Refined spreading history at the Southwest Indian Ridge for the last 92 Ma, with the aid of satellite gravity data Armelle Bernard, Marc Munschy, Yair Rotstein, Daniel Sauter

To cite this version:

Armelle Bernard, Marc Munschy, Yair Rotstein, Daniel Sauter. Refined spreading history at the Southwest Indian Ridge for the last 92 Ma, with the aid of satellite gravity data. Geophysi- cal Journal International, Oxford University Press (OUP), 2005, 162, pp.765-778. ￿10.1111/j.1365- 246X.2005.02672.x￿. ￿hal-00104269￿

HAL Id: hal-00104269 https://hal.archives-ouvertes.fr/hal-00104269 Submitted on 13 Feb 2021

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Geophys. J. Int. (2005) 162, 765–778 doi: 10.1111/j.1365-246X.2005.02672.x

Refined spreading history at the Southwest Indian Ridge for the last 96 Ma, with the aid of satellite gravity data

A. Bernard,1 M. Munschy,1 Y. Rotstein1,2 and D. Sauter1 1Ecole et Observatoire des Sciences de la Terre, Universite Louis Pasteur, 5 rue Rene Descartes, 67084 Strasbourg Cedex, France. E-mail: [email protected] 2The Geophysical Institute of Israel, Holon, Israel

Accepted 2005 April 27. Received 2004 July 15; in original form 2003 October 10 Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

SUMMARY The spreading history of the oceans is modelled mostly by using magnetic anomalies and the fracture zone geometry. The high-quality, satellite-derived gravity data, that became available in recent years, reveal the details of fracture zones, which can be used as flow lines to control spreading models. We have applied this approach to the Southwest Indian Ridge (SWIR) in order to refine its spreading history. This is particularly useful for the period of complex spreading between magnetic anomalies 33 and 23, where the magnetic anomalies alone cannot resolve the detailed spreading history. We find four main stages in the spreading history of the SWIR since 96 Ma, including two that were not noted previously, between 96 Ma and anomaly 33 (76.3 Ma) and between anomalies 23o (51.7 Ma) and 18o (40.1 Ma; o denotes old boundaries of normal magnetization period). We also find that the start of the period of complex spreading was at anomaly 33, somewhat earlier than previously proposed. We discuss

the characteristics of the extension that the old transform faults underwent during the complex GJI Marine geoscience spreading phase, in response to the counterclockwise rotation of spreading. New transform faults appeared at that time, considerably widening the transform zones. Key words: Indian Ocean, Southwest Indian Ridge, spreading, transform faults.

a long period of very slow spreading, which extend to the present INTRODUCTION time. Consequently, previous, simple, single-rotation pole models The sea floor morphology of the Southwestern Indian Ocean is dom- were replaced by a series of rotations representing a rather complex inated by the Southwest Indian Ridge (SWIR), which extends for spreading history, with significant changes in the direction and rate some 7700 km between the Bouvet triple junction at 55◦S, 0.5◦W of spreading (Patriat et al. 1985; Patriat & S´egoufin1988; Royer and the Rodrigues triple junction at 25◦S, 70◦E(Fig. 1). Spread- et al. 1988). These models were mostly based on the identification ing at the axis of the SWIR started 165 Ma ago with the breakup of the magnetic anomalies and less on the use of fracture zones between Africa and Antarctica (Livermore & Hunter 1996). The as indicators of flow lines of the plate motion. This approach was early development of the Southwest Indian Ocean, from breakup to adopted since the fracture zones, although quite prominent in places, anomaly 34 [83 Ma on the magnetic timescale of Cande & Kent appeared to be quite linear, in a marked contrast with the magnetic (1995)] was poorly resolved until recently. Recent works (Marks & anomalies, which clearly detailed a complex spreading history at the Tikku 2001; Tikku et al. 2002) significantly improved the under- ridge. Some of these works (Patriat et al. 1985; Patriat & S´egoufin standing of the early accretion history and, in particular, resolved 1988) have generally used fracture zones as an overall guide and the overlap problem of Madagascar Plateau. They also defined the placed more importance on their trend, only for periods where location of Madagascar with respect to Africa and Antarctica. In the magnetic anomalies proved to be insufficient for determining contrast, the post-chron 34 evolution of the SWIR appeared to be rotation parameters. They noted some inconsistencies and misfits in well constrained by the early studies of the region (Norton & Sclater their models, but related them to fundamental geological processes, 1979; Patriat 1979; Tapscott et al. 1980; Sclater et al. 1981; Fisher such as regional plate deformation and not to a lack of data. & Sclater 1983; Martin & Hartnady 1986). These generally pro- Royer et al. (1988) were the first to use the earliest satellite al- posed that a single pole of rotation describes the motion between timetry data to improve the fracture zone geometry along the SWIR. Africa and Antarctica during the entire period from anomaly 34 to However, the satellite profiles were not dense enough for precisely the present. When additional magnetic data were collected in the following the traces of conjugate fracture zones, particularly in the region, it became apparent that spreading at the SWIR was not con- complex area between 25◦E and 35◦Ewhere the original trend of stant, but rather quite complex, with periods of rapid spreading and the transform faults might have been overprinted by a younger trend.

