Earth and Planetary Science Letters 414 (2015) 156–163

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

The Cretaceous opening of the South Atlantic Ocean

Roi Granot a,∗, Jérôme Dyment b

a Department of Geological and Environmental Sciences, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israelb Institut de Physique du Globe de Paris, CNRS UMR 7154, Sorbonne Paris Cité, Université Paris Diderot, 75005 Paris, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 2 October 2014Received in revised form 8 January 2015Accepted 13 January 2015Available online xxxxEditor: P. Shearer

Keywords:plate motionsSouth Atlanticmagnetic anomaliesquiet zone

The separation of South America from Africa during the Cretaceous is poorly understood due to the long period of stable polarity of the geomagnetic field, the Cretaceous Normal Superchron (CNS, lasted between ∼121 and 83.6 Myr ago). We present a new identification of magnetic anomalies located within the southern South Atlantic magnetic quiet zones that have arisen due to past variations in the strength of the dipolar geomagnetic field. Using these anomalies, together with fracture zone locations, we calculate the first set of magnetic anomalies-based finite rotation parameters for South America and Africa during that period. The kinematic solutions are generally consistent with fracture zone traces and magnetic anomalies outside the area used to construct them. The rotations indicate that seafloor spreading rates increased steadily throughout most of the Cretaceous and decreased sharply at around 80 Myr ago. A change in plate motion took place in the middle of the superchron, roughly 100 Myr ago, around the time of the final breakup (i.e., separation of continental–oceanic boundary in the Equatorial Atlantic). Prominent misfit between the calculated synthetic flowlines (older than Anomaly Q1) and the fracture zones straddling the African Plate in the central South Atlantic could only be explained by a combination of seafloor asymmetry and internal dextral motion (<100 km) within South America, west of the Rio Grande fracture zone. This process has lasted until ∼92 Myr ago after which both Africa and South America (south of the equator) behaved rigidly. The clearing of the continental–oceanic boundaries within the Equatorial Atlantic Gateway was probably completed by ∼95 Myr ago. The clearing was followed by a progressive widening and deepening of the passageway, leading to the emergence of north–south flow of intermediate and deep-water which might have triggered the global cooling of bottom water and the end for the Cretaceous greenhouse period.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Seafloor spreading between South America and Africa began around Anomaly M13 (134 Myr ago, Channell et al., 1995) in the southernmost part of the South Atlantic Ocean and then propa-gated northward toward the Equatorial Atlantic Ocean. The ini-tial breakup probably resulted from the inception of the Tristan hotspot that formed the Paraná–Etendeka large igneous province (Renne et al., 1992). Despite their geodynamic and paleooceano-graphic importance, the relative motions between the two plates during most of the breakup period are still poorly resolved due to the lack of reversal-related magnetic anomalies on crust of age 121 (Anomaly M0y) to 83.6 (Anomaly C34y) Myr old (Malinverno et al., 2012; Ogg, 2012). This crust, formed during the Cretaceous Normal Superchron (CNS), is termed the magnetic “quiet zone” and is the focus of the present study.

* Corresponding author. Tel.: +972 8 6477509.E-mail address: rgranot@bgu.ac.il (R. Granot).

The available kinematic models for the opening of the South Atlantic during the CNS (Eagles, 2007; Heine et al., 2013; Jones et al., 1995; Moulin et al., 2010; Nürnberg and Müller, 1991)are based on interpolation between the rotation parameters of M0y and C34y, predicting an opening age of ∼110–104 Myr ago (clearing of continental–oceanic boundaries, COBs) for the Equa-torial Atlantic Gateway, even though equatorial sedimentary fa-cies and subsidence rates (Pletsch et al., 2001) and global com-pilation of benthic foraminifera stable isotopes (Friedrich et al., 2012) indicate that the establishment of deep-water circulation between the South and Central Atlantic Basins started roughly 10–15 Myr later. Recent studies have tried to overcome this con-tradiction by using either the seaward-edge of the salt basins (Torsvik et al., 2009) or gravity scars located within the South Atlantic quiet zones (Pérez-Díaz and Eagles, 2014) to constrain the rotation parameters during the superchron. These works had to assume that these markers were stable since their formation (i.e., isochrones-like features) and assign ages, which are poorly constrained.

http://dx.doi.org/10.1016/j.epsl.2015.01.0150012-821X/© 2015 Elsevier B.V. All rights reserved.

