The crustal structure of the continent-ocean transition zone in the South Atlantic salt basins is poorly understood. Current interpretations place the limits of oceanic crust at the distal salt limits, with sub-salt crust consisting of rifted continental crust and, in some versions, varying amounts of exhumed mantle. Plate reconstructions that map these limits of oceanic crust onto appropriate-age restorations show poor geometric fits, with unexplained gaps and overlaps. One possible reason for the poor fits is that the distal salt limits are not the real limits of oceanic crust. In this paper we investigate this option by mapping rift basins and seaward-dipping reflectors whose seaward edges mark significant structural boundaries as much as 300 km inboard of the distal salt limits. We interpret these boundaries, which match geometrically in a salt-age (Aptian) plate reconstruction, to be the limits of oceanic crust. We suggest that salt was deposited as seafloor spreading commenced and that, as the South Atlantic opened, salt flowed over the ridge axis, sealing off the extrusive component of oceanic crust, resulting in formation of intrusive oceanic crust. Seafloor spreading eventually broke through the thinning salt, forming breakthrough volcanoes preserved today as basement ramps at the distal salt limits. These ramps formed diachronously, so the distal salt limits are not isochrons, explaining the poor fit of these features in plate reconstructions.


Most crustal interpretations of the central South Atlantic (Davison, 2007; Davison et al., 2012; Quirk et al., 2012, 2013) regard the limit of oceanic crust (LOC; like Heine et al., 2013, we prefer this term to the more commonly used continent-ocean boundary) to be located at the distal autochthonous salt limit. However, this interpretation has several drawbacks. One is that gaps and overlaps of conjugate salt edges in plate restorations of the South Atlantic indicate that the edges of the salt basins cannot be isochrons (Fig. 1). Another is that the distal edge of autochthonous salt is in most places marked by a landward-dipping basement ramp (the outer basement ramp) having kilometers of relief in some places; that is, basement steps up onto oceanic crust (e.g., Fig. 2C). The existence of an outer basement ramp is typical of passive margins involving salt but uncommon at the LOC in salt-free margins (Hudec et al., 2013; Quirk et al., 2012, 2013; Davison et al., 2012). In this paper we reexamine the location of the LOC and suggest a tectonic scenario that provides simple explanations for these and other structural observations.


Prestack depth-migrated three-dimensional (3-D) seismic reflection data in the Campos Basin (central South Atlantic; survey location in Fig. 1) provide new clues to the location of the LOC. A northwest-southeast section from the data (Fig. 2A) suggests a major change in the structural framework of the pre-salt section in this region. At the landward (western) end of the line, the pre-salt succession is characterized by tilted fault blocks and hanging-wall growth wedges typical of rift basins. Extensional grabens are separated from the base of the salt by a fault-free, seaward-thinning unit termed the sag section (Jackson et al., 2000; Unternehr et al., 2010). At the east end of the line in Figure 2A, the sag section pinches out above a thick seaward-expanding and seaward-dipping reflector sequence. We argue that these are seaward-dipping reflectors (SDRs), i.e., reflectors produced by lavas erupted at a seaward-migrating volcanic center (Mutter et al., 1982). An alternative interpretation is that these reflections are from sediments in a half-graben, thickening into a landward-facing listric fault. We think that this alternative is unlikely because (1) the scale differs from the inboard rift grabens, i.e., >4 km thickness versus <1 km for the grabens to the west, and (2) the seismic character of our proposed SDRs differs from that of the sediments in the rift basins and is more consistent with known SDRs from other margins (Mutter et al., 1982; Jackson et al., 2000; Franke, 2013; Koopmann et al., 2014). For these reasons, we prefer interpreting the seaward-fanning reflector sequence as SDRs.

SDRs have been well studied in many basins, and the subaerial nature of the lava flows has been confirmed by drilling (Jackson et al., 2000; Franke, 2013). They are found along almost the full length of both South Atlantic margins south of the Walvis Ridge (Fig. 1; Franke, 2013; Koopmann et al., 2014; Blaich et al., 2013) and in the Jacuipe and Sergipe Alagoas Basins at the north end of the Brazilian salt basin (Mohriak et al., 1998). The important point for this study is that SDRs define the seaward edge of the continent-ocean transition zone; the limit of oceanic crust can be no farther outboard than the distal edges of the SDRs (Franke, 2013). Figure 2C is a seismic line from the rift-sag section through the Campos SDR region to oceanic crust. We place the LOC at the outboard edge of the SDRs; seaward of this, salt is directly on acoustic basement, which we interpret as oceanic crust. At the seaward limit of salt there is a large basement ramp, the outer basement ramp, beyond which salt forms an allochthonous sheet ∼20 km wide.