C 2005 RAS 765 766 A. Bernard et al.

0° 10° 20° 30° 40° 50° 60° 70°

-20°

-30°

-40° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-50°

-60°

Central Indian Ridge Rodrigues Triple ° -20 Junction AFRICA

MADAGASCAR

Melville

Novara

-30° Atlantis

Gallieni Southeast

Indomed Indian Ridge

Discovery -40°

PE

M Mid Atlantic ES Ridge Bouvet Triple Junction ° -50 Southwest AB Indian Ridge

-60° American Antarctic Ridge Enderby Basin ANTARCTICA 0° 10° 20° 30° 40° 50° 60° 70°

Figure 1. Ver tical gradient gravity map (VDGM; see Rotstein et al. 2001) of the SWIR, based on the free-air gravity data of the Geosat and ERS1 satellites (Sandwell & Smith 1997) showing in detail the transform fault and fracture zone pattern of the SWIR. Also marked on the map are the locations of the bathymetric features discussed in the text. AB = Andrew Bain FZ; M = Marion FZ; PE = Prince Edward FZ and ES = Eric Simpson FZ.

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 767

Using Seasat data was clearly an important step towords improv- data coverage is not uniform throughout the region and particularly ing the details of the tectonic fabric of the world ocean floor (e.g. south of the SWIR, in the Enderby Basin, new data were required to Sandwell & Schubert 1982; Gahagan et al. 1988). However, the qual- better constrain the reconstruction models [Fig. 2, and also compare ity of the maps at that time was not sufficient to reveal the details with fig. 1 in Royer et al. (1988)]. of the complex pattern of fracture zones associated with the SWIR In the IODCP database, all the magnetic anomaly profiles were because of their dense distribution and nonlinear nature. Since then, treated by removing the appropriate IGRF (Mandea et al. 2000) and the Geosat and ERS1 altimeter missions enabled the construction by filtering out the long wavelength anomalies. Only clear anomalies of a uniform and dense 2 gravity anomaly grid (McAdoo & Marks were picked, wherever possible, at intervals of about 10 Ma. These 1992; Sandwell & Smith 1997). The 2 satellite-derived free air grav- include for the Late Cretaceous—Present, anomalies 1, 3y, 5o, 6o, ity map now unveils many details of the structure and segmentation 13y, 18o, 20o, 21o, 22o, 23o, 24o, 25y, 26o, 29o, 31y, 32y, 33y, 33o of the SWIR (Marks et al. 1993; Fig. 1). All the fracture zones that and 34y (Fig. 2 gives the chron/age correspondence; y and o denote were previously observed as bathymetric features (Fisher & Good- the young and old boundaries of the normal magnetization period, willie 1997) are apparent on this gravity map and can now be traced respectively). with significantly more detail and accuracy. Additional small frac-

ture zones can now be recognized between the main fracture zones, Satellite altimetry data Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021 achieving a higher resolution of bathymetric detail. Since this work was done, a 1 satellite grid became available but is not expected to Recent satellite altimetry data (Sandwell & Smith 1997), which change the results that are presented in this work. combine Geosat and ERS1 data, represent a significant improve- The availability of this new data set and the observation that the ment as compared to the bathymetric data, mainly due to the uni- details of the accretion on the SWIR since the Late Cretaceous can be form and close spacing of the satellite tracks. This is particularly better resolved prompted this study.For example, with the new data it true at high latitudes, where satellite tracks become denser and ship- became apparent that, although the main evolution stages described board data scarcer. The limitation of satellite data is that the higher previously (Patriat et al. 1985; Patriat & S´egoufin1988; Royer et al. frequency variations in the gravity field are less apparent than the 1988) are correct, the flow lines that result from previously published low-frequency anomalies. In order to somewhat reduce this prob- poles are at places oblique to the fracture zones. This work attempts lem, we amplified the higher frequencies in the data by computing, to refine the available spreading models of the SWIR since the Late in the frequency domain, the vertical derivative of the gravity map Cretaceous. It uses a more complete set of magnetic anomalies than (VDGM; Rotstein et al. 2001). In the VDGM map (Fig. 1), we was previously available. More importantly, it uses the details of picked the troughs, since comparison with the bathymetry shows the fracture zone geometry as flow-lines in order to constrain the that they depict the fracture zone valleys. We note that although spreading history of the SWIR. The revised rotation parameters are many features already stood out in the new compilation of Sandwell generally similar to those of Patriat & S´egoufin(1988) and Royer & Smith (1997), some details appear to be clearer in our repre- et al. (1988), but better account for the details of fracture zone sentation of the data (Fig. 1). This is particularly important for the geometry. This is particularly true between anomalies 33 (76.3 Ma) SWIR because of its numerous short ridge segments and closely and 23o (51.7 Ma), where the details of the SWIR spreading were spaced fracture zones. In particular, the details of the fracture zones previously not resolved as well as they were for the younger period. that were formed during the period of complex spreading, between In this paper, after a brief review of data and methods, we present anomalies 33 and 23, clearly stand out. and discuss a set of eight new reconstructions that describe the main stages of the evolution of the Southwest Indian Ocean. We also use Computation of kinematic phases the detailed gravity data to discuss the effect that the changes in the Rotation parameters (coordinates of the rotation pole and angle of spreading direction have on the SWIR segmentation. rotation) that describe each kinematic phase were computed with the method of Chang (1987). This method determines rotation pa- MAGNETIC ANOMALIES rameters by minimizing the misfit between two pairs of homologous This study uses the IODCP (International Indian Ocean Compilation intersections of fracture zones with magnetic anomaly lines on con- Project) database of magnetic anomaly crossings for the Southwest jugate plates. Five poles of rotation at 96 Ma and at anomalies 34y Indian Ocean (Sclater et al. 1997). This database is a compilation of (83 Ma), 33 (76.3 Ma), 23o (51.7 Ma) and 18o (40.1 Ma) were numerous data sets, and includes older data not available to previous computed with this method (Table 1). During the period between studies of the SWIR. As noted by Royer et al. (1988), the magnetic anomalies 33y (73.6 Ma) and 23o, in which the rotation parameters

