On the basis of the analysis of seismic and well data (Figure 2), and the data acquired during the 2004–2005 cruises, the general evolution of the mixed system in the GoP is outlined with an emphasis on which processes can be considered predominant and which processes are minor in the system development during different Cenozoic time intervals. Four major phases can be distinguished in the geological evolution of the Cenozoic mixed GoP system based on the balance between tectonics, eustasy, carbonate production, and clastic sediment supply.
4.2.1. Tectonic Control of the System Evolution (Phase 1)
 The first phase in the mixed system evolution was mostly driven by large-scale tectonics, represented by periods of active rifting, subsidence, and uplift. This phase corresponds to the last stage of the Coral Sea spreading in the late Cretaceous-Paleocene and subsequent uplift resulted in intensive erosion (so-called base tertiary unconformity, BTU) [Hamilton, 1979; Symonds et al., 1991; Pigram and Symonds, 1993; Hall, 2002; Quarles van Ufford and Cloos, 2005; Morgan, 2005]. The gravity map of the GoP (Figure 3) helps to characterize the overall tectonic grain presently buried under the modern shelf and seaward of the modern shelf edge. The early physiography, molded by tectonics, displays two sets of preferential northeast and northwest trending orientations, which have influenced the GoP evolution throughout the entire Cenozoic. The first structures consist of a series of three northeast trending relatively continuous ridges (Pasca, Portlock ridges, and Eastern Fields Ridge (defined as such in this paper)) separated by intervening paleotroughs and modern troughs (Pasca, Flinders, Ashmore, and Pandora). The second set of structural features includes northwest oriented deep troughs (ancient Aure and Moresby) which served and currently act as major conduits for siliciclastic sediment transfer through Moresby Canyon down to the Coral Sea basin, the ultimate sediment sink for the sediments not stored on the shelf and slope of the GoP. The gravity map and two schematic geological cross sections (Figure 4) illustrate that the morphology of paleoridges and troughs buried beneath the GoP shelf under several kilometers thick late Pliocene-Pleistocene siliciclastics is almost identical to the morphology in the deep-water part of the GoP directly adjacent to the shelf (Ashmore and Pandora Troughs adjacent to Portlock and Eastern Fields ridges). This tectonic morphology later influenced the late Oligocene-early Miocene establishment and distribution of the Miocene carbonate platforms as well as the invasion of the late Miocene-Holocene clastics.
Figure 4. Schematic cross sections showing generalized structures and sedimentary sequences in the GoP (modified from Davies et al. ). Note long- and short-lived carbonate systems and mixed carbonate-siliciclastic and pure siliciclastic parts of the GoP shelf. Shelf edge position is controlled by the distribution of long-lived carbonate platforms and structural features. Siliciclastic shelf edge did not prograde, being “anchored” along Borabi Reef Trend (BRT) or Pasca Ridge, until the next adjacent troughs (Pasca and Flinders Troughs, respectively) were infilled (see graph). High gradient slope of the curve shows low rate of shelf edge progradation, while low gradient curve shows high rate progradation of the shelf edge per million years. In this model, the modern shelf edge is “anchored” along Portlock Ridge including Portlock Reef (PR) and will not prograde until Pandora Trough is infilled.
Download figure to PowerPoint
4.2.2. Carbonate Deposition and Eustatic Control (Phase 2)
 In general, during the Cenozoic, two types of carbonate systems can be identified in the GoP, long-lived and short-lived systems (Figure 4). The long-lived systems include several large isolated long-lived carbonate platforms which have evolved since the late Oligocene-early Miocene, and observed today as typical large atolls (e.g., Ashmore, Boot, Portlock, and EFR). Over longer times, these or similar long-lived carbonate edifices, buried under the GoP shelf, appear to act as compartments, controlling the GoP margin evolution, guiding and restraining siliciclastic sediment fluxes from the continent. This has been clearly the case in the recent past for Portlock Reef and the Miocene drowned reefs on the northeastern extension of EFR in Pandora Trough (Figure 5b), where drowned platforms directly interact with the slumping masses along the slope of the central Pandora Trough. Some of the isolated carbonate platforms were drowned starting in the early Miocene and were subsequently buried by siliciclastics (e.g., Pasca and Pandora reefs) [Sarg et al., 1996; Morgan, 2005], while others (the Ashmore-Boot-Portlock Reef complex and EFR) remained active and not buried from the late Oligocene until present. On the other hand, short-lived carbonate systems represent buildups of much smaller size that lived over short periods of time (Figure 4). These include the northern extension of the GBR, which covers the western shelf and, like its more southern counterpart (e.g., Ribbon Reef 5 on the GBR), is probably not older than mid-Brunhes in age (<0.5 Ma) [International Consortium for Great Barrier Reef Drilling, 2001; Webster and Davies, 2003]. Moreover, during the 2004 and 2005 cruises, a series of early transgressive barrier reefs was discovered (Figures 5c and 5d), established on top of coastal Last Glacial Maximum siliciclastic deposits. Because of short life of these reefs, they are considered to be ephemeral sources and sinks of neritic carbonates.
