Lowstand wedges in carbonate platform slopes (Quaternary, Maldives, Indian Ocean)

Seismic, hydroacoustic and sedimentological data were used to analyse the response of atoll‐slope sedimentation in the Maldives to the late Quaternary sea‐level change. The slope deposits, as imaged in multichannel seismic profiles, are arranged into stacked aggrading to backstepping basinward thinning wedges. In a piston core recovered at the lower slope of one of the atolls, the sediment texture ranges from packstone to rudstone. Major components are blackened bioclasts, the large benthic foraminifers Operculina and Amphistegina, together with Halimeda debris and red algae. Radiocarbon dating at a core depth of 66 cm indicates that the wedge sedimentation stopped or was largely reduced after 16 ka BP. Therefore, the atoll‐slope deposits largely consist of sediment formed in situ and deposited during the last glacial lowstand in sea‐level. This is in apparent contradiction to the concept of highstand shedding of tropical carbonate platforms, which requires slope sedimentation during sea‐level highstands, when the platform is flooded. Rather than intrinsic factors, such as sediment bypass along the steep slope, the extrinsic process of current winnowing of the slope appears to be a major controlling factor in the production of this feature. This process may be relevant for other case studies of carbonate platforms, as currents may be accelerated around such edifices, leading to slope winnowing and sediment deposition in more current‐protected zones. The study results also have consequences for the interpretation of outcrop and seismic subsurface data of carbonate platform slope series, because such slope sediment wedges are not necessarily formed during sea‐level highstands, but can consist of lowstand wedges only.


INTRODUCTION
When tropical carbonate platforms are flooded and the platform interior is occupied by a neritic carbonate factory, sediment is exported into the adjacent basins, especially towards the leeward flanks of the platform where this sediment is redistributed through wind-driven currents and waves (Eberli & Ginsburg, 1987). This process has been defined as highstand shedding (Schlager et al., 1994), which is the main mechanism to form the slope sediments that accumulate into wedgeshaped bodies. Because the amount of sediment deposited during sea-level highstands is larger than the amount formed during sea-level lowstands (Grammer & Ginsburg, 1992), carbonate platform slope wedges mainly consist of highstand deposits. Intrinsic factors, such as an oversteepening of the slope or differences in the grain size of the exported particles, were evoked as controlling factors to allow slope sediment bypass or even erosion (Schlager & Camber, 1986;Kenter, 1990;Rendle-B€ uhring & Reijmer, 2005). This concept of a highstand origin for slope deposits is well established and applied when interpreting subsurface and outcrop data from carbonate platform slopes. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
There is, however, growing evidence that gravitationally controlled off-bank transport mechanisms are only one factor controlling carbonate platform slope deposits and that alongslope contour currents are another major driver of slope deposition. This has been shown at the different slopes of the Bahamas carbonate platform, where the highstand sediments accumulate in periplatform drifts, i.e. drift bodies lining the platform flanks (Betzler et al., 2015;Tournadour et al., 2015;Chabaud et al., 2016;Principaud et al., 2016;Wunsch et al., 2016), irrespective of the windward or leeward exposure of the slope. Slope segments with elevated contour current velocities show no sediment cover or reduced sedimentary thickness (Neumann & Ball, 1970;Mulder et al., 2012).
The Maldives archipelago consists of atolls bathed by vigorously flowing and seasonally reversing currents (Betzler et al., 2009(Betzler et al., , 2013aL€ udmann et al., 2013). Therefore, the atoll slopes appear to be a good location to study the interaction of currents and sea-level controlled slope sedimentation and to expand the insights into the interaction of carbonate slope sedimentation and contour currents. It will be shown that the established lowstandhighstand partition is not applicable because virtually no highstand material is deposited along the slope.

