Large along-axis variations in magma supply and tectonism of the Southeast Indian Ridge near the Australian-Antarctic Discordance

1. Introduction

The Oceanic Core Complex (OCC) features were interpreted to be the result of prolonged localized tectonic extension in oceanic lithosphere, resulting in rotated and uplifted foot wall of a normal fault exhibiting corrugated surface (Tucholke and Lin, 1994; Cann et al., 1997; Tucholke et al., 1998; MacLeod et al., 2009). Most of OCC features were observed at slow- (full spreading rates of 20–40 mm/a) or ultraslow-spreading (<20 mm/a) ridges, where the magma supply is relatively low comparing to the fast- (>80 mm/a) and intermediate fast-spreading (40–80 mm/a) ridges. Globally, nearly 200 Oceanic Core Complexes (OCCs) have been identified, most of them were found on slow- and ultraslow-spreading ridges, e.g., the northern Mid-Atlantic Ridge (MAR) and Southwest Indian Ridge (SWIR) (Ciazela et al., 2015). Several OCCs were also found on ridges of moderate spreading rates, e.g., the Southeast Indian Ridge (SEIR) (Christie et al., 1998), the Chile Ridge (Okino et al., 2004), and the Parece Vela backarc basin (Ohara et al., 2001). Noticeably, a cluster of OCCs, including a prominent example of 50-km-long corrugation, were observed in the AAD on the intermediate fast-spreading SEIR (Okino et al., 2004).

Our study region on the SEIR is located at 54°–44°S and 108°–134°E. It has relatively constant full spreading rates of 72–76 mm/a (Weissel and Hayes, 1971) and includes eight spreading segments that are separated by nine prominent transform faults (Fig. 1). Several second-order offsets also exist, which are separated by small transform and non-transform discontinuities. Weissel and Hayes (1971) first divided the SEIR south of Australia into three zones. We adopted the same zone names but modified the zone boundaries as Zone A (~128°–134°E), Zone B (~120°–127°E), and Zone C (~108°–120°E). The overall topography of Zones A and C is shallower than that of Zone B. Zones A and C include a series of short segments separated by propagating rifts and transform faults. Zone B, which is also called the AAD, can be further divided into five segments (B1–B5) from east to west, bounded by the Birubi, Euroka, Lutana, Warringa, Nalkari, and Baroga transform faults, respectively (Hayes and Conolly, 1972; Vogt et al., 1983). The AAD region is characterized by anomalously deep seafloor compared with the average seafloor age-depth relation (Hayes, 1988). About 30 Ma ago, chaotic topography began to develop in the AAD, indicating relatively weak magma supply, which might be caused by cold spot, mantle convective downwelling, and/or pre-Cretaceous remnant slab (Anderson et al., 1980; Klein et al., 1991; Christie et al., 1998; Gurnis et al., 1998). The AAD region also shows prominent axial valleys (Fig. 1), which are typical of slow- and ultraslow-spreading ridges, e.g., the MAR and SWIR (Lin et al., 1990; Macdonald, 1990; Dick et al., 2003).

Figure  1.  Topography map of the study region. The Southeast Indian Ridge (SEIR) and the study region are shown as red lines and red rectangle, respectively, on the regional setting map (inset map). Black dashed lines, black short double lines, and green lines are the main fracture zones, ridge axes, and propagating rifts, respectively. The study region is divided into Zones A, B, and C; the Australian–Antarctic Discordance (AAD) includes 5 segment corridors of B1 to B5; Zones A and C have 3 and 5 segments, respectively. The boundaries are modified from Weissel and Hayes (1971) and Vogt et al. (1983). Red curve is the trace of the Indian/Pacific MORB mantle boundary from Christie et al. (2004). AF represents the African plate,AN the Antarctica plate, and AU the Australian plate.

DownLoad: Full-Size Img PowerPoint

Since the spreading rate is relatively uniform along our study region on the SEIR, other factors such as magma supply are expected to play an important role in shaping the seafloor morphology and tectonic deformation. Buck et al. (2005) modeled seafloor morphology and normal faulting, showing that 95% of the plate separation on the SEIR at 115°E of Zone C is result of magma intrusion (i.e., M=0.95). Behn and Ito (2008) found M=0.7–0.8 on the SEIR at 114.2°–114.3°E of Zone C. However, the detail variations in tectonic deformation along the SEIR and the potential controlling factors are still poorly known.

