerous faults (e.g., Valparaíso and Quintay faults) (Ch26–Ch32). Another low-elevation area follows southward, with values as low as 17 m (Ch33–Ch35). Farther south, between 34 and 38∘ S, broad (∼ 200 km at Arauco), medium (∼ 45 km at Topocalma), and narrow (Carranza) upwarped zones occur that are manifested by variable terrace elevations. These include prominent high-terrace elevations at Topocalma with a maximum of 180 m (Ch36–Ch39), slightly lower levels of 110 m at Carranza (Ch42–Ch47), exceptionally low values near the Itata River (< 10 m) (Ch48–Ch64, Ch66), and the most extensive and highest shoreline-angle elevations on the Arauco Peninsula with elevations in excess of 200 m (Ch67–Ch73). Additionally, we measured MIS 5c terrace elevations in the three higher exposed areas with a range of 20–55 m at Carranza and a few locations at Topocalma (76–81 m) and Arauco (117–123 m). The medium-wavelength structure of Topocalma is bounded by the Pichilemu and Topocalma faults, and near Carranza several fault offsets (e.g., Pelluhue and Carranza faults) are responsible for the short-wavelength changes in terrace elevation. In contrast, crustal faulting is nearly absent in the high-elevation and long-wavelength structure at Arauco. MIS 5e terrace elevations are highly variable within a small area south of the Arauco Peninsula near the Tirúa fault, increasing rapidly from 27 to 78 m and decreasing thereafter to approximately 20 m (Ch74, Ch75). The continuity of terraces is interrupted by the absence of terrace levels between Imperial River and Toltén River but resumes afterward with a highly frequent zigzag pattern and multiple faults (e.g., Estero Ralicura and Curinanco faults) from as low as 18 m to a maximum of 40 m (Ch76, Ch77). In this area locations with the highest terrace levels comprise the terraces near Mehuín and Calfuco. A final increase in shoreline-angle elevations from about 20–30 m up to 76 m near Valdivia coincides with the southern terminus of our terrace-elevation measurements (Ch78).

The spatial distribution of the MIS 5 marine terrace elevations along the convergent South American margin has revealed several high-amplitude and long-wavelength changes with respect to tectonically controlled topography. Long-wavelength patterns in terrace elevation (∼ 102 km) are observed at the Manta Peninsula in Ecuador, central Peru between Chiclayo and Lima, San Juan de Marcona (Peru), and on the Arauco Peninsula in Chile, while medium-wavelength structures occur at Mejillones Peninsula and Topocalma (Chile). Instead, short-wavelength patterns in MIS 5 terrace elevations are observed, for instance, near Los Vilos, Valparaíso, and Carranza in Chile.

The subduction of bathymetric anomalies has been shown to exert a substantial influence on upper-plate deformation (Fryer and Smoot, 1985; Taylor et al., 1987; Macharé and Ortlieb, 1992; Cloos and Shreve, 1996; Gardner et al., 2013; Wang and Bilek, 2014; Ruh et al., 2016), resulting in temporally and spatially variable fault activity, kinematics, and deformation rates (Mann et al., 1998; Saillard et al., 2011; Morgan and Bangs, 2017; Melnick et al., 2019). When comparing the uplift pattern of MIS 5 marine terraces and the bathymetry of the oceanic plate, we observe that the two long-wavelength structures in this area, on the Manta Peninsula and at San Juan de Marcona, both coincide with the location of the subducting Nazca and Carnegie ridges, respectively (Fig. 11a and b); this was also previously observed by other authors (Macharé and Ortlieb, 1992; Gutscher et al., 1999; Pedoja et al., 2006a; Saillard et al., 2011). In summary, long-wavelength structures in coastal areas of the upper plate may be associated with deep-seated processes (Melosh and Raefsky, 1980; Watts and Daly, 1981) possibly related to changes in the mechanical behavior of the plate interface. In this context it is interesting that the high uplift rates on the Arauco Peninsula do not correlate with bathymetric anomalies, which may suggest a different deformation mechanism. The scarcity of crustal faults described in the Arauco area rather suggests that shallow structures associated with crustal bending and splay faults occasionally breaching through the upper crust (Melnick et al., 2012; Jara-Muñoz et al., 2015; Jara-Muñoz et al., 2017; Melnick et al., 2019) may cause long-wavelength warping and uplift there (Fig. 11a).