Figure 5. High-resolution multibeam bathymetry maps of different portions of the GoP, showing modern analogs of vertically stacked sedimentary features interpreted in the Miocene-Pleistocene infilling of Pasca and Flinders paleotroughs (see Figure 9). (a) Submarine fan and channel-levee system in deep-water Moresby Trough. (b) Ponded turbidites and slope deposits in Pandora Trough. Drowned isolated platforms serve as barriers for siliciclastics. (c) Prograding lowstand shelf edge delta deposits in Ashmore Trough. (d) Three-dimensional perspective of the transgressive drowned barrier reef on the modern shelf break in northern Ashmore Trough. (e) Location map. Depositional features in Figures 5a–5c demonstrate a lateral trend which is comparable to the vertical stacking of the Pliocene-Pleistocene depositional environments, observed in the prograding several kilometers thick siliciclastic sediment pile infilling the paleotroughs and burying the drowned carbonate platforms.
Download figure to PowerPoint
 Carbonate deposition in the mixed GoP system was initiated during the Eocene. Interpretation of the seismic and well data suggests that during the late Oligocene-early Miocene, the initial distribution of major carbonate sources and sinks in the GoP was controlled by a system of preexisting northeast oriented structural ridges. Such carbonate platforms as Uramu, Pasca, Pandora, Ashmore-Boot-Portlock, and EFR reefs were established during this time on the uplifted blocks (Figures 3 and 4). The northeast trending Borabi Reef rimmed a large carbonate shelf in the western part of the GoP by the middle of the early Miocene [Pigram et al., 1989, 1990; Carman, 1993]. We assume that a late Oligocene-early Miocene overall sea level transgression [Vail et al., 1977; Haq et al., 1987, 1988; Billups and Schrag, 2002] triggered the large-scale establishment of major carbonate platforms in the GoP. Although the New Guinea mountain chains possibly started emerging at that time, the area of neritic carbonate production was not influenced by siliciclastics, because their accumulation was mostly restricted in Aure Trough (Figure 3), the foreland basin located in the early Miocene further northeast from the carbonate province.
 Our study reveals that the evolutionary history of the neritic carbonate system in the GoP was very similar to the evolution of the pure carbonate systems along the northeastern Australian margin and particularly on the Queensland and Marion plateaus [McKenzie et al., 1991; Droxler et al., 1993; Feary et al., 1993; Betzler et al., 1993, 2000; Brachert et al., 1993; Isern et al., 2002; John and Mutti, 2005]. We observe that during the overall late Oligocene and early Miocene sea level transgression, most of the platforms in the GoP experienced a general back stepping of environments, resulting in a systematic decrease of the surface area of neritic carbonate production. Some of the platforms were partially or completely drowned at that time. For example, EFR platform started as a much larger system extending toward the southwest and northeast relative to the modern EFR (Figures 1 and 5b) and later partially drowned in the early Miocene. On the basis of newly acquired high-resolution bathymetry data, several narrow, drowned, isolated, high-relief carbonate platforms were discovered to the northeast of EFR in the central Pandora Trough. Analyses of dredge samples from one of them (Figures 5b and 6) demonstrated that the original EFR platform was drowned as early as 20 Ma [Droxler et al., 2004].
Figure 6. Thin sections showing early Miocene larger foraminifera (Lepidocyclina (Nephrolepidina) and Spirockypena margaritatus) (upper (uTe) zone, ∼20 Ma) observed in dredge samples collected from the slope of a drowned carbonate platform in Pandora Trough (dredge sample is located in Figure 5b).