In this study, we investigated the role of melt supply in the development of morphologic features and crustal structures along the SEIR. We first conducted a detailed analysis of morphology and gravity anomalies. We then developed a new 2D map-view method to quantify the percentage of seafloor surface area occupied by fault scarps and to calculate the percentage of seafloor area created by magmatic accretion (i.e., the M factor). We also identified the exact boundary of individual OCCs and calculated the M factor of these areas. Finally, we analyzed the results of gravity anomalies and M factors together with geochemical constraints. The combination of the results indicates a strong link among cold mantle source, limited magma supply, thin crust, and widespread development of OCCs.

2. Data and methods

2.1 Bathymetry

The topographic data used in this study are the merged datasets of global bathymetry data in 15″ grid and multi-beam bathymetry data where available. The global bathymetry data were obtained from the University of California, San Diego (SRTM15_PLUS Version 1, http://topex.ucsd.edu/marine_topo/). Multi-beam bathymetry data were obtained mostly from the Marine Geoscience Data System (MGDS, http://www.marine-geo.org/portals/gmrt/) and National Geophysical Data Center (NGDC, http://maps.ngdc.noaa.gov/viewers/autogrid/). Part of the multi-beam bathymetry data for Segment B3 were collected by Okino et al. (2004). The resampled grid size of all multi-beam bathymetry data is about 100 m × 100 m. Multi-beam bathymetry data cover several areas including regions at 108°–118°E, 124°–126°E, and a northern box at 126°–128°E. Most of the multi-beam bathymetry data cover regions of more than 50 km in width to both sides of the SEIR axis. The global bathymetry of 15″ grid size was used for areas without multi-beam bathymetry.

2.2 Sediment thickness and crustal age

The sediment thickness data were obtained from the NGDC (http://www.ngdc.noaa.gov/mgg/sedthick/sedthick.html) with 5′ grid size. The sediment layer is relatively thin in our study area with average thickness of ~130 m. The average sediment thickness north of the SEIR is smaller than the conjugate southern flank by ~100 m. The crustal age data were obtained from the NGDC (http://www.ngdc.noaa.gov/mgg/ocean_age/ocean_age_2008.html) with 2′ grid size(Müller et al., 2008).

2.3 Gravity anomalies and crustal thickness

The free-air gravity anomaly (FAA) data were obtained from the global marine gravity model data base with 1′ grid size (Sandwell et al., 2014). We calculated the mantle Bouguer gravity anomaly (MBA) by subtracting from the FAA the gravitational effects of (1) water-sediment; (2) sediment-crust; and (3) crust-mantle interfaces with an assumed 6-km-thick reference crust. The sediment density is assumed to increase with depth (e.g., Wang et al., 2011) from 1 950 kg/m³ to 2 600 kg/m³ with an assumed depth gradient of 15 kg/m³ per 100 m. The densities of water, crust, and mantle were assumed to be 1 030, 2 700, and 3 300 kg/m³, respectively. The Fourier spectrum method of Parker (1973) was used in the calculation.

We further calculated the residual mantle Bouguer gravity anomaly (RMBA) by subtracting from the MBA the gravitational effects of lithospheric cooling (Wang et al., 2011). We assumed a model of 1D vertical plate cooling for the corresponding crustal age, assuming temperature of 1 350°C at the mantle depth of 100 km and thermal expansion coefficient of 3.5×10^–5 °C^–1. The 3D temperature field was then converted into a density anomaly grid by Δρ=ρ₀αΔT, where ρ₀, α, and ΔT are the reference mantle density, thermal expansion coefficient, and temperature anomaly relative to a reference point in the mantle, respectively (Parker, 1973; Kuo and Forsyth, 1988; Lin et al., 1990).

Finally, we calculated an end-member crustal thickness model by downward continuing the RMBA signals to the bottom of the assumed 6-km-thick reference crust. A cosine taper low-pass filter between 25 and 100 km was used to ensure stability in downward continuation (Oldenburg, 1974; Kuo and Forsyth, 1988).