GEOLOGICAL SETTING
The Maldives archipelago south-west of India, in the central equatorial Indian Ocean, is an isolated tropical carbonate platform (Fig. 1). The north-south-oriented double row of atolls encloses the Inner Sea of the Maldives (Fig. 1A). Atolls are separated from each other by inter-atoll channels, which deepen towards the Indian Ocean (Purdy & Bertram, 1993). The Inner Sea is a bankinternal basin with water depths of up to 550 m (Fig. 1B). The Maldives carbonate sedimentary succession is almost 3 km thick and has accumulated since the Eocene, away from any terrigenous input (Aubert & Droxler, 1992;Purdy & Bertram, 1993;Belopolsky & Droxler, 2004;Betzler et al., 2009Betzler et al., , 2013b. The archipelago comprises about 1200 smaller atolls. Discontinuous marginal rims formed by smaller atolls (faros) surround lagoons with water depths of up to 50 to 60 m (Betzler et al., 2015). The oceanward margins of the Maldives archipelago are generally steeply inclined, with dips of 20 to 30°down to 2000 m of water depth. On the Inner Sea side, stepped atoll slopes have the same dip angles, but reach down to water depths of a few hundred metres, where the gradient rapidly declines (F€ urstenau et al., 2010). The Inner Sea is characterized by periplatform ooze deposition (Droxler et al., 1990), locally accumulated into sediment drift bodies (Betzler et al., 2009(Betzler et al., , 2013aL€ udmann et al., 2013).
The climate and oceanographic setting of the Maldives is dictated by the seasonally reversing Indian monsoon system (Tomczak & Godfrey, 2003). South-western winds prevail during the northern hemisphere summer (April to November), whereas northeastern winds prevail during winter (December to March). Winds generate ocean currents, which are directed westwards in the winter and eastwards in the summer. Interseasonally, a band of Indian Ocean Equatorial Westerlies establishes strong, eastward-flowing surface currents with velocities of up to 1Á3 m s À1 . Currents reach down to the sea floor (L€ udmann et al., 2013), especially in the inter-atoll passages, where submarine dunes and moats occur (Betzler et al., 2009(Betzler et al., , 2013b.

METHODS
Seismic signals were generated by means of two clustered GI-Guns. The volume of each GI-Gun was 45 cin for the generator with a 105 cin injector volume. The GI-Guns were operated in 'true GI mode' and synchronized by a SureShot trigger system, which displays the source signal of each airgun. The digital streamer used for the survey was a Hydroscience Technologies SeaMUX 144-channel array with an active length of 600 m and an asymmetric group interval. The selected shooting distance during the entire cruise was 25 m.
Sub-bottom data were recorded with the RV METEOR parametric sediment echo sounder (PARASOUND P70; Atlas Elektronik, Bremen, Germany). The system was operated with two frequencies (18 kHz and 22 kHz). The software PS32segy (Hanno Keil, University of Bremen, Germany) was used to cut and convert the data. Data processing was performed with the software package ReflexW (Sandmeier Software, Karlsruhe, Germany), comprising automatic gain control (AGC) and along-profile amplitude normalization.
Sediment samples were acquired with a piston corer during Cruise M74/4 with RV METEOR and a videograb sampler during Cruise SO236 with RV SONNE. Components in the sediments were analysed quantitatively: Samples were wet-sieved and 200 components from the 500 to 1000 lm and the >1000 lm fraction were counted. The following components were differentiated: Amphistegina, Homotrema, Operculina, Heterostegina, Gyrodinoides, miliolids, other benthic foraminifers, planktonic foraminifers, bryozoans (encrusting, robust branching, vagrant), bivalves, gastropods, echinoderms, serpulids, crustaceans, bioclasts and lithoclasts. Especially bioclasts in some samples have a dark grey to black stain. Such components were counted separately. Tables with the counts are deposited at www. pangaea.de.    (Table 1). Samples were ultrasonically cleaned in deionized water and visually inspected for cements, overgrowths and fills. Conventional radiocarbon ages were calibrated using Calib (v.7.0.4, Stuiver & Reimer, 1993) and the calibration curve Marine13 (Reimer et al., 2013) with no local reservoir correction applied. Calibrated ages were rounded to the next decade, and in the text, the median of the probability distribution is used in conjunction with the two-sigma range (95Á4% probability).