In map view, the SDRs can be traced using the 3-D data complemented by some 2-D data. They occur along an 80-km-long, northeast-southwest zone in the central Campos Basin and were used to map the LOC between the red arrows in Figure 1. Mapping of the LOC outside the SDR area must rely on other criteria. We note that the landward end of the SDRs coincides with seaward pinch-out of the pre-salt sag sequence (Fig. 2A). We interpret the sag section to be deposited above inactive rift structures, so that as rifting migrates away from the continent it leaves inactive rifts behind that are then covered by sag section; this is why the sag thins toward the rift axis (there is less time for deposition). Sag deposition ends when salt is deposited. As we interpret this to occur when seafloor spreading began, the LOC must be close to the distal limit of sag. The sag pinch-out can be mapped on 2-D data beyond the limits of the SDRs (Fig. 1). Figure 3 is an example seismic line from the Santos Basin, showing data across the Tupi structure. The sag sequence pinches out on the Tupi high, and seaward of this there is a large 2-km-high seaward-facing ramp. This ramp was mapped for more than 250 km along strike in the Santos Basin by Gomes et al. (2009; see shaded area in Fig. 1). Gomes et al. (2009) showed that it consists of several segments, and based on the pinch-out of sag section we interpret the ramp to be close to the LOC. Exactly where is poorly constrained, as there is some sub-salt reflectivity close to the ramp, but further outboard this reflectivity disappears; we place our LOC at the point of disappearance (Fig. 3). In seismic data between the Santos and Campos Basins, we map the LOC based on the rift-sag pinch-out. In some locations, there is a basement ramp close to the LOC; Figure 2C shows an example, where there is a ramp ∼700 m high at our LOC. We term this ramp the inner basement ramp. Note that in Figures 2C and 3, the base of the salt is very rugose outboard of our LOC.

We have not observed SDRs in seismic data available to us on the conjugate African margin, but have mapped the edge of the rift-sag basin (Fig. 1) using 11 2-D CongoSpan seismic lines (courtesy of GXTechnology–ION Geophysical; http://www.iongeo.com/About_Us/Corporate_Overview/GX_Technology/) and our interpretations of published data (Quirk et al., 2013; Unternehr et al., 2010; Kumar et al., 2013). The edge of the rift-sag basin in the line of Unternehr et al. (2010) is particularly clear. Mapping the edge in the area conjugate to the northern Campos Basin is poorly constrained partly because the Angola margin includes a string of seamounts, the Sumbe trend (Dickson et al., 2003), which obscures deep structure on seismic data. We have a gap in our LOC interpretation in this region. By analogy with the Brazilian margin, we suggest that the rift-sag basin edge that we mapped is also the African LOC. Figure 4 is a plate reconstruction that shows an excellent fit between these two conjugate LOCs. This fit is for the time of Chron M0r, 121 Ma, and was found by trial and error using rotation poles from Heine et al. (2013).


The improved plate fit (Fig. 4) suggests that our reinterpretation of LOCs on both sides of the South Atlantic may be valid; however, several questions remain. Why do the inner (seaward facing) and outer (landward facing) basement ramps exist and what is the relationship between the inner ramp and LOC? Why is the oceanic crust between the two ramps so rugose? At what point in this evolution was salt deposited, and what role did it play in the development of these structures?

We begin by discussing the timing of salt deposition relative to the crust on which it sits. Sag sediments progressively onlap the top of the SDR package and pinch out at the seaward end of the SDRs (Fig. 2). In the Santos Basin (Fig. 3) the sag section pinches out onto the Tupi high. In both locations salt directly overlies basement outboard of the sag pinch-out. Salt was therefore deposited shortly after the end of sag basin deposition and extrusion of the lava flows contained in the SDRs. We thus conclude that salt was deposited during the transition from rifting to seafloor spreading, after SDR formation but before significant generation of oceanic crust. This places salt deposition at the time of the reconstruction in Figure 4, Chron M0, which is, by definition, the base of the Aptian stage (Erba, 1996). The absolute age of salt is poorly constrained (Davison, 2007; Davison et al., 2012). A base Aptian age is consistent with some, but not all, observations constraining the age of salt. For more discussion of the absolute age of salt and problems with the geologic time scale in the Aptian, see the GSA Data Repository1.

We next discuss our diverse structural observations: the LOC, ramps, and base of salt rugosity. We suggest that all these are interrelated, and have developed a model for evolution of the salt basin. This is presented in Figure 5, which is drawn on the basis of output of basin modeling of two cross sections (for details, see the Data Repository).