2.4 Estimation of M factors

The M factor is defined as the fraction of total seafloor spreading accommodated by magma accretion (Buck et al., 2005; Behn and Ito, 2008). In this study, we analyzed multi-beam bathymetry to estimate the spatial variations in the M factor. Since the sediment layer is relatively thin (usually <500 m) near the SEIR axis, we neglected the sediment when analyzing faults.

The M factor along an individual ridge-perpendicular profile could be estimated by calculating the cumulative fault heaves along the profile normalized by the profile length: M=1–∑X_f/(∑X_f+∑X_m), where ∑X_f is the cumulative length of fault heaves along the profile, while ∑X_m is the cumulative length not covered by fault heaves (Fig. 2).

Figure  2.  Illustrasion map of M factors calculation. The M factor is calculated from topographic profile, as modified from Smith (2013). The blue stripe (X_f) is the horizontal projection of fault surface. The blue and red (X_m) stripes represent tectonic extension and magmatic accretion regions, repectively. M factor can be evaluated by M=$\frac{{\displaystyle\sum {X_{\rm m}} }}{{\displaystyle\sum {X_{\rm f} + \displaystyle\sum {X_{\rm m}} } }}$. The blue arrow is the moving direction relative to the conjugated fault block. The water, upper crust, and the lower crust/upper mantle are represented by white, light gray, and light brown areas, respectively. Red vertical box represents the ridge axis.

DownLoad: Full-Size Img PowerPoint

We also calculated the M factors in two-dimensional map view. The study region with multi-beam bathymetry data was divided into a mosaic of small boxes. Each box has spatial dimension of ~12 km×7 km and contains several faults. The fault surfaces (i.e., fault scarps) exposed in each box were manually identified and digitized. We calculated the map-view M factors as M=1–∑S_f/(∑S_f+∑S_m), where ∑S_f is the total area of fault surfaces in map view, while ∑S_m is the total area not covered by faults surfaces.

For a certain sliding sampling box K×L, the fault scarp F(i) can be treated as enveloped by curves of h(x) and g(x), while L and K are parallel and perpendicular to the ridge axis, respectively; x is parallel to the ridge axis. The area S(i) can be calculated by Eq. (1):

Dividing Eq. (1) by the box area S=K×L yields:

According to the Mean Value Theorem, there is at least one ξ in domain (0, L) that can satisfy Eq. (3):

The left-hand-side of Eq. (3) is the map view ratio of faults while the right-hand-side of Eq. (3) is a single profile ratio of faults.

We calculated the map-view M factors for all ridge segments in the study region (Fig. 1), except for Segments B1 and B2, where multi-beam bathymetry is lacking. The sampling box size is ~12 km × 7 km, and the sliding step length is ~9 km and 5 km, respectively, along and perpendicular to the ridge axis. The mean map-view M factors of each segment were also calculated. Through calculating the spatial variations in the 2D map view, we could obtain critical information on magma-to-tectonic ratio both along and across the SEIR axis.

2.5 Simulation of lithospheric deformation as a function of M factor

To better understand the causes of the observed large variations in the SEIR tectonics, we conducted geodynamic simulations of lithospheric deformation for average spreading rates similar to that of the SEIR, but varying the M factor. The simulations were carried out using the 2D Fast Lagrangian Analysis Continua (FLAC) that has previously been used to model ridge deformation (Cundall, 1989; Buck et al., 2005; Behn and Ito, 2008; Tucholke et al., 2008; Olive et al., 2010). Our model is 40 km in width and 10 km in depth with grid size of ~ 0.18 km × 0.2 km. The magma is injected and fixed at the center of the ridge axis with a narrow rectangle injection region of ~ 0.18 km × 6 km. The temperatures of the top and bottom boundaries are set to 0°C and 1 300°C, respectively. The heat source associated with magma injection at the ridge axis is added in proportion to the magma injection rate (Behn and Ito, 2008). The spatial variations in the elastic, viscous, and plastic deformation of the crust and mantle depend on the temperature, strain rate, and accumulated plastic strain. The faulting-related plastic yielding follows the Mohr-Coulomb criterion, while the cohesion is assumed to decrease linearly from 44 MPa to 4 MPa when a fault develops, as was required to model the fault formation processes in mid-ocean ridges (Lavier et al., 2000) and subducting plates (Zhou et al.,2015; Zhou et al., 2018; Zhou and Lin, 2018). The thermal diffusion perturbation due to faulting is taken into consideration. The modeling time step is 25 ka, and remeshing occurs when any distorted triangles are greater than 5° (Lavier and Buck, 2002).