Table 1. Rotation parameters for the main phases of SWIR history and intermediate poles of the oblique spreading period. The rotation parameters are calculated using (1) Chang’s method (1987) and (2) a least square fit of fracture zones trends for the period of complex spreading. Angles are positive for counterclockwise rotation, and motions are relative to a fixed Africa. Confidence regions for each computed pole are shown in Fig. 11. Time Anomaly Pole latitude Pole longitude Rotation Method Royer et al. (1988) Patriat & S´egoufin(1988) (Ma) (◦N) (◦E) angle (◦) 40.1 18o 13.6 −41.4 7.47 1 15.3 −50.4 6.92 51.7 23o 8.5 −40.8 10.01 1 6.7 −40.6 9.97 10.4 −44.4 9.72 63.1 28 11.3 −49.6 11.11 2 0.6 −39.2 11.32 71.1 32y −1.2 −42.4 12.38 2 −1.8 −41.4 13.47 75.5 Prior to 33y −4.0 −40.9 14.03 2 76.3 33 −4.6 −40.6 14.39 1 −4.7 −39.7 16.04 83 34y −1.3 −34.7 17.78 1 −2.0 −39.2 17.85 96.0 Prior to 34 3.1 −38.5 26.50 1

C 2005 RAS, GJI, 162, 765–778 768 A. Bernard et al.

AFRICA -20° M20 M15 M10 MADAGASCAR 18o 6o M3 20o M1 M0 24o 6o -30° 96 Ma

24o 18o 34y 20o 33o 20o Magnetic Anomalies 18o 34y 83 Ma 33y 6o -40° 18o 33o 79.1 Ma 20o 33y 73.6 Ma Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021 24o 34y 32y 71.1 Ma 33o 18o 31y 67.7 Ma 20o 29o 64.7 Ma 24o 18o 26o 57.9 Ma 20o 25y 55.9 Ma A33y ° 33o 24o 53.3 Ma -50 6o 34y 23o 51.7 Ma 22o 49.7 Ma 6o 18o 21o 47.9 Ma 20o 20o 43.8 Ma 24o 18o 40.1 Ma 33o 13y 33.1 Ma -60° 34y 96 Ma 6o 20.1 Ma 5o 10.9 Ma M0 3y 5.9 Ma M1 M3 1 0.8 Ma M10

M15 ANTARCTICA 0° 10° 20° 30° 40° 50° 60° 70°

Figure 2. Magnetic anomalies identified along the SWIR, from present time to anomaly 34y (83 Ma), based on the IODCP database (Sclater et al. 1997). Detailed presentations of the magnetic picks in five sub-regions are presented in Fig. 3.