Download figure to PowerPoint
 An overall back stepping pattern of the carbonate system during the early Miocene is also shown on Borabi platform (Figure 7). During the end of the early Miocene and the earliest middle Miocene, Borabi platform, as many carbonate platforms in the GoP, vertically aggraded and therefore was able to keep up with the rise of sea level. In the middle Miocene, the carbonate deposition on the northeastern Borabi margin shifted downward which most likely signals a systematic sea level lowering [Billups and Schrag, 2002]. Subsequent deposition resulted into well developed progradational patterns. The platform was then reflooded during a major transgression at the very beginning of the late Miocene.
Figure 7. (a) Seismic profile showing different geometries of the Miocene carbonate system in the GoP (back stepping, aggradation, progradation, reflooding). TWT, two-way traveltime. (b) Basic geometries of tropical platforms as a response to rates of change in accommodation space and platform carbonate growth rates [Schlager, 2005].
Download figure to PowerPoint
 The late Oligocene-middle Miocene stacking patterns observed in the seismic profiles crossing the Borabi Reef trend (Figures 7a and 8) correspond to the identical and contemporaneous pattern of back stepping, aggradation, downward shift, progradation, and reflooding observed in other pure carbonate systems such as the Bahamas (Figure 8a) [Eberli and Ginsburg, 1987; Eberli et al., 1997], the Maldives (Figure 8b) [Belopolsky and Droxler, 2003, 2004a, 2004b], and also in pure siliciclastic system on the New Jersey continental margin [Miller et al., 1996]. This so-called Neogene global stratigraphic signature [Bartek et al., 1991] is represented in the schematic diagrams shown in Figure 8d by (1) aggradation and back stepping and partial drowning in the late Oligocene-early Miocene, (2) vertical growth or aggradation in the latest early Miocene and earliest middle Miocene, (3) a downward shift of deposition in the middle Miocene, (4) systematic lateral growth or progradation in the middle Miocene, and (5) reflooding and aggradation from the late Miocene until the early Pliocene. Because the Maldives and the Bahamas platforms, as well as the New Jersey margin are considered to be tectonically stable during the Oligocene-Neogene, the common sedimentary geometries observed in these different systems must be produced by eustatic sea level fluctuations.
Figure 8. (a) Overall sequence stratigraphic signature observed in the interpretation of the Bahamas western line (modified from Eberli et al. ). (b) Neogene stratigraphic signature along the West Maldives Inner Sea carbonate margin (modified from Belopolsky and Droxler [2003, 2004a, 2004b]). (c) Neogene stratigraphic signature on the Borabi platform margin and adjacent slope in the GoP. (d) Stratigraphic signature of the Neogene (modified from Bartek et al. ). Numbers in circles: 1, late Oligocene-early Miocene aggradation, back stepping and partial drowning; 2, latest early Miocene-earliest middle Miocene vertical growth or aggradation; 3, middle Miocene downward shift of deposition; 4, latest middle Miocene systematic lateral growth or progradation; and 5, late Miocene-early Pliocene reflooding and aggradation.
Download figure to PowerPoint
 The Neogene signature is also observed in the GoP, and therefore eustatic sea level fluctuations apparently influenced the major carbonate sources and sinks in the GoP during maximum development of the carbonate system in the late Oligocene-Miocene. Our interpretation of the seismic data shows that in general the study area had relatively stable tectonics with the exception of some local zones of late Oligocene and earliest Miocene extensional faulting and more intensive differential subsidence (e.g., Pasca Trough). Accommodation space was produced by a combination of a low subsidence rate and eustatic sea level fluctuations. The stacking pattern of sedimentary sequences (Figures 7 and 8c) observed in the GoP carbonate system (back stepping, vertical aggradation, downward shift, progradation, reflooding) is identical to the contemporaneous pattern identified in the Bahamas and Maldives and can only be explained by eustasy influencing the neritic carbonate production. Siliciclastics did not influence the system, since at that time they were isolated to the east in the deepest part of the foreland basin referred to as Aure Trough. The late Oligocene-Miocene evolution of the GoP mixed system therefore was preferentially controlled by intensive carbonate production and eustasy that resulted in thick carbonate successions with sequence geometries identical to the well-established Neogene global stratigraphic signature observed worldwide.