In contrast, small-scale bathymetric anomalies correlate in part with the presence of crustal faults perpendicular to the coastal margin near, for instance, the Juan Fernandez, Taltal, and Copiapó ridges (Fig. 11b); this results in short-wavelength structures and a more localized altitudinal differentiation of uplifted terraces. This emphasizes also the importance of last interglacial marine terraces as strain markers with respect to currently active faults, which might be compared in the future with short-term deformation estimates from GPS or the earthquake catalog. In summary, short-wavelength structures in the coastal realms of western South America may be associated with faults that root at shallow depths within the continental crust (Jara-Muñoz et al., 2015, 2017; Melnick et al., 2019).

The thickness of sediment in the trench is an additional controlling factor on forearc architecture that may determine which areas of the continental margin are subjected to subduction erosion or accretion (Hilde, 1983; Cloos and Shreve, 1988; Menant et al., 2020). Our data show that the accretionary part of the WSAC (south of the intersection with the Juan Fernandez Ridge at 32.9∘ S) displays faster median uplift rates of 0.26 m kyr-1 than in the rest of the WSAC (Fig. 11b and c). However, no clear correlation is observed between trench fill, uplift rates, and the different structural patterns in the erosive part of the margin. On the other hand, we observe lower uplift rates for greater distances from the trench at the Arica bend, in central Peru, and in the Gulf of Guayaquil, while higher uplift rates occur in areas closer to the trench, such as near the Nazca and Carnegie ridges and the Mejillones Peninsula.

The latitudinal climate differences that characterize the western margin of South America may also control coastal morphology and the generation and preservation of marine terraces (Martinod et al., 2016b). In order to evaluate the influence of climate in the generation and/or degradation of marine terraces, we compared the number of marine terrace measurements, which is a proxy for the degree of marine terrace preservation, and climatically controlled parameters such as wave height, tidal range, coastline orientation, and the amount of precipitation.

The maximum wave height along the WSAC decreases northward from ∼ 8 to ∼ 2 m (see Sect. 3.3, Fig. 11d). Similarly, the tidal range decreases progressively northward from 2 to 1 m between Valdivia and San Juan de Marcona, followed by a rapid increase to 4 m between San Juan de Marcona and the Manta Peninsula. We observe an apparent correlation between the number of measurements and the tidal range in the north, between Illescas and Manta (Fig. 11f). Likewise, the increasing trend in the number of measurements southward matches with the increase in wave height (Fig. 11d). An increase in wave height and tidal range may lead to enhanced erosion and morphologically well-expressed marine terraces (Anderson et al., 1999; Trenhaile, 2002), which is consequently reflected in a higher number of measurements that can be carried out. Furthermore, we observe low values for the reference water level (< 0.7 m) resulting from tide and wave-height estimations in IMCalc (Lorscheid and Rovere, 2019), which are used to correct our shoreline-angle measurements in the WALIS database (see Sect. 3.3).

The control of wave-erosion processes on the morphological expression of marine terraces may be counteracted by erosional processes such as river incision. We note that the high number of preserved marine terraces between Mejillones and Valparaíso decreases southward, which coincides with a sharp increase in mean annual precipitation from 10 to 1000 mm yr-1 (Fig. 11e and f) and fluvial dissection. However, in the area with a high number of measurements between the Illescas Peninsula and Manta we observe an opposite correlation: higher rainfall associated with an increase in marine terrace preservation (Fig. 11e). This suggests that the interplay between marine terrace generation and degradation processes apparently buffer each other resulting in different responses under different climatic conditions and coastal settings.

The greater number of marine terraces between Mejillones and Valparaíso and north of Illescas corresponds with a SSW–NNE orientation of the coastline (azimuth between 200 and 220∘). In contrast, NW–SE- to north–south-oriented coastlines (azimuth between 125 and 180∘), such as between the Arica and Huancabamba bends, correlate with a lower number of marine terrace measurements (Fig. 11e and f). This observation appears, however, implausible considering that NW–SE-oriented coastlines may be exposed more directly to the erosive effect of storm waves associated with winds approaching from the south. We interpret the orientation of the coastline therefore to be of secondary importance at regional scale for the formation of marine terraces compared to other parameters, such as wave height, tidal range, or rainfall.