underwent large changes, the scarcity of well-identified magnetic region started before, we try extending the kinematic reconstruc- anomalies precludes the definition of conjugate points. However, tions into the magnetic quiet zone. We identify the early fracture as the fracture zones are well identified on the VDGM, it is still zones in the magnetic quiet period (Figs 2–4) and find that they possible to compute the location of the pole of rotation, but not the extend beyond anomaly 34y without a noticeable change in their rotation angle. For this interval we compute three poles at about trend or shape. Farther into the magnetic quiet zone, the fracture anomalies 28 (63.1 Ma), 32y (71.1 Ma) and at 75.5 Ma, prior to zones show a few degree change in trend, indicating a modifica- anomaly 33y. We first reconstruct the SWIR and its fracture zones tion of spreading parameters. We choose the point where the trend at anomaly 23o using the Chang (1987) method. Each of the three changes as the oldest reconstruction model, assuming that it is rea- pole locations is then computed by minimizing the misfit between sonably safe to extrapolate up to this point, using the spreading the flow lines that result from that pole and the corresponding con- parameters of anomaly 34y. The age of the change in the trend jugate fracture zones. This procedure is nonlinear and we use a can be approximated by extrapolating the well-determined spread- mean square criterion to minimize the sum of distances between the ing parameters of anomalies 34y and 33, assuming that the earlier identified portions of fracture zones and the associated flow lines spreading had approximately the same characteristics. An age of (Le Pichon et al. 1973). The angle of rotation is then estimated 96 Ma is estimated in this way for the cusp in the fracture zones, from the length of each portion of fracture zone. The correspond- and the plate reconstruction for this age is shown in Fig. 4. The ing time intervals are constrained by the few magnetic anomalies proximity in time to anomaly 34 (Fig. 5) and the continuity of the closest to the boundaries of each of these portions of fracture zones fracture zones from the cusp to anomaly 33, both appear to support (Table 1). the validity of this age approximation. 96 Ma also corresponds to a marked change in spreading parameters in the Southeast Indian Ocean (M¨uller et al. 2000), which is likely to have left a notice- RESULTS AND DISCUSSION able mark in the adjacent SWIR, such as the change in its trend of spreading. Main stages in the spreading history of the SWIR Prior to anomaly 33y, our model is in agreement with that of Anomaly 34y (83 Ma) is the oldest identified anomaly before the Royer et al. (1988). As noted by these authors, this was a period magnetic quiet period (Figs 2 and 3). However, as spreading in the of intermediate spreading rate and we calculate a half spreading

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 769

25° 30° 35° 40° 45°

34y ° ° -35 33o -35 34y 33y -45° -45° 34y 32y 31y 33o 33o 33y 32y -40° -40° 55° 60° 24o 11y 25y 24o 22o 20o ° 21o ° -50 18o -50 -30° -30° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-45° -45° 35° 40° 45°

25° 30° 20o 30° 35° 40° ° ° 15 20 18o -35° 20o -35° -45° ° 21o 5o -45

5o

6o 6o 18o 24o 20o 13y 25y -55° -55° 18o 55° 60° 18o 20o 20o 21o 21o ° 22o -50 24o -50° 18o 24o 27y 20o 20o 29o 32y 32y 33y 29o 31y 33y 33o 32y Magnetic Anomalies

33y 33o 34y 83 Ma 22o 49.7 Ma 33o 79.1 Ma 21o 47.9 Ma -60° -60° 34y 33y 73.6 Ma 20o 43.8 Ma 33o 32y 71.1 Ma 18o 40.1 Ma 34y -55° -55° 34y 31y 67.7 Ma 13y 33.1 Ma 29o 64.7 Ma 6o 20.1 Ma 30° 35° 40° 26o 57.9 Ma 5o 10.9 Ma 25y 55.9 Ma 3y 5.9 Ma 24o 53.3 Ma 1 0.8 Ma 15° 20° 23o 51.7 Ma

Figure 3. To enable a more detailed presentation of the magnetic crossings, the SWIR was divided into five subregions. Note the scarcity of anomalies for most of the SWIR, between anomalies 33 and 23, during the period of complex spreading.

velocity of about 3 cm yr−1.Wefind that the change in spreading Complex spreading ended at anomaly 23o (Fig. 8) and slow parameters, leading to the well-known period of slow and complex spreading continued, but with the present direction. This period spreading at the SWIR, started between anomaly 33o and 33y (Figs 2 appears to be contemporaneous with the period of increased ve- and 3). This is somewhat earlier than what was previously suggested locity of India with respect to Antarctica (Fig. 9) and the related (Royer et al. 1988) and being a major stage in the evolution of rapid migration of the Rodrigues triple junction eastward in the ref- the Southwest Indian Ocean, an updated reconstruction model for erence frames of both Africa and Antarctica (Royer et al. 1988; anomaly 33 (76.3 Ma) is presented in Fig. 6. Dyment 1993). The rapid increase in velocity of India with respect For the complex plate motion between anomalies 33 and 23o we to Antarctica started between anomalies 33 and 32o. The sudden propose a set of three poles (Table 1). These are generally simi- slowing down of India at anomaly 23 is associated with the onset of lar to the previously published poles of Royer et al. (1988) but the the collision between India and Eurasia (Patriat & Achache 1984), new flow lines fit better the fracture zone traces (Fig. 7). The new ending at anomaly 20 with welding of the two plates. These major pattern of spreading indicates that the complex spreading period events are likely to have caused the changes in spreading parameters begins with a gradual change of spreading direction and ends with along the adjacent SWIR. an abrupt change. In the absence of good magnetic anomaly control The long period between anomaly 23o and the present can be between anomalies 33 and 23o, the details of the fracture zones be- further divided into two phases. The first one, between anomaly 23o come crucial in resolving the spreading history for which a constant and anomaly 18o (40.1 Ma), is expressed by a series of short but spreading rate was assumed. The corresponding kinematic model distinct segments of the fracture zones, which exhibit a different is consistent with the few magnetic anomalies from this spreading trend than at younger ages (Fig. 3). These are apparent only near the period, and the predicted flow lines generally match the details of Rodrigues triple junction because of the relative position of the poles the complex fracture zones (Fig. 7). of rotation during this period. However, a change in the spreading

C 2005 RAS, GJI, 162, 765–778 770 A. Bernard et al.