R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163 157

Fig. 1. Tectonic map of the South and Equatorial Atlantic Oceans superimposed on a satellite-derived free-air gravity field (Sandwell et al., 2014). Magnetic picks and FZs crossings are shown with empty symbols for the six anomalies used (rounded ages are shown in Myr). Filled symbols show the rotated picks (rotation parameters from Table 1). The gray flowlines are calculated on the basis of the rotation parameters of Cande et al. (1988), covering the last 79 Myr (Anomaly C33o to present), and the red lines are flowlines calculated on the basis of the new stage rotations (Fig. 3b, Anomalies M2o to C33o, marked with filled black circles). Red stars denote the initiation points of the displayed flowlines. For uncertainties on selected flowline locations see Fig. 4. Dashed red flowlines delineate the expected location of FZs using interpolation between the rotation parameters of Anomalies M0y and C34y. Blue lines mark the oldest conjugate anomalies formed along the fossil spreading axes (black dashed lines) (Cande et al., 1988; Cande and Rabinowitz, 1978). Arrows show the expected position of the flowlines based on the gravity data. Gray shaded area demarcates a sliver of transferred crust (Sandwell et al., 2014).

Although geomagnetic reversal-related anomalies are the most prominent magnetic signature of the oceanic crust, past changes in the strength of the dipolar geomagnetic field are also recorded, re-sulting in globally correlatable wiggles within geomagnetic chrons (Bouligand et al., 2006; Cande and Kent, 1992; Granot et al., 2012;Pouliquen et al., 2001). Two globally traceable magnetic anoma-lies from the CNS (Granot et al., 2012) are used here to define internal ages within the South Atlantic quiet zones: the oldest is a long-wavelength wiggle (Anomaly Q2, 108 ± 1 Myr ago) and the youngest (Anomaly Q1, 92 ±1 Myr ago) is a short-wavelength wig-gle, which is followed by subdued magnetic anomalies toward the younger end of the quiet zones. The ages (and uncertainties) of these anomalies are based on the ages of the oldest sediments that were drilled in the Central Atlantic Ocean (Granot et al., 2012). We have picked the location of these CNS magnetic anomalies and combined them with fracture zone (FZ) locations to compute inter-nal rotation parameters for the superchron period. We complement these two rotations with a new set of picks for Anomalies M2o,

M0y, C34y and C33o. From these finite rotations we derive stage rotations that constrain the relative motion of South America and Africa at 5 to 15 Myr intervals between 124 and 79 Myr ago. We then discuss the evolution of seafloor spreading in the South At-lantic and conclude with the geodynamic and paleooceanographic implications of our results.

2. Data and methods

We constrain South America–Africa motion by using magnetic anomaly picks and FZ traces found south of the Tristan da Cunha FZ, where both quiet zones are fully preserved (Figs. 1, 2 and S1). Locations of FZs were selected from the satellite gravity map (version 23) of Sandwell et al. (2014). Magnetic anomaly data were obtained from the National Geophysical Data Center (NGDC) and contain both shipboard and aeromagnetic (Project MAGNET) profiles. The locations of the reversal-related anomalies were picked based on forward modeling whereas the location of

158 R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163

Fig. 2. Magnetic anomaly profiles used in this study. Location of Anomalies Q2 and Q1 are shown with arrows. The profiles are located between FZs to prevent tectonic complications at segment ends.

the geomagnetic intensity-related anomalies were determined by comparison against profiles from other oceanic basins (Granot et al., 2012). We identified Anomalies Q1 and Q2 within each quiet zone by (1) tracing the general, long wavelength shape of the mag-netic profiles and (2) picking their exact locations by correlation of the short wavelength signal. We note that tectonic-related fea-tures (e.g., buried seamounts, variability in the basalt iron content, undetected small offsets of the spreading centers, etc.) may alter the shape of the anomalies enough to locally mask the intensity-related signal. We therefore assigned conservative data uncertain-ties of 4 km and 5 km to the magnetic and FZ picks, respectively, to account for these difficulties while computing the finite rotation parameters.