4.2.3. Partial Demise of Carbonate Platforms (Phase 3)
 The demise of carbonate systems in either pure carbonate or mixed carbonate-siliciclastic environments can be caused by one or a combination of several factors such as significant eustatic sea level changes, increased tectonic activity, siliciclastic burial, and/or climatic/environmental changes. Tectonic activity in the PNG region increased during the late Miocene-early Pliocene with intensive fold and thrust belt development which resulted in significant crust loading and related subsidence [Home et al., 1990; Quarles van Ufford and Cloos, 2005]. This increased subsidence was probably a major reason explaining the partial demise of a large part of the carbonate system in the northern part of the GoP. Although the early partial demise (drowning) of the neritic carbonate system at some locations of the GoP was probably already initiated in the early Miocene, the late Miocene-early Pliocene interval was the time of maximum drowning of the carbonate platforms, when such large systems as Borabi and Uramu platforms drowned.
 When the ages of the drowning of Pandora Reef and the first major influx of siliciclastic sediments into Flinders paleotrough are compared (Figure 9) [Sarg et al., 1996], it is clear that a time gap of 10–15 Ma occurred between the carbonate demise and major siliciclastic arrival. This suggests that siliciclastics did not cause the cessation of the carbonate production, but the platforms possibly drowned due to eustatic sea level rise enhanced by increased tectonic subsidence and changes in ocean environment conditions. During this phase, two specific times, at the beginning of the late Miocene (Tortonian) and beginning of the early Pliocene, were characterized by high rates of eustatic sea level rise. These episodes, most likely enhanced by contemporaneous basin subsidence, related to the loading effect of the PNG mountain belt forming at that time, correspond to the intervals of ultimate demise of several carbonate platforms.
Figure 9. Interpreted seismic profile showing depositional features infilling Flinders paleotrough and Pandora slope and burying drowned early Miocene Pandora Reef. The features include submarine fans, ponded turbidite, slope, prograding shelf edge deltas, and aggrading shelf deposits. These vertically stacked depositional environments are observed as modern analogues laterally juxtaposed in the Moresby Trough, central Pandora Trough, Pandora slope, and northern Ashmore shelf edge (Figures 5a–5c). BTU is base tertiary unconformity. Dates are obtained from Sarg et al. . The seismic profile is courtesy of Fugro Multi Client Services, Inc.
Download figure to PowerPoint
4.2.4. Siliciclastic Influx in the Mixed System (Phase 4)
 Since the late Pliocene, siliciclastics have dominated the deposition in the GoP mixed system. This interval is considered to be an overfilled phase in the evolution of the foreland basin. As a main result, the proximal foredeep (Aure Trough) was infilled by clastic sediments and many carbonate platforms in the distal part of the foreland basin (e.g., Uramu, Pasca, Pandora reefs) already drowned for 5–15 Ma [Pigram et al., 1989, 1990] became buried by the prograding siliciclastics. During this phase, the huge influx of siliciclastic sediments, originating from the denudation of the New Guinea uplifted mountains, was linked to the intensified tectonical uplift during the last 3 Ma and associated monsoonal wet tropical climate. The wet climate generated high rainfall resulting in high rates of weathering and erosion as well as high levels of runoff. Since the Pliocene, the siliciclastic shelf edge has prograded about 80 km to the southeast (Figures 3 and 4). Seismic interpretation shows that the rate of progradation was lower when the shelf edge was located on top of the preexisting northeast oriented ridges and carbonate platforms where it was temporarily anchored until the adjacent trough filled. The rate accelerated when the shelf edge was prograding over previously infilled trough (Figure 4).
 In addition, the long-term (∼2 Ma) late Pliocene to mid-Brunhes sea level regression and more than 120 m cyclic sea level fluctuations characteristic of the late Pleistocene (Figure 10) influenced the volume and spatial distribution of siliciclastics, accumulating in the GoP, possibly following the reciprocal model of mixed carbonate-siliciclastic sedimentation [Wilson, 1967; Dolan, 1989; Handford and Loucks, 1993; Schlager et al., 1994; Jorry et al., 2008]. According to this model, during sea level regressions and lowstands, the neritic carbonate production in the GoP was minimized or completely ceased, and the slope and basins became starved of neritic carbonate sediments. River channels incised the exposed continental shelf, and siliciclastic sediments bypassed it, initiating prograding lowstand shelf edge deltas and thick basinal deposits. Unconsolidated siliciclastic sediments accumulated during previous sea level highstands on the inner shelf (e.g., modern prograding clinoforms) were reworked and transported to the slopes and basin floor during the early parts of regressions. During regressions and lowstands, siliciclastic sediment fluxes to the basin floor significantly increased whereas the neritic carbonate fluxes dwindled [Jorry et al., 2008]. During sea level transgressions, the accommodation space on the reflooded shelf increased, and siliciclastics accumulated on the continental inner shelf, along the coast, and in the fluvial plains. As a result, siliciclastic sedimentation on the slope and in the basin floor adjacent to the shelf edge was dramatically reduced. Carbonate bank tops exposed during lowstands reentered the photic zone and neritic production was reinitiated, resulting in greatly increased carbonate exports to the slope and basin floor. In the GoP, the discovery of relict barrier reefs existing along the modern shelf edge [Droxler et al., 2004; Dickens et al., 2006], demonstrates that early transgressions can be optimum intervals for neritic carbonate accumulation on low-latitude siliciclastic shelves (Figures 5c, 5d, and 11). During highstands of sea level, the carbonate factory production on isolated carbonate platform tops and back barrier and barrier reefs on the mixed shelves is still very high, maximizing the carbonate deposition on the surrounding slopes and basin floor.