Sedimentation during rifting was nonmarine (Chaboureau et al., 2013), and our basin modeling suggests that the area of the future salt basin was below sea level; the rift axis was ∼1000 m below sea level by the end of rifting (Fig. 5A). Figure 5A shows the configuration without water and Figure 5B shows the basin filled with water, which isostatically lowered the rift axis to ∼1500 m below sea level. Estimates of duration of salt deposition are poorly constrained, ranging from 600 k.y. (Dias, 2005) to 5 m.y. (Davison et al., 2012). Given Aptian plate spreading rates of 50–60 km/m.y. (Heine et al., 2013), longer intervals of salt deposition would mean that the conjugate salt basins separated before the end of salt deposition; this is implausible. We thus prefer a short period of salt deposition. We also assume that isostatic loading, occurring as it does over thousands of years as opposed to hundreds of thousands of years for salt deposition (Van den Belt and de Boer, 2007), was instantaneous (Fig. 5C). As separation between Africa and South America continued (Fig. 5D), new oceanic crust was produced, although salt continuously flowed out over the newly forming crust. Because the salt sealed off contact with seawater, this oceanic crust was unable to produce the extrusive volcanics of layer 2A, resulting in all the crust being intrusive under the salt. We call this crust intrusive oceanic crust, and expect that the only difference between this crust and normal oceanic crust is the lack of layer 2A; therefore, we think of it as a different flavor of oceanic crust. Data consistent with our model come from recent seismic refraction studies in the Santos Basin (Evain et al., 2015). Evain et al. (2015) mapped several crustal zones, one of which coincides with our area of intrusive oceanic crust. In discussing this zone, Evain et al. (2015, p. 5427) stated, “If oceanic crust is present, it is, however, atypical and without its layer 2 (basalt)…”; this is consistent with our concept of intrusive oceanic crust.

The seaward-dipping inner basement ramp is not found throughout the salt basin. Where seen, it represents a step downward from rifted continental crust or SDRs to intrusive oceanic crust. The intrusive crust is anomalously deep because (1) it was isostatically loaded by the salt, and (2) it was unable to pierce the thick salt sequence, instead cooling at depth beneath the highly conductive salt layer. We suggest that variations in these processes control whether there is an inner basement ramp. Where salt is in contact with intrusive oceanic crust, we suggest that the rugosity is a consequence of variations in emplacement level of the oceanic crust through time, presumably owing to temporal variations in pressure, temperature, or volatile content of the magma. As the basin widened, salt continued to spread into the newly forming basin, thinning as it did. For simplicity, we assume no salt dissolution in our model, keeping the cross-sectional area of salt constant. Eventually salt thinned enough that oceanic crust broke through the salt (Fig. 5E), forming a breakthrough volcano that is preserved today as the outer basement ramp (Fig. 2C). Breakthrough was diachronous, resulting in distal salt limits that are not isochrons (Fig. 1). As seafloor spreading continued (Fig. 5F), normal (i.e., subaqueous) oceanic crust was formed in a process that continues today at the mid-ocean ridge. Note that both the inner and outer ramps are not faults; they are constructional features formed during crustal accretion.


Our new LOC has some important implications for hydrocarbon exploration. Almost all the discoveries in the Brazilian salt basins (Mello et al., 1988), and some recent offshore Angola discoveries (Cazier et al., 2014), are sourced from pre-salt organic-rich sediments. As this sedimentation ended once oceanic crust started to accrete and salt deposition began, there can be no pre-salt source rocks on oceanic crust. Our LOCs are well inboard of the distal salt limits, meaning that prospectivity of the outer salt basins for pre-salt sourced oils is essentially eliminated.


In this paper we present a new model for separation of the South Atlantic salt basins. Seismic data show that rift-sag basins occur along both the Brazilian and African margins, the distal limits of which we interpret to be close to the inner limits of oceanic crust. In our study area in the Campos Basin, a zone of SDRs formed along this limit. The LOC is locally marked by seaward-facing scarps. Because salt is above sag-phase sediments that pinch out near the SDRs, we suggest that salt was deposited over a brief period as seafloor spreading commenced. As the South Atlantic opened, because salt flowed continuously onto the developing oceanic crust, this crust was emplaced intrusively beneath the salt. Eventually the salt thinned by stretching enough that volcanism broke through the salt, forming breakthrough volcanoes preserved today as the outer basement ramp near the distal limit of salt. This final breakthrough was diachronous and, in some areas, very asymmetric. In the Santos Basin, for example, final breakthrough was close to the African LOC so that all the salt remained on the South American side of the ocean.

We thank PGS (Oslo, Norway), Fugro (Netherlands), and GXTechnology–ION Geophysical (Texas, USA) for permission to publish seismic lines, Landmark (Halliburton; Houston, Texas) for access to DecisionSpace, and the Rothwell Group (Colorado, USA) for access to PaleoGIS software. We thank Tony Watts for providing the MATLAB version of his flexural modeling code and Nancy Cottington for help drafting figures. We also acknowledge helpful discussions with Martin Jackson and Chris Jackson on South Atlantic passive margin evolution and rift structure. This paper greatly benefited from comments by R. Wilson, G. Karner, and an anonymous reviewer. This project was sponsored by AGL (Bureau of Economic Geology) and the PLATES project (Institute for Geophysics, UTIG), both part of the Jackson School of Geosciences, University of Texas. This is UTIG contribution 2846. Publication is authorized by the Director, Bureau of Economic Geology, Texas.

1GSA Data Repository item 2016013, discussion of Aptian time scale and details about basin modeling, is available online at www.geosociety.org/pubss/ft2016.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.