30° 40° 50° -20° 96 Ma

-30° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-40°

-20° M20 M15 INDIA AFRICA M10 MAD.

M3 M1 M0 -30°

Rodrigues Triple Junction

Discovery

Bouvet Triple Junction M0 M1 M3 -40° M10 Andrew Bain

M15

ANTARCTICA ANTARCTICA

30° 40° 50°

Figure 4. Reconstruction of the Southwest Indian Ocean at 96 Ma. Rotation parameters for the India closure are 44.8◦N, 207.1◦E and 27.2◦ (Rotstein et al. 2001). Madagascar is in its present location relative to Africa.

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 771

20° 30° 40° -20° Chron 34, 83 Ma

-30° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-40°

-50°

-20° M20 M15 INDIA M10 AFRICA MAD.

M3 M1 M0 -30° Rodrigues Triple Junction

Discovery Bouvet -40° Triple Junction

Andrew Bain

Magnetic Anomalies

M0 34y 83 Ma M1 M3 M10

-50° M15 ANTARCTICA

20° 30° 40°

Figure 5. Reconstruction of the Southwest Indian Ocean at anomaly 34y (83 Ma). Rotation parameters for the India closure are 7.8◦N, 10.9◦E and 65.1◦ (Royer & Sandwell 1989). C 2005 RAS, GJI, 162, 765–778 772 A. Bernard et al.

10° 20° 30° 40° 50° 60° -20° Chron 33, 76.3 Ma

-30°

-40° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-50°

-60° -20° AFRICA M20 INDIA M15

M10 MAD. AFRICA

M3 M1 M0 -30° 96 Ma

Rodrigues Indomed Triple Junction

33o 34y

-40°

34y Andrew Bain

Discovery 33o Bouvet 96 Ma Triple Magnetic Anomalies Junction 34y 83 Ma M0 33o 79.1 Ma -50° M1 M3 M10

M15

ANTARCTICA

-60° 10° 20° 30° 40° 50° 60°

Figure 6. Reconstruction of the Southwest Indian Ocean at anomaly 33 (76.3 Ma), the start of the period of complex spreading phase. Note the continuity of the fracture zones through anomaly 34y, without any change in the trend. A change in the trend is seen further into the basins, indicating a different pole of rotation. We estimate the time of the change to be 96 Ma (see text).

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 773

2a. This paper

1a.

3a.

-20° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021 Rodrigues AFRICA MAD. Triple Junction 3

-30°

2 -40°

Bouvet 1 -50° Triple Junction

-60°

ANTARCTICA 0° 10° 20° 30° 40° 50° 60° 70° 2b. 1b. 3b.

Patriat et al. (1985) Royer et al. (1988)

Figure 7. In the centre, a comparison of the detailed fracture zones (grey lines) with the flow lines (black lines with tick marks) computed using the rotation parameters suggested in this work and shown in Table 1. Small figures [1(a)–3(a)] above the map are enlargements of the corresponding areas in the map. Small figures [1(b)–3(b)] below the maps are the enlargements of the same areas but using the flow lines that result from the rotations of Patriat et al. (1985) for anomalies younger than 20, and Royer et al. (1988) for older times. Note the obliquity of the flow lines to the fracture zones in 1b–3b and the irregular nature of the flow lines in 1b–2b as compared with the same in 1a–3a. pattern in the Southwest Indian Ocean, some time between anoma- We have not been able to identify magnetic anomaly 19 and lies 21 and 15, has been previously suggested (Bergh & Norton we start the subsequent phase at anomaly 18o (40.1 Ma) and pro- 1976) from magnetic anomalies in the Mozambique Basin. Our pose that it lasts to the present time (Fig. 10). Seafloor spreading separate pole for this period roughly corresponds to the period of during this entire period can be modelled with the single pole of collision between India and Eurasia. Patriat & S´egoufin(1988), but a small modification of the pole

C 2005 RAS, GJI, 162, 765–778 774 A. Bernard et al.