We modeled five cases with M factor of 0.2, 0.4, 0.6, 0.8, and 1.0, respectively, to investigate the resultant faulting style characteristics (Fig. 4). After multiple modeling steps, the temperature structure and fault style become steady state, and thus the resultant fault pattern could be considered as representing the specific M factor. In our models, we chose t = 0.7 Ma as the end time for comparison, at which time all simulations have reached steady state.

Figure  4.  Results of FLAC geodynamic simulation for M factors of M=0.2 (a, b), M=0.4 (c, d), M=0.6 (e, f ), M=0.8 (g, h), and M=1.0 (I, j). Left and right columns are the claculated strain rate and temperature, respectively, at the model running time of 0.7 Ma, when the models have reached steady state. Magma injection is imposed at the ridge axis in a dike of 0.18 km wide and at depth of 0–6 km from the seafloor. The simulation grid size is 0.18 km ×0.22 km. The cohesion of the deforming layer is assumed to decrease with plastic strain after yielding. The intial layer cohesion is 44 MPa, and the deformation plane is assumed to lose its conhesion linearly to 4 MPa at deformation offset of ~200 m. Intial axial lithospheric thickness is ~6 km.

DownLoad: Full-Size Img PowerPoint

3. Results

3.1 Variations in M factor

The regional scale lithospheric strength and faulting styles are controlled by the lithospheric thermal structure and stress conditions for a given M factor (Shaw and Lin, 1996; Lavier and Buck, 2002; Buck et al., 2005; Behn and Ito, 2008; Olive et al., 2010). Different faulting patterns could result from varying spreading rate and/or M factor.

The map-view M factors were calculated for all segments of the study region with sufficient multi-beam bathymetry, including Segment A1 (Fig. 3a, b), C1 and C2 (Figs 3c-f), C3 and C4 (Fig. 3g-j), C5 and B3 (Figs 3k–n), and B4 and B5 (Fig. 3o-r). Segments B1 and B2 lack sufficient multi-beam bathymetry and thus were not included in the analysis. While all results (Fig. 3) are shown for a specified sampling box size of ~12 km×7 km, analyses using varying sampling box sizes have yielded generally similar conclusions.

Figure  3.  Map of shaded topography, identified fault scarps, and calculated map-view M factor for Segments A1 ( a, b ), C1 ( c, d ), C2 ( e, f ), C3 ( g, h), C4 ( i, j ), C5 ( k, l), B3 ( m, n ), B4 ( o, p ), and B5 ( q, r ), respectively. For each ridge segment, the shaded topographic maps are based on multi-beam bathymetry data ( a, c, e, g, i, k, m, o and q ). The map-view M factors (MVM, color shades) were calculated using a sliding sampling box of ~12 km ×7 km, while the fault scarps (black stripes) were identified from 3D topography and topographic slopes ( b, d, f, h, j, l, n, p, and r ). Both the white and green double lines are the ridge axes.

DownLoad: Full-Size Img PowerPoint

For ridge segments outside the AAD, including Segments A1 (Figs 3a,b), C1 and C2 (Figs 3c-f), and C3 and C4 (Figs 3g–j), the calculated map-view M factors are in the range of 0.7–1.0 with peak values of ~0.85–0.92. These five segments have similar observed faulting style that appears to resemble the predicted lithospheric deformation pattern by the FLAC numerical simulation for M=0.8 (i.e., Figs 4g-h). For this relatively high M value, faults are predicted to develop successively on both ridge flanks with fault spacing of about several kilometers. Buck et al. (2005) compared the results of FLAC simulations with a topographic profile across the SEIR axis, yielding an estimated M factor of ~ 0.95 for a profile at 115°E within Segment C4. Meanwhile, Behn and Ito (2008) estimated a M factor of ~ 0.7–0.8 for a profile at 114.2°–114.3°E, also within Segment C4. Thus the average values from our map-view M factor analysis are consistent with the previous estimates based on selected profiles across Segment C (Buck et al., 2005; Behn and Ito, 2008).