Finite rotation parameters were calculated by implementing the statistical routines of Kirkwood et al. (1999) and Royer and Chang(1991). We define the location of six anomalies (M2o, M0y, Q2, Q1, C34y and C33o, Supplementary Table S1), and together with FZ crossings we calculate a set of finite rotation parameters (poles and angles), with their 95% uncertainty intervals (Fig. 3a and Ta-ble 1). From the finite rotations we compute five stage poles that

describe the evolution of relative motions between the two plates throughout most of the Cretaceous.

The ages of the Cenozoic anomalies follow the geomagnetic po-larity timescale of Ogg (2012). The ages of the Mesozoic anomalies (M-sequence) are less certain. We adopt 121 Myr old as the age for Anomaly M0y and 124 Myr old as the age for Anomaly M2o (He et al., 2008; Malinverno et al., 2012). We note that the choice of the Mesozoic timescale affects only the calculated spreading rates and tectonic inferences for the period bounded between Anomalies M2o and Q2, since Anomalies Q2 and Q1 were inde-pendently dated (Granot et al., 2012).

3. Finite rotations

The six finite rotations and their uncertainty ellipses are shown in Fig. 3. The rotations and their covariance matrices are presented in Table 1. Generally, the locations of the poles are well-defined and their positions form a clear trend. The values of the statistical parameter κ̂ (Royer and Chang, 1991) are near one (Table 1) in-

R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163 159

Fig. 3. Kinematics of the South Atlantic Ocean. a, Finite rotation poles with their 95% confidence limits. Dashed red line corresponds to the 95% confidence limits for the solution calculated on the basis of data obtained south of the Rio Grande FZ. b, Stage poles for South America–Africa displacements. Poles are shown for South America (red) and Africa (green). c, Spreading rates between South America and Africa along the Rio Grande FZ synthetic flowlines (blue). Gray bars show 95% confidence intervals. Bars along the abscissa show chrons. Gray line shows spreading rates that are based on Mesozoic (Nürnberg and Müller, 1991) and Cenozoic (Cande et al., 1988) kinematic solutions, with interpolated spreading rates for the CNS. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1

Finite rotations and covariance matrices for the relative motion of Africa (fixed) and South America. The covariance matrix is given by the formula 1κ̂

∗!

a b cb d ec e f

"∗ 10−7 rad2.

Age (Ma)

Mag. ano. Lat. (◦N)

Long. (◦E)

Angle (◦)

κ̂ a b c d e f Points Segs.

79.9 C33r 62.47 326.08 30.94 0.59 7.73 −4.31 −4.83 4.77 3.69 4.26 63 783.6 C34 61.98 325.32 33.48 1.19 6.45 −3.61 −4.01 5.48 3.77 3.83 58 792.0 Q1 60.84 323.50 38.58 0.69 18.93 −14.70 −15.72 14.84 14.23 16.17 33 7

108.0 Q2 62.17 319.83 47.35 1.41 491.62 −531.78 −653.66 579.8 709.15 874.31 26 4120.6 M0y 53.55 324.36 52.75 0.87 51.66 −58.28 −64.08 74.76 75.82 83.10 28 4124.0 M2o 52.16 324.90 53.86 2.33 38.17 −45.92 −48.62 60.63 60.91 64.60 31 4

dicating that the uncertainty assigned to the data points used to calculate the solutions were reasonable.

The identification of anomalies within the quiet zones is not straightforward and requires special care. We validate the kine-matic solution of Anomaly Q1 by re-calculating the rotation pa-rameters by including magnetic picks from an additional spreading segment located between the Bode Verde and Rio de Janeiro FZs, for which there are sufficient data on both flanks (Fig. 1). A re-cent update of the satellite-derived world gravity field (Sandwell et al., 2014) unraveled conjugate tectonic features of pseudofaults and a sliver of transferred crust within this spreading segment (Fig. 1). This suggests that an episode of northward ridge propa-gation possibly related to the Tristan hot spot activity has affected the crust located ∼100 km west of the (soon to be failed) spread-ing axis. To avoid the possibility of a biased kinematic solution, we have limited our identifications of Anomaly Q1 to the region lo-cated south of the tectonically-affected crust. The new solution is identical to the one constrained solely by the southern spreading segments, but it is better defined (Fig. 3a), confirming the Q1 iden-tifications. The identifications and kinematic solution of Anomaly Q2 are less certain (Table 1) due to its long wavelength nature and the general tendency for strong and varied wiggles within the middle part of the quiet zones (Granot et al., 2012). Although less reliable, the Q2 solution provides a useful constraint on the