Figure 10. Graph constructed from 57 stacked, globally distributed benthic δ18O records. This curve is the best proxy for ice volume changes and therefore eustatic sea level fluctuations during the last 5 Ma. The record demonstrates an overall long-term increase of global ice volume or sea level regression from 2.7 to 0.5 Ma and high-amplitude ∼120 m sea level (ice volume) cyclic changes at a frequency of 100 ka during the last 0.5 Ma (modified from Lisiecki and Raymo ).
Download figure to PowerPoint
 The several kilometers thick siliciclastic sediments deposited during different periods of phase 4 also include a series of ephemeral (short-lived), thin, laterally limited neritic carbonate lenses (Figures 4, 5c, 5d, and 11). These ephemeral carbonate deposits are interpreted to have lived over relatively short time intervals, and are much smaller in size when compared with the long-lived carbonate platforms which originated already during the late Oligocene–early Miocene and were able to survive until today (e.g., EFR and Portlock reefs). These ephemeral neritic carbonate accumulations were often first established during early transgression on top of lowstand coastal deposits on the shelf edges, drowned during late transgression, and then were buried by prograding siliciclastics during late highstand, regression and lowstand. Short-lived transgressive carbonate banks buried in thick siliciclastic pile of sediment are not uncommon in low-latitude siliciclastic passive margins. They have been described, for instance, by Belopolsky and Droxler  along the south Texas shelf edge offshore Corpus Christi.
 One of the best examples of ephemeral carbonate accumulations in the GoP is the drowned barrier reef established on top of the lowstand shelf edge delta deposits in the northern Ashmore Trough (Figures 5c, 5d, and 11). This transgressive barrier reef was first established during the early part of the last transgression (∼14.5 ka) on top of lowstand siliciclastic coastal features such as beach coastal ridges. Once established, the reef grew 30 to 80 m high as it kept up with the rising sea level during one of the periods of rapid melting of the Northern Hemisphere ice sheets (Meltwater Pulse 1A) [Droxler et al., 2006]. Finally, this transgressive barrier reef drowned, most likely during Meltwater Pulse 1B (∼11 ka). The transgressive origin of ephemeral (short-lived) carbonate systems on top of lowstand coastal deposits is a simple mechanism that would explain the initiation, often contemporaneous, of many modern and ancient barrier reefs in the world [Droxler et al., 2003].
Figure 11. Interpreted seismic profile showing prograding shelf edge delta and drowned transgressive barrier reef, which formed at the beginning of the last transgression on top of last lowstand coastal beach ridges and subsequently drowned during the end of the transgression. The seismic profile is courtesy of Fugro Multi Client Services, Inc.
Download figure to PowerPoint
 In the GoP, distinct siliciclastic depositional environments, juxtaposed laterally from the deepest to the shallowest settings, are observed seaward of the modern shelf edge. They are represented by the submarine fan and deep sea channels in Moresby Trough, the flat seafloor of the ponded turbidite basin in the central Pandora, the muddy slope slumping deposits in Pandora Trough, and the prograding lowstand shelf edge delta in the northern Ashmore Trough (Figures 5a–5c). Identical environments can be interpreted in the kilometers thick siliciclastic infill of Flinders paleotrough. In this trough, more than 3 km thick siliciclastic deposits of the late Pliocene–Pleistocene represent a vertically stacked succession of depositional environments representing deep sea fans, flat floored ponded turbidite basin, slumping slope, prograding lowstand shelf edge delta, and aggrading shelf (Figure 9).