Seismic data
During the M74/4 cruise, the rims of the Ari and Male atolls were crossed three times at different positions in passages separating the faros lining the atoll's borders (Fig. 1B); overview lines are presented in Fig. 1C and D. Both atolls are around 50 m deep, and the passages between the faros have a slightly shallower sill before the stepped slopes dip at 10 to 50°into the Inner Sea. The Inner Sea is between 200 and 430 m deep, with the deepest area in its central part. Whereas seismic imaging of the lagoonal stratigraphy is reduced because of the seafloor multiple, data from the slopes and the Inner Sea give a good insight into the stratigraphic succession (Figs 1C, D, and 2). The Inner Sea succession consists of a drowned carbonate bank lined and overlain by drift deposits (Fig. 1C), similar to the succession of the Kardiva Channel located further north in the Maldives archipelago (Betzler et al., 2009(Betzler et al., , 2013aL€ udmann et al., 2013).
Line 1, located at the northeastern margin of Ari Atoll is oriented NE-SW and crosses the position of Core M74/ 4-1135 and of videograb sample SO236-51 (Figs 1B, C and 2A). Down to a water depth of 100 m, the line images an irregular sea-floor relief on top of a succession with discontinuous to chaotic reflections. From ca 97 to 128 m, there is a submarine cliff. Basinwards, the slope progressively flattens out to water depths of 180 m. Here, reflections are laterally more or less continuous with moderate amplitudes. Strong, slightly inclined reflections occur in the upper part of the succession, which appear truncated downslope. Around 8 km NE of the platform edge, the deposits are arranged into drift bodies (Fig. 1C).
Line 2 (Figs 1B, D and 2B) is located at the eastern margin of Ari Atoll. Similar to the northeastern margin of the atoll (Line 1), there is a sea-floor step at ca 97 m. In front of this step, there is a sediment wedge with strong basinward reflections, which are also truncated downslope. This wedge overlies a horizon with a strong acoustic impedance, which corresponds to the limit between the Inner Sea drift units 8 and 9 (L€ udmann et al., 2013), which was formed at ca 2Á3 Ma . Drifts sediments form the succession further away from the atoll margin ( Fig. 1D).
A sediment wedge is also imaged in Line 3 (Figs 1B and 2C), which crosses the south-western margin of North Mal e Atoll. Towards the Inner Sea, the smooth pattern of reflection changes into a discontinuous to wavy Table 1. Results of radiocarbon dating. Calibration was performed using Calib (v7.0.4, Stuiver & Reimer, 1993) and the calibration curve Mar-ine13 (Reimer et al., 2013)  stratification which corresponds to an area where the sea floor is covered by submarine dunes moved by bottom currents (Betzler et al., 2013a;L€ udmann et al., 2013).

Sedimentology
Piston Core M74/4-1135 was retrieved at 4°16 0 59Á592″N, 72°50 0 4Á992″E, i.e. around 2 km NE of the margin of Ari Atoll at a water depth of 172 m (Figs 2A and 3A, B). It recovered 12Á22 m of unlithified carbonates with a rudstone to grainstone texture (Fig. 4) deposited in the sediment wedge located in front of the 30 m high submarine cliff which forms the seaward limit of the terrace at a water depth of ca 97 m. Figure 3 depicts the linkage between sedimentological and high-resolution parasound data. The parasound profile images a surface with a high impedance contrast at a depth of 7 mbsf (metre below sea floor). In the core, this surface correlates with a colour change from dark greenish grey to grey. The darker core colours are due to the high abundance of dark components giving the sediment a 'salt-and-pepper' texture ( Fig. 5A). The dark colour of the grains is a consequence of an elevated organic and clay content, as resolved in smear slides of acid residues.
The lower 30 cm of the succession consists of a rhodolith-rich rudstone with a fining-upward trend of the components. Between 11Á92 mbsf and 8Á80 mbsf, the core is a grey rudstone with some dispersed rhodoliths up to 5 cm in size. Above 8Á8 mbsf, a rudstone with some rhodoliths is mottled with a greenish grey to grey colour. A platy coral was recovered at 6Á7 mbsf. Large components disappear upcore, with the last large rhodoliths registered at 5Á5 mbsf. The top of the succession is a light grey packstone to grainstone with Halimeda flakes, planktic foraminifers, pteropods, benthic foraminifers, serpulids and echinoid debris (Fig. 5B).