relative motion between the plates for the middle part of the su-perchron.

4. Stage rotations

To evaluate the predictive quality of the new finite rotations we calculate stage rotations (Fig. 3b) and generate synthetic flowlines (Figs. 1 and 4) that, ideally, should trace the actual position of FZs. Indeed, the southern flowlines (north of the Falkland-Agulhas and Gough FZs) trace their positions within uncertainties (Fig. 4). Sim-ilarly, north of the Ascension FZ, in the northern part of the South Atlantic Ocean and in the Equatorial Atlantic Ocean (Figs. 1 and 4), the synthetic flowlines trace the actual location of FZs. Altogether, these results indicate that the kinematic solutions presented here are robust, even for the Cretaceous quiet zones.

In contrast to the southern and northern sections of the spread-ing system, there is a growing misfit between the FZs and the synthetic flowlines on the African flank between (and including) the Rio Grande and Ascension FZs. The misfit starts at Anomaly Q1 and grows to about 100 km towards Anomaly Q2 (Figs. 1 and 4). This one-sided misfit could have resulted from an asymmetric spreading episode that would have taken place between Anoma-lies Q2 (or possibly before) and Q1, whereby eastward migration of the ridge there would have led to more lithosphere on the South American flank. Alternatively, the misfit might have formed due to

160 R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163

Fig. 4. Oblique Mercator projections (Anomaly C34y pole, 325◦N, 62◦E) of selected FZs. Synthetic flowlines (red) are based on new stage poles (Fig. 3b). Dashed red lines show interpolated flowlines; green lines are flowlines computed assuming internal motion within Africa; purple lines are flowlines assuming internal motion within South America; and light blue lines are flowlines assuming asymmetric spreading (70 percent of lithosphere created on the South American plate between Q2 and Q1). Details of the parameters used are given in the text. Confidence intervals for the locations of Q1 and Q2 are shown. Arrows show the expected positions of the flowlines based on the gravity data.

a missing plate boundary that straddled the Rio Grande FZ or a nearby FZ, cutting through either of the two continents.

Kinematic simulations (Fig. 4) enable us to choose the most suitable of the three possible tectonic scenarios. First, for the flow-lines to fit the FZs of the African flank due to continental defor-mation within the African plate, it is necessary to have an inter-nal African rift that could be described by a 1.3◦ rotation over a 345◦E/15◦S pole. In such a scenario, the misfit between the syn-thetic flowlines and the FZs of the African flank is indeed reduced, but it introduces an unacceptable misfit to their conjugate FZs located on the South American flank (Fig. 4, green lines). Further-more, this scenario would require an unrealistic opening motion within Africa (∼110 km of N–S opening near the East African Rift). This prediction is not supported by geological observations (Heine et al., 2013; Moulin et al., 2010) and we thus consider it unlikely.

Introducing ∼106 km of dextral strike slip motion along the Rio Grande FZ and further west within the South American Plate (−1◦ rotation over a 306◦E/79◦N pole) provides a better fit to the flowlines of the African flank while not affecting the fit to the South American FZs (Fig. 4, purple lines). Importantly, the region where the Rio Grande FZ meets the South American continent is

considered to be a deformed zone that might have accommodated some (tens of kilometers) strike slip motion in the Cretaceous (Eagles, 2007; Moulin et al., 2010; Torsvik et al., 2009).