10° 20° 30° 40° 50° 60° ° -20 Chron 23o, 51.7 Ma

-30°

-40° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-50°

-60°

-20° M20 Rodrigues M15 Triple M10 MAD. Junction AFRICA

M3 M1 M0 24o -30° Gallieni 96 Ma 24o

Indomed

Atlantis 34y Discovery 33o 33y

33y -40° 33o 34y Magnetic Anomalies

34y Andrew Bain 34y 83 Ma Bouvet 33o 33o 79.1 Ma

Triple 33y 33y 73.6 Ma Junction 24o 32y 71.1 Ma 31y 67.7 Ma 33y 24o -50° 33o 96 Ma 29o 64.7 Ma 34y 26o 57.9 Ma M0 25y 55.9 Ma M1 24o 53.3 Ma M3 M10

M15 ° -60 ANTARCTICA

10° 20° 30° 40° 50° 60°

Figure 8. Reconstruction of the Southwest Indian Ocean at anomaly 23o (51.7 Ma), the end of the period of complex spreading and the return to normal spreading. Note that the larger and older transform faults (Andrew Bain, Marion, Prince Edward, Eric Simpson) turned during the phase of complex spreading into wide fracture zones, each including several distinct fractures. A number of new fracture zones also originated at this period in the old crust. Also, note the migration of the Rodrigues triple junction eastward, in the reference frames of both Africa and Antarctica, since anomaly 33, resulting in the rapid growth of the SWIR.

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 775

Opening rate : 220 Southeast Indian Ridge Southwest Indian Ridge Spreading direction : 200 Southwest Indian Ridge

180 30

160 spreading direction (degrees N) 20 140

10 120 Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

100 0

Full opening rate (km/Ma) 80 -10

60 -20

40 -30 20

age (Ma) 0 40 45 50 55 60 65 70 75 80 85 20o 21o 22o 23o 24o 25y 26o 27y 29y 29o 31y 32y 33y 33o 34y

Figure 9. A comparison between the full opening rate of India with respect to Antarctica (Southeast Indian Ridge) during the period of complex spreading on the SWIR, and the direction and full opening rate of Africa with respect to Antarctica (SWIR). Error bars reflect the standard deviations associated with the mean values computed from several different ridge segments.

parameters allows us to better follow the fracture zone geometry extension as they are in the newly created part of the SWIR result- (Fig. 7). ing from the easterly migration of the triple junction. In contrast, the VDGM map (Fig. 1) shows that the change in the direction of SWIR spreading was accompanied by the appearance of multiple Restraining and releasing bends on the SWIR fracture zones across much of the SWIR, replacing the distinct, sin- transform faults gle trace transform faults. Some 27 new transform faults appeared As Menard & Atwater (1969) first pointed out, the change in the between 23◦E and 43◦E, instead of the 10 existing faults at the start direction of plate motion will induce a component of either ten- of change in the spreading during anomaly 33. This process ap- sion or compression into an existing transform fault. These effects, pears to be particularly apparent in the larger offset transform faults which are commonly known as releasing and restraining bends on along the SWIR. In the Andrew Bain transform fault, which was the strike-slip faults, are determined by the sense of rotation coupled largest transform fault associated with the SWIR, at least five new with the direction of relative motion across the transform fault. In closely spaced segments appeared with an average width of some 80 the particular case of the SWIR, which is characterized by left lat- km. This is consistent with the observation (Tucholke & Schouten eral transform faults, extension will result from a counterclockwise 1989) that, for a constant angle of rotation, the larger the initial rotation in the direction of sea floor spreading, and compression by fracture offset is, the wider the transform fault zone becomes. Un- a clockwise rotation. In the period of complex spreading, charac- fortunately, in the absence of sufficient identified magnetic anomaly terized from a counterclockwise rotation of the spreading direction, crossings between anomalies 33 and 23o, it is impossible to deter- a component of extension must have affected the SWIR transform mine the individual offset associated with any of these many trans- faults. Of course, this effect will occur only in older crust and pre- form segments. We note that although this phenomenon is clearly existing transform faults. At the same time, new transform faults that more evident in the large offset Andrew Bain transform, it is not developed in a newly added part of the ridge will not be affected limited to the largest transform faults. This observation is in con- by this extension. For example, Gallieni and Melville fracture zones trast to Ligi et al. (2002) who suggested that broad transforms are that were created at the onset and the end of this period, respectively only those with a large offset and with an age offset greater than (Dyment 1993; Sauter et al. 1997), did not experience transverse 20 Ma.

C 2005 RAS, GJI, 162, 765–778 776 A. Bernard et al.