Out of the differentiated and counted components in the fraction >1 mm, some show distinct trends with depth. The variation in abundance of these components is presented in Fig. 4. From the bottom of the core, Amphistegina and Operculina decrease in abundance upcore from around 15% and 25% of the components, respectively, to almost disappearing in the upper part of the sequence. Red algae and the green alga Halimeda show an inverse trend increasing upcore to reach 10% to 15% and 40% of the particles, respectively. Bioclasts decrease in abundance upcore, as is the case for the stained dark bioclasts, which make up 40% to 45% of the samples below 9Á6 mbsf. In contrast to the trend described for the other components which extend over the entire core, the abundance of the dark bioclasts rapidly decreases between 9Á6 mbsf and 7Á4 mbsf to values of 0% to 5% above, except for the interval between 6Á8 mbsf and 6Á2 mbsf, where 10% to 20% were counted.
Red algae from selected rhodolith samples (Table 2) were analysed in thin section for identification at the most precise taxonomic level possible. Among coralline red algae, the dominating genus is Lithothamnion, followed by Lithophyllum and minor Lithoporella. The aragonitic red alga Peyssonnelia occurs in several samples as well.
Five samples were used for age dating the succession (Fig. 4, Table 1) suggesting that the interval between 0Á66 mbsf and 9Á16 mbsf was deposited between approximately 16 and 22 ka BP. Deposits below 9Á16 mbsf were not dated, as fossils without encrustation were not found. Three intervals with distinct sedimentation rates can be differentiated: below 7 mbsf, the rate is around 0Á6 mm year À1 , between 7 and 0Á66 mbsf the values increase to 3Á9 mm year À1 before dropping sharply to 0Á04 mm year À1 .
Videograb sample SO236-51 is located at the toe of a 31 m high terrace which is positioned at a water depth of ca 97 m (Fig. 3A). The 132 m deep location is 500 m upslope of Site M74/4-1135. The sea floor at the sample locality and along the flank of Ari Atoll is an irregular rocky surface (Fig. 6A and B). The rock is a rudstone (Fig. 6C) with large benthic foraminifers (Amphistegina, Heterostegina, Alveolinella), encrusting and articulated red algae, encrusting foraminifers and bryozoa, Halimeda flakes, serpulids and planktonic foraminifers. The irregular rock surface, colonized by gorgonians, sponges and bryozoa, is characterized by holes and depressions with irregular shapes (Fig. 6B). For age dating of these deposits, several Amphistegina specimens were isolated which yield an age of around 13Á32 AE 0Á11 ka cal BP.

Depositional geometry
The three seismic lines depicted in Fig. 2 cover the transition from the reef buildups of the atoll margins to the stratified succession of the fore-reef area and the basin. This transition is characterized by a change from discontinuous and chaotic reflections to more continuous reflections from the sediment wedges. The limit between both acoustic facies is placed at the toe of the 31 m high wall of the 97 m terrace, which has been attributed to reefs which drowned during Meltwater Pulse 1A (14Á3 to 14Á65 kyrs BP; Deschamps et al., 2012Deschamps et al., ) by F€ urstenau et al. (2010. The successions of the different slopes analysed are arranged into an aggrading to slightly backstepping pattern (Fig. 2).

Lowstand sediment wedges
The sediment wedges were formed during the last glacial sea-level lowstand, as indicated by the dating of the core M74/4-1135, which shows that the succession was deposited between ca 22 and 16 kyrs BP (Table 1, Fig. 7). Applying the reconstruction for the sea-level position during and after the last glacial maximum (Deschamps et al., 2012;Lambeck et al., 2014), the site would have been located at water depths of ca 48 m at 22 kyrs BP, ca 55 m at 18 kyrs BP, ca 52 m at 16Á9 kyrs BP and ca 57 m at 16 kyrs BP (Figs 7 and 8).