Alternatively, asymmetry in seafloor spreading may have been fueled by the proximity to the active Tristan da Cunha hotspot (Rohde et al., 2013), resulting in an eastward migration of the ridge axis. We modeled asymmetric seafloor spreading by assign-ing roughly 70 percent crustal accretion on the South American flank between Anomalies Q2 and Q1. We assigned 60 percent asymmetry for the spreading segment confined between the Bode Verde and Rio de Janeiro FZs due to the ridge propagator and cap-ture of transferred crust there (Sandwell et al., 2014). Asymmetric seafloor spreading helps to fit (yet not perfectly) the FZs of the African flank without affecting the South American fit while still complying with the location of the anomalies (Fig. 4, blue lines). Two Cretaceous fossil spreading axes dated by magnetic anomalies to be active prior (Cande and Rabinowitz, 1978) and after (Cande et al., 1988) the CNS are indeed found on the current South Amer-ican flank (Fig. 1), supporting the scenario that seafloor spreading asymmetry must have started before and continued after the CNS, with a preferentially eastward motion of the spreading axis, i.e., more lithosphere accreted to the South American flank. Moreover,

R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163 161

Fig. 5. Plate reconstructions and global bottom water temperatures. Plate reconstructions are shown at Anomalies Q2, 108 ± 1 Myr ago (a), Q1, 92 ± 1 Myr ago (b), and C34y, 83.6 Myr ago (c). South America is fixed. Contours correspond to 1000- and 2000-m isobaths. The geometry of the mid-ocean ridges are constrained only where we have magnetic anomaly picks (Fig. 1) and interpolated elsewhere. d, Benthic foraminifera oxygen isotopic data from Friedrich et al. (2012). Arrow delineates the initiation of cooling of global deep-water masses.

gravity scars located within the South American quiet zone may preserve evidence for the abandoned spreading center that was left behind when the ridge migrated eastward (Pérez-Díaz and Eagles, 2014).

To conclude, our new rotations suggest that up to ∼100 km of dextral motion has been accommodated within the South Amer-ican Plate, west of Rio Grande FZ, after 121 Myr ago (Anomaly M0y). Additionally, the central part of the South Atlantic was prob-ably formed with some degree of asymmetric seafloor spread-ing resulting in preferential accumulation of crust to the South American Plate. The exact amount of internal dextral deforma-tion within South America depends on the magnitude of asym-metric seafloor spreading and therefore still remains uncertain. Finally, the excellent fit of our flowlines to the actual position of FZs on crust younger than Anomaly Q1 (Figs. 1 and 4) im-plies that internal continental deformation must have stopped

by 92 Myr ago, in agreement with geological observations from Africa and South America (Heine et al., 2013; Moulin et al., 2010;Torsvik et al., 2009).

5. Geodynamic implications

The new kinematic solutions presented here provide impor-tant constraints on the relative motions between Africa and South America during their final stages of separation. The close prox-imity of the early rotation poles (Fig. 3b) to the Equatorial At-lantic implies that a slow rotational motion governed the equa-torial area while the continents were still partly attached (Fig. 5). Applying linear interpolation between the different rotations, our plate reconstructions indicate that the transition from rift to drift (breakup) took place at around 99 ± 1 Myr, assuming rigidity of both plates. Independent palaeomagnetic data indicate that a vir-

162 R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163

tual geomagnetic pole standstill of South America ended at around 100 Myr ago and was followed by a westward drift of the con-tinent (Somoza and Zaffarana, 2008). This finding supports the ∼100 Myr timing for the final breakup between Africa and South America. Rapid northward migration of the stage rotation poles af-ter Anomaly Q2 (Fig. 3b) reveals a major plate reorganization of the South Atlantic spreading system. This change in plate motion has previously been linked to a supposedly global plate reorgani-zation event at around 105–100 Myr, possibly triggered by the dy-namics of subduction and mantle plumes (Matthews et al., 2012). Our new reconstructions provide an alternative explanation sug-gesting that the plate motion reorganization in the South Atlantic coincided with the final separation of South America and Africa.