For ridge segment within the AAD, including Segments C5 and B3 (Figs 3k–n) and B4 and B5 (Figs 3o–r), however, the map-view M factor exhibits a wide range of variations from 0 to 1. The M factors for profiles across the OCCs are calculated to be in the range of ~0.3–0.6.

Meanwhile, for cases of limited magma supply, i.e., M≤~0.6 (Figs 4a-f), the FLAC simulations predict long-lived faults of large fault scarp (tens of kilometers) and relatively small dip angle. These results are consistent with previous FLAC simulations for M factor of 0.3–0.5 (Tucholke et al., 2008). The observed strong tectonism and asymmetrical accretion at the ridge segments within the AAD are similar to that of the MAR at 12.5°–35°N where a single detachment fault can accommodate 50% of the plate separation on one flank of the ridge axis (Escartín et al., 2008). In particular, Segment B4 (Figs 3o-p) and the eastern part of Segment B3 (Figs 3m-n), both of which are located within the AAD, exhibit irregular topography of faulting. Such topographic features are consistent with the faulting pattern predicted by the FLAC simulations for M≤0.6 (Figs 4a-f). Such areas of extensive faulting would facilitate lithospheric weakening and water penetration into the mantle (e.g., Olive et al., 2010; Zhou and Lin, 2018).

Although the fault scarps might be affected by the subsequent geological process, we employed the same approach for calculating the M factor for the entire study region, and thus the relative differences in the M factor between sub-regions are of geological significance. Furthermore, we focused on the variations of the M factors of the AAD in comparison to adjacent segments. Therefore, the uncertainties in the calculation of absolute value of the M factor have little influence on our main results.

In summary, the ridge segments within the AAD (Zone B, Fig. 1) appear to have limited magma supply, as indicated by the smaller map-view M factors than the adjacent segments. Within the AAD, Segment B4 and the eastern part of Segment B3 have the smallest M values, implying that they may be undergoing the strongest tectonic extension in comparison to other segments.

3.2 OCCs and correlation with RMBA

The RMBA varies significantly in the study region (Fig. 5b), indicating strong spatial variations in magma supply. In the flow line direction, the area of more positive RMBA (red shaded areas in Fig. 5b) are much wider for ridge segments within the AAD than the adjacent segments. The off-axis distance of the regions with RMBA≥20 mGal is 2–6 times greater within the AAD than the adjacent segments. Furthermore, the RMBA contours show systematic westward along-axis trends. Such trends are consistent with the westward direction of propagating rifts at Segment A1 (Figs 1 and 5a).

Figure  5.  Map of gravity-derived crustal thickness (a), RMBA of the study region in b correlation between RMBA and M factors (c), and correlation between residual depth and RMBA (d). Small regions with blue enclosed contous in b. indicate OCCs. Black points in (c) represents mean map-view M factor and average RMBA of ridge segments. Horizontal errorbars represent the minmum and maxmum RMBA values within individual ridge segments. Vertical errorbars corespond to 50% confidence interval for fault recognition. Red points in (d) are the aveage RMBA and aveage residual depth from each grid bins (~20 km×20 km) of the study region. Blue points indicate regions containing OCCs with the horizontal and vertical errorbars reflecting the minmum and maxmum values of residual depth and RMBA, repectively.

DownLoad: Full-Size Img PowerPoint

A total of 22 OCCs with typical megamullion structure have been found in the study region with 17 OCCs occurring in Segments B3, B4, and B5 of the AAD, while another 5 OCCs occurring at Segments C4 and C5 of Zone C at 116°–117°E (Fig. 5b). Furthermore, several possible OCCs without obvious megamullion have been found in Zone C. The OCCs occur only in regions of relatively positive RMBA (Fig. 5c), while no OCCs have been found in areas of relatively more negative RMBA.

The OCCs within the AAD have greater size than those of Segment C5. Most OCCs are 5–20 km in length along the flow line direction. The largest OCC, which is located at the inside corner of Segment B3, extends by ~50 km in the flow line direction. The RMBA values of OCCs are in the range of 20–70 mGal with an average of ~40 mGal. The relatively positive RMBA values of the OCCs, which might result from thinner crust and/or exposed upper mantle, are consistent with previous studies at the MAR (e.g., Cann et al., 1997; Tucholke et al., 1998) and global settings (Tucholke et al., 2008). A refraction seismic experiment in Segment B4 (Kojima et al., 2003) revealed crustal thickness of ~3.6 km, which is consistent with our gravity-derived crustal thickness model of ~3.0 km (Fig. 5a).