In general, Core M74/4 displays an upcore trend from a floatstone to rudstone to a packstone to grainstone texture. The red algal associations and the large benthic foraminifers provide two lines of evidence indicating that the deposits recovered in Core M74/4-1135 were formed in situ within this palaeobathymetric range, or are parautochthonous at most. The samples marked by grey shading in the red algal overview of Table 2 include components common in the 'intermediate water assemblage' of Webster et al. (2009), namely Lithophyllum gr. acrocamptum and Lithothamnion prolifer. In the Pacific Ocean, in the absence of shallow-water species, such as Porolithon onkodes, these coralline algae are typical of 20 to 60 m water depths; a similar depth range, however, can be expected for the Indian Ocean, where less data are available. This covers the water depth of ca 50 m for the site location applying the Deschamps et al. (2012) sealevel curve. In any case, this depth range obviously can change in extremely clear waters, which however, in the discussed case study is irrelevant, as the maximum water depth at the time of sedimentation is constrained by the sea-level position (Fig. 8).
The large benthic foraminifer Operculina ammonoides, which is frequent to abundant in the lower and middle part of the succession (Figs 4 and 7), lives in water depths of up to 70 m with an optimum around 40 m (Hohenegger, 2000). Amphistegina lessoni and A. radiata are also frequent to abundant and thrive in water depths of up to 70 m, with highest abundance around 40 m (Hohenegger, 2000). As is the case for the red algae, the palaeobathymetry indicated by the large benthic foraminifers in the succession falls within the water depth for the site (Deschamps et al., 2012). Both large benthic foraminifer decrease in abundance as the water depth at Site M74/4-1135 deepened (Fig. 7). This trend is paralleled by increased amounts of Halimeda and red algae. In the Indo-Pacific, Halimeda and red algae may be more abundant at water depths exceeding those occupied by large benthic foraminifers (Wilson & Vecsei, 2005). Therefore, the upcore increase in algal abundance may well be a consequence of the deepening of Site M74/4-1135 during the post-glacial sea-level rise.
The dark-stained grains (Fig. 4) record a certain sediment influx from the exposed carbonate platform. In carbonates, such grains blackened by organics and clays are known from intertidal and subtidal environments (Fl€ ugel, 2004). It is proposed that in Core M74/4-1135, these grains were shed from the emerging parts of the platform, which were just some hundreds of metres away from the site location. This export was reduced with the transgression due to the beginning of sea-level rise around 20Á5 ka BP (Lambeck et al., 2014) to recover a little bit around 18Á5 ka BP with the temporary stop of sea-level rise at that time (Fig. 7). The base of this transgression correlates with the high-amplitude reflection in the parasound profile which crosses the core location (Fig. 3).
In summary, data indicate that the sediment wedge at position of Site M74/4-1135 is a lowstand wedge and that no or only very minor sediment is deposited in this area during the latest Pleistocene and Holocene. The parallelism in shape and seismic stratigraphy of this wedge compared with the other wedges as imaged in the seismic lines (Fig. 2) indicate that this also applies to other atoll flanks.

Intrinsic and extrinsic control on slope deposition
The seminal publication by Schlager et al. (1994) established the concept of the highstand shedding of Quaternary tropical flat-topped carbonate platforms, where the  platforms export sediment into the adjacent basins when they are flooded and a shallow-water carbonate factory produces particles. Limitations were seen for carbonate ramps and carbonate platforms undergoing sea-level lowstands for extended periods of time. Applying the highstand shedding scenario, carbonate platform slope angles are seen as a function of grain size (Kenter, 1990), with coarser-grained sediments developing steeper slope angles than finer-grained ones. Both aspects are now widely accepted paradigms of carbonate sedimentology and stratigraphy applied when interpreting geological and subsurface geophysical data.
In the carbonate sedimentary succession of the Maldives, a differentiation of highstand and lowstand deposition has been demonstrated for the Pleistocene and Holocene drifts of the Inner Sea by Paul et al. (2012) and Betzler et al. (2013b). This is based on variations in the amount of fine-grained, platform-derived aragonite mud which at different basinal localities of the Maldives is most abundant in the highstand deposits. Therefore, it remains to be resolved why the slope and toe of slope deposits do not bear highstand sediments.