Seafloor spreading rates increased steadily from slow rates (full spreading rate of 20 mm/yr along the Rio Grande FZ) before the CNS and peaked at intermediate rates (77 mm/yr, Fig. 3c) immedi-ately after the superchron. Motion has decreased to intermediate spreading rates (40 mm/yr) towards 60 Myr ago (Cande et al., 1988). Interestingly, spreading rate has peaked some 20 Myr af-ter the completion of separation of the plates. A similar tectonic pattern, i.e. reorientation of relative plate motion direction and an increase followed by a delayed decrease in spreading rates, oc-curred in the Northeast Atlantic during and after the breakup of Greenland from Europe (Gaina et al., 2009). These two examples suggest that the development of plate motions during (and up to 20 Myr after) the transition from rift to drift might be influenced by the mechanical setting of the rift zone and not governed solely by far-field (i.e., subduction-related) forces.

Our kinematic model has tectonics implications for the South and Equatorial Atlantic Basins. The new stage poles suggest that, until the middle of the CNS, the relative motions between South America and Africa were slower than previously expected (Nürnberg and Müller, 1991) (Fig. 3c). Since the apparent overall overlap between the continents in the Equatorial Atlantic, as is ev-ident from the rotation parameters of Anomalies M2o and M0y, is similar to that suggested by previous studies (Moulin et al., 2010;Nürnberg and Müller, 1991), our results imply that internal de-formation lasted longer than previously expected, probably until roughly 100 Myr ago and no later than 92 Myr ago. This age stands in general agreement with geological observations from the rift zones of Africa (Fairhead et al., 2013; Heine et al., 2013).

The seaward edge of the Aptian salt basins found along the Brazilian and Angola–Congo–Gabon margins have been used by Torsvik et al. (2009) as fixed conjugate isochrons for 112 Myr time. Linear interpolation between our M0y and Q2 finite rotations brings these basins to match at 116 Myr ago. We note that al-though it seems that the age of breakup of the basins is somewhat older than previously assumed, our model is not well constrained for that period of time.

Finally, the opening of the Equatorial Atlantic Gateway had a profound effect on the Cretaceous climate (Frank and Arthur, 1999;MacLeod et al., 2011), yet current magnetic anomaly-based kine-matic models have failed to match the timing of the opening to ge-ological observations. Our new reconstruction model suggests that the final clearing of the continent–ocean boundaries in the Equato-rial Atlantic Ocean was completed by 95 ± 1 Myr, followed by the opening of the gateway (Fig. 5). The initial deep-water circulation might have been restricted by the dynamically uplifted seafloor (Flament et al., 2014) and the major FZs (Jones et al., 1995) found in the Equatorial Atlantic. During that stage, the northern section of the South Atlantic Basin was already rather wide (500–800 km), and therefore we believe that the seamount chains resulting from the volcanic activity at Tristan hotspot had only a minor impact on water circulation. An increase in water exchanges between the Atlantic basins developed progressively, while the oceanic crust within the gateway continued to age and deepen. As a result, a

major reorganization of oceanic circulation patterns emerged, lead-ing to circulation system similar to that of the modern Atlantic Ocean, where north–south flow of intermediate and deep-water dictates water-mass and heat exchanges (Frank and Arthur, 1999). Oxygen isotopic signatures of benthic foraminifera from different oceanic basins (Friedrich et al., 2012) (Fig. 5d) show that a global cooling of bottom water started at about 92 Myr and continued progressively until ∼80 Myr.

6. Conclusions

Magnetic anomalies and FZ locations from the southern and central South Atlantic are used to construct a set of six rotation parameters that provide new constraints on the relative plate mo-tion between South America and Africa during the Cretaceous. The rotations are consistent with the pattern of magnetic anomalies found in other parts of the basin (for Anomaly Q1) and fracture zone locations.

The rotations presented here indicate that Lower Cretaceous slow spreading rates have gradually increased and peaked at in-termediate spreading rates shortly after the CNS, followed by a decrease in spreading rates. A change in the direction of relative motions between the two plates took place in the middle of the CNS, at around 100 Ma. This event generally coincided with the time of final separation between the plates, and hence might be linked with the final transition into complete drift motion. Misfit of synthetic flowlines in the central South Atlantic indicates that internal strike-slip deformation within the South American Plate (at around the Rio Grande FZ) has lasted until roughly 92 Myr ago. This process was probably active in conjugation with seafloor asymmetry. Finally, the results indicate, without taking a-priori as-sumptions, that the separation of the plates was later than previ-ously thought. This in-turn seems to agree with the global deep-water isotopic signature that suggest an age of roughly 95–90 Myr for the establishment of deep water exchange across the Equatorial gateway.