10° 20° 30° 40° 50° 60° -20° Chron 18o, 40.1 Ma

-30°

-40° Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021

-50°

-60°

-20° M20 M15 M10 MAD. AFRICA Melville 20o M3 Atlantis M1 20o M0 24o ed Gallieni -30° 96 Ma 24o

Indom Rodrigues 34y Triple Discovery 24o Junction 33o 20o 33y Magnetic Anomalies -40° 34y 83 Ma 20o 20o 33y 33o 79.1 Ma 33o 34y 33y 73.6 Ma 33o 34y 32y 71.1 Ma Bouvet 33y 31y 67.7 Ma Triple 29o 64.7 Ma Junction 24o 20o 26o 57.9 Ma Andrew Bain -50° 20o 25y 55.9 Ma 24o 33y 24o 53.3 Ma 33o 96 Ma 23o 51.7 Ma 34y 22o 49.7 Ma M0 47.9 Ma M1 21o M3 20o 43.8 Ma M10 -60° M15

ANTARCTICA

10° 20° 30° 40° 50° 60°

Figure 10. Reconstruction of the Southwest Indian Ocean at anomaly 18o (40.1 Ma), approximately at the end of the collision between India and Eurasia and reorganisation of spreading in the Southeastern Indian Ocean. Note, by comparing Figs 8 and 10, the decrease in the rate of the eastwards migration of the Rodrigues triple junction since anomaly 23. The period between anomalies 23 and 18 is characterized by the formation of numerous small, new transform faults close to the Rodrigues triple junction.

C 2005 RAS, GJI, 162, 765–778 Refined spreading history at the Southwest Indian Ridge 777

60°W 55°W 50°W 45°W 40°W 35°W 30°W 25°N 25°N

20°N 20°N

15°N 18o 15°N

28

10°N 23o 10°N

5°N 5°N Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021 prior to 34

0 32y 0 34y

prior to 33y 5°S 5°S

33

10°S 10°S 60°W 55°W 50°W 45°W 40°W 35°W 30°W

Figure 11. 95 per cent confidence region computed for each pole of rotation. Five poles (for 18o, 23o, 33, 34y and prior to 34) were calculated using the method of Chang (1987) yielding contour maps with an interval of 0.1◦ of the upper value (solid line) and lower value (shaded line) of the rotation angle versus latitude and longitude. Three poles (for 28, 32y and prior to 33y) were calculated for the period of complex spreading using the least square fit of fracture zone portions, which allows to calculate only the boundary of the 95 per cent confidence region (the confidence region is shaded).

CONCLUSIONS Chang, T., 1987. On the statistical properties of estimated rotations, J. geo- phys. Res., 92, 6319–6329. We find four main spreading periods at the SWIR since 96 Ma, each Dyment, J., 1993. Evolution of the Indian Ocean Triple Junction between 65 characterized by a noticeably different trend in the fracture zones: and 49 Ma (anomalies 28 to 21), J. geophys. Res., 98, 13 863–13 877. between 96 Ma and anomaly 33 (76.3 Ma); the complex spreading Fisher, R.L. & Sclater, J.G., 1983. Tectonicevolution of the Southwest Indian period between anomalies 33 and 23o (51.7 Ma); between anomalies Ridge since the Mid-Cretaceous: plate motions and stability of the pole 23o and 18o (40.1 Ma) and between anomaly 18o and the present of Antarctica/Africa for at least 80 Ma, Geophys. J. R. Astron. Soc., 73, time. A new rotation pole can be attributed to each of these periods, 553–576. except for the complex spreading period where several poles are Fisher, R.L. & Goodwillie, M.A., 1997. The physiography of the Southwest needed to describe the variable spreading pattern. The stages in Indian Ridge, Mar. Geophys. Res., 19, 451–455. the development of the SWIR were associated with changes in the Gahagan, L.M. et al., 1988. Tectonic fabric map of the ocean basins from satellite altimetry data, Tectonophysics, 155, 1–26. regional spreading pattern. These include such major events as the Le Pichon, X., Francheteau, J. & Bonnin, J., 1973. Plate Tectonics, 309 p. start of a rapid separation of India from Antarctica and the rapid Elsevier, New York. migration of the Rodrigues Triple Junction (77 Ma), and the onset Ligi, M., Bonatti, E., Gasperini, L. & Poliakov, A.N.B., 2002. Oceanic broad and end of the collision of India with Eurasia (approximately 52 multifault transform plate boundaries, Geology, 30, 11–14. and 44 Ma, respectively). The period of complex spreading between Livermore, R.A. & Hunter, R.J., 1996. Mesozoic seafloor spreading in the anomalies 33 and 23o is characterized by the appearance of many Southern WeddellSea, in WeddellSea Tectonics and Gondwana Break-up, new transform faults in the old parts of the SWIR. Vol. 108, pp. 227–241, eds Storey, B.C., King, E.C. & Livermore, R.A., Geol. Soc., London. Mandea, M. et al., 2000. International geomagnetic reference field-2000, ACKNOWLEDGMENTS Phys. Earth planet. Int., 120, 39–42. The authors thank TotalFinaElf for financial support for this work. Marks, K.M. & Tikku, A.A., 2001. Cretaceous reconstructions of East Antarctica, Africa and Madagascar, Earth planet. Sci. Lett., 186, 479– This is EOST contribution No. 2005.01-UMR7516. 495. Marks, K.M., McAdoo, D.C. & Smith, W.W.F., 1993. Mapping the south- REFERENCES west Indian Ridge with Geosat, EOS Transactions, American Geophysical Union, Washington DC, USA, 74, 81–86. Bergh, H.W. & Norton, I.O., 1976. Prince Edward fracture zone and the Martin, A.K. & Hartnady, C.J.H., 1986. Plate tectonic development of the evolution of the Mozambique Basin, J. geophys. Res., 81, 5221–5239. Southwest Indian Ocean: a revised reconstruction of East Antarctica and Cande, S.C. & Kent, D.V., 1995. Revised calibration of the geomagnetic Africa, J. geophys. Res., 91, 4767–4786. polarity timescale for the Late Cretaceous and Cenozoic, J. geophys. Res., McAdoo, D.C. & Marks, K.M., 1992. Gravity fields of the Southern Ocean 100, 6093–6095. from Geosat data, J. geophys. Res., 97, 3247–3260.