Sediment export by highstand shedding does not necessarily imply deposition at the slope of the carbonate platform, because other processes may outweigh gravitational transport processes. In carbonates, slope steepening behind the angle of repose has been suggested as an intrinsic mechanism allowing slope bypass of sediment exported from the platform into the basin (Schlager & Camber, 1986). The Bahamas escarpment has been reported as the type example to illustrate the case of slope sediment bypass. This, however, has to be revised because a number of recent studies show that along-slope contour currents winnow the slope and toe of slope (Mulder et al., 2012;Betzler et al., 2015;Jo et al., 2015;L€ udmann et al., 2016;Principaud et al., 2016;Wunsch et al., 2016). Consequently, it is reworked deposits that accumulate in periplatform drifts (Betzler et al., 2015) and in detached drifts in the basin (Bergman et al., 2010;L€ udmann et al., 2016).
Large drifts in the Inner Sea together with submarine dunes show that currents also dictate sedimentation patterns in the Maldives (Fig. 1C and D) (Betzler et al., 2009(Betzler et al., , 2013a(Betzler et al., ,b, 2016L€ udmann et al., 2013). Current speeds in the channels between the atolls are typically in the range of 0Á5 to 0Á8 m s À1 and accelerate in the interfaro channels and along the atoll flanks to 1Á5 to 2Á6 m s À1 (Preu & Engelbrecht, 1991;Owen et al., 2011). These currents are the result of the interaction of ocean, tidal and wind-induced currents. As noted by Darwin (1842, p. 108), based on the reports of Capt. Moresby, 'The currents of the sea flow across these atolls (. . .) yet the currents sweep with greater force round their flanks'.
The effect of these currents is well-documented in the video survey performed along the flank of Ari Atoll around the location of Sample SO236-51, where the sea floor is rocky, with no sediment cover (Fig. 6A). A question which remains unanswered based on available data is if the current regime flowing around the atolls was different during the past sea-level lowstand thus allowing lowstand deposition.
The margin aggradation or backstepping (Fig. 2) in the Maldives can be traced back to the lower Pliocene (Betzler et al., 2013a). Such a pattern elsewhere is interpreted to reflect carbonate platform growth under increasing accommodation, but the long-term platform trend discards sea-level changes as a trigger as it also applies to the rate of subsidence which has been calculated to be around 0Á03 to 0Á04 mm year À1 over the long term (Belopolsky & Droxler, 2004) and to 0Á15 mm year À1 for the short term (Gischler et al., 2008). The onset of the growth pattern correlates rather with the early Pliocene widening of the inter-atoll passages as mapped by L€ udmann et al. (2013). Inner Sea drift geometries indicate that the opening and widening of the passages allowed the current system that controls sedimentation nowadays to be established.
Summarizing, it is proposed that alongslope currents redistribute the sediment formed during the highstand shedding away from the atoll slopes, where the highest current speeds are to be expected, towards the more protected and tranquil areas of the Inner Sea. These are the zones of the Inner Sea, where large drift bodies accumulate ( Fig. 1C and D). These drift sediments show a clear differentiation into aragonite-mud rich highstand deposits and aragonite poor lowstand deposits (Paul et al., 2012;Betzler et al., 2013b). Self-erosion of the slope (Schlager & Camber, 1986) is excluded because the seismics and parasound lines do not show the corresponding signatures (Fig. 2).

CONCLUSIONS
The slope successions of atolls in the Maldives archipelago consist of a series of basinward thinning wedges which were deposited during sea-level lowstands, whereas highstand deposits are condensed. Rather than intrinsic factors such as slope steepness, the vigorous currents around and in the atolls control this pattern. Highstand sediment is not accumulated along the slope but reworked and transported to be accumulated into drift deposits forming in current-protected areas. This observation confirms that contour currents are of equal importance for determining carbonate sedimentation patterns on carbonate platform slopes as is, for example, the windward-leeward orientation of the slope. This effect of contour currents is expected to be more