The AAD region is deeper by ~0.8–1.0 km than the adjacent ridge segments (Hayes, 1988; Mark et al., 1990; Klein et al., 1991; Pyle et al., 1992). The stretched, rotated, and uplifted footwall of the OCCs might also cause relatively shallow seafloor and large topographic gradients. We calculated the residual bathymetry by subtracting the effect of thermal subsidence and sediment loading from the observed bathymetry to illustrate if the regions with OCCs were elevated by detachment faults. The OCCs have ~2.5–3.5 km of residual depth (Fig. 5d). The topographic variations within the individual OCCs are in the range of 1.0–2.5 km. In comparison, the topographic variations within typical abyssal hills are in the range of only 0.1–0.5 km (e.g., abyssal hills in Segments C1, C2, C3 and A1). Only very few abyssal hills may reach topographic variations of ~1 km (e.g., part of abyssal hills in Segment C4). The observed topographic variations within the OCCs are similar to the predictions by the FLAC simulations for moderate to small M factors (Fig. 4). The RMBA and topography of Segments B1 and B2 are similar to that of Segments B3, B4, B5, and C5. It is possible that OCCs might exist in Segments B1 and B2 of the AAD, but we cannot identify them yet due to the lack of high-resolution bathymetry data.

In summary, the mean map-view M factors show inverse correlation with the average RMBA values, i.e., the M factor decreases with increasing RMBA with a gradient of –0.003 mGal^–1 (Fig. 5c). The inverse correlation between the observed M factor and RMBA implies that the limited magma supply of the AAD, as reflected in thinner crust, is related to stronger tectonism, as reflected in extensive faulting and widespread occurrence of OCCs.

4. Discussion

4.1 Comparison with geochemical indicators

Multiple geochemical indicators can be used to infer mantle source compositions and partial melting processes beneath mid-ocean ridges (e.g., Klein and Langmuir, 1987; Klein et al., 1988, 1991). For example, the initial depth and average degree of partial melting can be inferred from Fe_8.0 and Na_8.0, which are calculated by normalizing the measured wt% FeO and Na₂O values, respectively, to 8.0 wt% MgO (Klein and Langmuir, 1987). We plotted 69 sets of Na_8.0 and Fe_8.0 data (Klein et al., 1988, 1991; Mahoney et. al., 2002) and compared the axial topography, RMBA, M factors, and geochemical indicators of Na_8.0 and Fe_8.0 to explore the geological processes at the SEIR (Fig. 6).

Figure  6.  Along-ridge-axis variations in seafloor depth (a), RMBA (b), Na_8.0 (c), Fe_8.0 (d), and map-view M factor (e) along the SEIR in the study region. Black arrows in (a) indicate location of transform faults and non-transform discontinuties. Light gray dash lines in (c) and (d) are envelope lines of geochemical data points from Klein et al. (1989, 1991) and Mahoney et al. (2002). Black dots in (e) indicate the mean map-view M factors of ridge segments.

DownLoad: Full-Size Img PowerPoint

Several positive correlations were found in the SEIR datasets: (1) Topography vs. Fe_8.0; (2) Topography vs. M factor; (3) Fe_8.0 vs. M factor; and (4) RMBA vs. Na_8.0. Meanwhile, RMBA and Na_8.0 exhibit inverse correlations with topography, Fe_8.0, and M factor (Fig. 6). The local extreme values of Na_8.0 and Fe_8.0 in Segment C5 (~116°E) and the AAD (~123°E) indicate shallower initial melting depth and lower partial melting extent than adjacent ridge segments. This is consistent with models of limited magma supply for these two regions, as reflected in the overserved relatively positive RMBA and deeper seafloor in Segment C5 and AAD (Fig. 6).

4.2 Causes and consequences of limited magma supply at the AAD

Although OCCs could sometime develop in regions of relatively high magma intrusion rate (e.g., Escartín et al., 2003; Olive et al., 2016), most OCCs do develop in amagmatic regions, e.g., at the slow-spreading MAR and ultraslow-spreading SWIR. Results of the above analysis revealed that the AAD is indeed such a region of limited magma supply even though it is located on an intermediate fast-spreading ridge.