Acknowledgements

This work was supported by the European Union Seventh Framework Programme (FP7/2007–2013) under grant agreement n◦ 320496 (GEOPLATE). We thank Nicolas Flament, an anonymous reviewer and the Editor for their constructive comments. Chris Heine is thanked for his comments on the early version of this manuscript. This is IPGP contribution 3606.

Appendix A. Supplementary material

Supplementary material related to this article can be found on-line at http://dx.doi.org/10.1016/j.epsl.2015.01.015.

References

Bouligand, C., Dyment, J., Gallet, Y., Hulot, G., 2006. Geomagnetic field variations between chrons 33r and 19r (83–41Ma) from sea-surface magnetic anomaly pro-files. Earth Planet. Sci. Lett. 250, 541–560.

Cande, S.C., Kent, D.V., 1992. Ultrahigh resolution marine magnetic anomaly pro-files: a record of continuous paleointensity variations? J. Geophys. Res. 97, 15075–15083.

Cande, S.C., LaBrecque, J.L., Haxby, W.F., 1988. Plate kinematics of the South Atlantic: chron C34 to present. J. Geophys. Res. 93, 13479–13492.

Cande, S.C., Rabinowitz, P.D., 1978. Mesozoic seafloor spreading bordering conju-gate continental margins of Angola and Brazil. In: Proc. 10th Ann. Offshore Tech. Conf., pp. 1869–1876.

Channell, J.E.T., Erba, E., Nakanishi, M., Tamaki, K., 1995. Late Jurassic–Early Creta-ceous time scales and oceanic magnetic anomaly block models. In: Berggren, W.A., Kent, D.V., Aubry, M.-P., Hardenbol, J. (Eds.), Geochronology, Time Scales and Global Stratigraphic Correlation. SEPM Spec. Publ., pp. 51–64.

R. Granot, J. Dyment / Earth and Planetary Science Letters 414 (2015) 156–163 163

Eagles, G., 2007. New angles on South Atlantic opening. Geophys. J. Int. 168, 353–361.

Fairhead, J.D., Green, C.M., Masterton, S.M., Guiraud, R., 2013. The role that plate tectonics, inferred stress changes and stratigraphic unconformities have on the evolution of the West and Central African Rift System and the Atlantic conti-nental margins. Tectonophysics 594, 118–127.

Flament, N., Gurnis, M., Williams, S., Seton, M., Skogseid, J., Heine, C., Muller, R.D., 2014. Topographic asymmetry of the South Atlantic from global models of man-tle flow and lithospheric stretching. Earth Planet. Sci. Lett. 387, 107–119.

Frank, T.D., Arthur, M.A., 1999. Tectonic forcings of Maastrichtian ocean-climate evo-lution. Paleoceanography 14, 103–117.

Friedrich, O., Norris, R.D., Erbacher, J., 2012. Evolution of middle to Late Cretaceous oceans—a 55 m.y. record of Earth’s temperature and carbon cycle. Geology 40, 107–110.

Gaina, C., Gernigon, L., Ball, P., 2009. Palaeocene—recent plate boundaries in the NE Atlantic and the formation of the Jan Mayen microcontinent. J. Geol. Soc. Lond. 166, 601–616.

Granot, R., Dyment, J., Gallet, Y., 2012. Geomagnetic field variability during the Cre-taceous Normal Superchron. Nat. Geosci. 5, 220–223.

He, H.Y., Pan, Y.X., Tauxe, L., Qin, H.F., Zhu, R.X., 2008. Toward age determination of the M0r (Barremian–Aptian boundary) of the Early Cretaceous. Phys. Earth Planet. Inter. 169, 41–48.

Heine, C., Zoethout, J., Müller, R.D., 2013. Kinematics of the South Atlantic rift. Solid Earth 4, 215–253.