C 2005 RAS, GJI, 162, 765–778 778 A. Bernard et al.

Menard, H.W. & Atwater, T., 1969. Origin of fracture zone topography, Royer, J.Y., Patriat, P., Bergh, H.W. & Scotese, C.R., 1988. Evolution of the Nature, 222, 1037–1040. Southwest Indian Ridge from the Late Cretaceous (anomaly 34) to the M¨uller, R.D., Gaina, C., Tikku, A., Mihut, D., Cande, S.C. & Stock, J.M., Middle Eocene (anomaly 20), Tectonophysics, 155, 235–260. 2000. Mesozoic/Cenozoic tectonic events around Australia, Monograph Sandwell, D.T. & Schubert, G., 1982. Lithosphere flexure at fracture zones, 121, The History and Dynamics of Global Plate Motions, pp. 161–188 J. geophys. Res., 87, 4657–4667. American Geophysical Union, Washington, DC. Sandwell, D.T. & Smith, W.W.F.,1997. Marine gravity anomaly from Geosat Norton, I.O. & Sclater, J.G., 1979. A model for the evolution of the Indian and ERS1 altimetry, J. geophys. Res., 102, 10 039–10 054. Ocean and the break up of Gondwanaland, J. geophys. Res., 84, 6803– Sauter, D., Mendel, V.,Rommevaux, J-C., Patriat, P. & Munschy, M., 1997. 6830. Propagation of the Southwest Indian Ridge at Rodrigues triple junction, Patriat, P.,1979. L’Oc´eanIndien occidental: la dorsale Sud-Ouest Indienne, Mar. geophys. Res., 19, 553–567. Mem. Mus. Hist. Nat., 43, 49–52. Sclater, J.G., Fisher, R.L., Patriat, P., Tapscott, C.R. & Parsons, B., 1981. Patriat, P. & Achache, J., 1984. India-Eurasia collision chronology has im- Eocene to recent development of the Southwest Indian Ridge: a conse- plications for crustal shortening and driving mechanism of plates, Nature, quence of the evolution of the Indian Ocean triple junction, Geophys. J. 311, 615–621. R. astr. Soc., 64, 587–604. Patriat, P. & S´egoufin,J., 1988. Reconstruction of the central Indian Ocean. Sclater, J.G., Munschy, M., Fisher, R.L., Weatherall, P., Cande, S.C., Tectonophysics, 155, 235–260. Patriat, P., Bergh, H. & Schlich, R., 1997. Geophysical synthesis of Downloaded from https://academic.oup.com/gji/article/162/3/765/2098160 by guest on 13 February 2021 Patriat, P., S´egoufin,J., Goslin, J. & Beuzart, P., 1985. Relative positions the Indian/Southern Oceans: Part I, The Southwest Indian Ocean, of Africa and Antarctica in the Upper Cretaceous: evidence for non- Scripps Inst. Oceanography Ref. Series 97–6,LaJolla, California, USA, stationary behaviour of fracture zones, Earth planet. Sci. Lett., 75, 204– 68 p. 214. Tapscott, C.R.P.,Patriat, P.,Fisher, R.L., Sclater, J.G., Hoskins, H. & Parsons, Rotstein, Y., Munschy, M. & Bernard, A., 2001. The Kerguelen province B., 1980. The Indian Ocean triple junction, J. geophys. Res., 85, 4723– revisited: additional constraints on the early development of the southeast 4739. Indian Ocean, Mar. geophys. Res., 22, 81–100. Tikku, A.A., Marks, K.M. & Kovacs, L.C., 2002. An Early Cretaceous extinct Royer, J.Y. & Sandwell, D.T., 1989. Evolution of the Eastern Indian Ocean spreading center in the northern Natal Valley, Tectonophysics, 85, 87–108. since the Late Cretaceous: constraints from Geosat altimetry, J. geophys. Tucholke, B.E. & Schouten, H., 1989. Kane fracture zone, Mar. geophys. Res., 94, 13 755–13 782. Res., 10, 1–40.

C 2005 RAS, GJI, 162, 765–778