Several hypotheses were proposed to explain the mantle and seafloor depth anomaly at the AAD, including “cold spot”, “mantle convective downwelling”, and “pre-Cretaceous remnant slab” (Hayes, 1976; Anderson et. al., 1980; Klein et al., 1991; Gurnis et al., 1998; Gurnis and Müller, 2003). Melting is more difficult in depleted or cold mantle, which could lead to less magma supply and thus thinner crust and stronger tectonism. Combination of the multiple geological, geophysical, and geochemical constraints leads to the following interpretation of the systematic variations in magma supply and tectonism of the SEIR near the AAD:

(1) Relatively cold or depleted mantle might exist beneath the AAD. Thus, partial melting beneath the AAD might be at shallower initial depth and of lesser extent than that of Zones A and C.

(2) Under the relatively constant spreading rates of the SEIR, the limited partial melting beneath the AAD would lead to insufficient melt supply and thinner than normal crustal thickness.

(3) The relatively magma-starved state of the AAD could lead to an increase in tectonism through normal faulting. At sites of extreme low local magma supply, OCC features would then develop either at isolated locations or as a cluster of OCCs as observed at the AAD.

4.3 Westward migration of the OCC Zone

Several lines of evidence suggest that the observed zone of concentrated OCCs at the AAD might have migrated westward: (1) The westward spindle-shaped envelopes of relatively high values of RMBA; (2) The relatively wider spread of the OCCs in the AAD than at Segment C5; (3) The westward-pointing depth anomaly; (4) the westward-pointing propagating rifts; (5) The westward-pointing magma-rich region at Segment B5 (Fig. 7a).

Figure  7.  Illustration of the OCC development and regional tectonics. a. Mjor tectonic elements of the study region on top of the RMBA map. b. Simplified illustration of OCC development, illustrating the developing tendency of the new OCC region (brown area with arrows) and magma-rich (red area with arrows) regions, respectively. c Illustration of the axial mantle underlying the AAD. Red and long blue curves in (a) are the inferred boundary between the Indian and Pacific MORB mantle domains from Christie et al. (2004). Yellow ellipse in (b) indicates a region of concentrated OCCs at the AAD. Black ellipses in (a) and (b) indicate the approximate locations of the 30 and 50 mGal contours of the RMBA. The red semi-circle at the boundary between Segments B5 and A1 in (b) indicates that the magma from the enriched Pacific mantle domain tends to migrate westward.

DownLoad: Full-Size Img PowerPoint

Based on the observed chaotic topography at the AAD, it was suggested that the chaotic topography of the AAD might have developed for at least 30 Ma (Christie et al., 1998). Furthermore, it was inferred that the observed chaotic topography, and thus the associated zone of concentrated OCCs, could have migrated westward at a rate of 19–27 mm/a (Mark et al., 1990; Christie et al., 1998). Meanwhile, the RMBA envelopes indicate an apparent westward rate of ~40 mm/a (Figs 7a,b). Therefore, the westward migration rate of the zone of concentrated OCCs could be in the range of ~19–40 mm/a. Clearly more in-depth studies are needed to investigate the causes and consequences of the possible migration of mantle domains and tectonic zones.

5. Conclusions

In this study, we analyzed along-axis variations in seafloor morphology, geophysical and geochemical anomalies, and the development of OCCs on the SEIR from 108°E to 134°E, yielding the following conclusions:

(1) The AAD is deeper than the adjacent ridge segments of the SEIR and is associated with more positive RMBA, indicating thinner crust and/or colder mantle. The higher Na_8.0 and lower Fe_8.0 values of the AAD indicate relatively shallow initial melting depth and small amount of mantle melting.

(2) We developed a new method of calculating 2D map-view M factor and illustrated that ridge segments within the AAD have smaller average M factors than the adjacent segments. The relatively small average M values of the AAD indicates limited magma supply and strong tectonism through faulting.

(3) Most OCCs occur in the AAD along the SEIR, especially at Segments B3 and B4. The OCC features are associated with relatively large values of local positive RMBA, indicating that the OCCs are developed at sites of limited magma supply. It is anticipated that the zone of concentrated OCCs might migrate westward along the ridge axis.