Jones, E.J.W., Cande, S.C., Spathopoulos, F., 1995. Evolution of a major oceanographic pathway: the equatorial Atlantic. In: Scrutton, R.A. (Ed.), The Tectonics, Sedi-mentation and Palaeoceanography of the North Atlantic Region. Geol. Soc., Lon-don, pp. 199–213.

Kirkwood, B.H., Royer, J.Y., Chang, T.C., Gordon, R.G., 1999. Statistical tools for es-timating and combining finite rotations and their uncertainties. Geophys. J. Int. 137, 408–428.

MacLeod, K.G., Isaza Londoño, C., Martin, E.E., Jiménez Berrocoso, A., Basak, C., 2011. Changes in North Atlantic circulation at the end of the Cretaceous greenhouse interval. Nat. Geosci. 4, 779–782.

Malinverno, A., Hildebrandt, J., Tominaga, M., Channell, J.E.T., 2012. M-sequence geo-magnetic polarity time scale (MHTC12) that steadies global spreading rates and incorporates astrochronology constraints. J. Geophys. Res. 117, B06104.

Matthews, K.J., Seton, M., Muller, R.D., 2012. A global-scale plate reorganization event at 105–100 Ma. Earth Planet. Sci. Lett. 355–356, 283–298.

Moulin, M., Aslanian, D., Unternehr, P., 2010. A new starting point for the South and Equatorial Atlantic Ocean. Earth-Sci. Rev. 98, 1–37.

Nürnberg, D., Müller, R.D., 1991. The tectonic evolution of the South-Atlantic from Late Jurassic to present. Tectonophysics 191, 27–53.

Ogg, J.G., 2012. Geomagnetic polarity time scale. In: Gradstein, F.M., Ogg, J.G., Schmitz, M., Ogg, G. (Eds.), The Geological Time Scale 2012. Elsevier, pp. 85–114. Chapter 115.

Pérez-Díaz, L., Eagles, G., 2014. Constraining South Atlantic growth with seafloor spreading data. Tectonics 33, 1848–1873.

Pletsch, T., Erbacher, J., Holbourn, A.E.L., Kuhnt, W., Moullade, M., Oboh-Ikuenobede, F.E., Soding, E., Wagner, T., 2001. Cretaceous separation of Africa and South America: the view from the West African margin (ODP Leg 159). J. South Am. Earth Sci. 14, 147–174.

Pouliquen, G., Gallet, Y., Patriat, P., Dyment, J., Tamura, C., 2001. A geomagnetic record over the last 3.5 million years from deep-tow magnetic anomaly pro-files across the Central Indian Ridge. J. Geophys. Res. 106, 10941–10960.

Renne, P.R., Ernesto, M., Pacca, I.G., Coe, R.S., Glen, J.M., Prévot, M., Perrin, M., 1992. The age of Paraná Flood Volcanism, rifting of Gondwanaland, and the Jurassic–Cretaceous boundary. Science 258, 975–979.

Rohde, J.K., Bogaard, P.v.d., Hoernle, K., Hauff, F., Werner, R., 2013. Evidence for an age progression along the Tristan–Gough volcanic track from new 40Ar/39Ar ages on phenocryst phases. Tectonophysics 604, 60–71.

Royer, J.Y., Chang, T., 1991. Evidence for relative motions between the Indian and Australian Plates during the last 20 My from plate tectonic reconstructions – im-plications for the deformation of the Indo–Australian Plate. J. Geophys. Res. 96, 11779–11802.

Sandwell, D.T., Müller, R.D., Smith, W.H.F., Garcia, E., Francis, R., 2014. New global marine gravity model from CryoSat-2 and Jason-1 reveals buried tectonic struc-ture. Science 346, 65–67.

Somoza, R., Zaffarana, C.B., 2008. Mid-Cretaceous polar standstill of South America, motion of the Atlantic hotspots and the birth of the Andean cordillera. Earth Planet. Sci. Lett. 271, 267–277.

Torsvik, T.H., Rousse, S., Labails, C., Smethurst, M.A., 2009. A new scheme for the opening of the South Atlantic Ocean and the dissection of an Aptian salt basin. Geophys. J. Int. 177, 1315–1333.