1.Sidescan and underway geophysical surveys have provided bathymetry, acoustic imagery, magnetization, and gravity maps that permit detailed reconstructions of the spreading history (Taylor et al., 1995, 1996; Goodliffe et al., 1997, and A. Goodliffe, unpubl. data).
2.Multichannel seismic reflection surveys reveal the upper crustal architecture of the rifting region, including the presence of low-angle normal faults (Figs. 3, 4, 5; Mutter et al., 1996; Taylor et al., 1996).
3.The PACLARK and SUPACLARK series of cruises in 1986-1991 (Binns et al., 1987, 1989, 1990; Lisitsin et al., 1991; Benes et al., 1994) included dredging, coring, camera and video observations, and seven Mir submersible dives. The bottom samples include normal mid-ocean ridge basalt (N-MORB) from the youngest spreading segments, as well as greenschist facies metamorphics from the lower north flank of Moresby Seamount. In contrast, a 1995 site survey dredged late Pliocene (synrift) sedimentary rocks from the upper south flank of Moresby Seamount‹apparently precluding a metamorphic core complex origin for this feature (Taylor et al., 1996).
4.Abers (1991) and Abers et al. (1997) determined source parameters and relocated earthquakes in the rifting region. The focal mechanisms are all extensional or strike-slip with northerly tension axes (T-axes; Fig. 2). Several are consistent with slip on shallow-dipping normal faults.
5.Studies of metamorphic core complexes on the Papuan Peninsula, D'Entrecasteaux, and Misima Islands (Fig. 2) show that (A) they are associated with Pliocene/Pleistocene granodiorite intrusions and amphibolite-facies ductile shear zones, (B) they have been rapidly exhumed from ~30 km depth (7-11 kilobar [kb]) in the last 4 m.y., (C) uplift continues (forming topography up to 2.5 km), and (D) they are very three dimensional and regionally discontinuous (or varying in grade) along strike (Davies and Warren, 1988, 1992; Hill, 1987, 1990, 1994, 1995; Hill et al., 1992, 1995; Hill and Baldwin, 1993; Baldwin et al., 1993; Lister and Baldwin, 1993; Baldwin and Ireland, 1995).
6.The Papuan Ultramafic Belt (PUB) is a late Paleocene to early Eocene supra-subduction zone ophiolite with gabbros and boninites 40Ar/39Ar dated at 59 Ma (R. Duncan, pers. comm., 1993; Walker and McDougall, 1982) and P4 (late Paleocene) foraminifer-bearing micrites overlying the basalts with tonalite-diorite-dacite intrusions K/Ar dated at 57-47 Ma (Rogerson et al., 1993). This revision to dating of the Papuan Peninsula basement allows a simplified geological evolution for the region, as outlined below.
Papuan Crustal Evolution
Paleogene Subduction and Collision
Much of the Papuan Peninsula is 1-3 km above sea level and is underlain by crust 25-50 km thick (Finlayson et al., 1976). Major orogenic thickening of the crust occurred following the northeast subduction and partial accretion of a thick sequence of dominantly Cretaceous to Eocene strata beneath a late Paleocene-early Eocene island arc that includes the PUB, Milne Basic Complex, and Cape Vogel boninites (Davies and Jaques, 1984; Davies et al., 1984; Rogerson et al., 1987, 1993). Collision of the Australian (Papuan) continental margin plateau caused subduction to cease and uplifted the accretionary complex (Owen Stanley metamorphics) by the early Miocene (Rogerson et al., 1987). Metabasites in the Emo metamorphics and Suckling-Dayman massif (the latter with a cover of Maastrichtian micrites) have been exhumed from 7-12 kb (25-35 km) and may represent slivers of the subducted Cretaceous oceanic crust (Davies, 1980; Worthing, 1988).
The preMiocene geology of the islands on the Pocklington and Woodlark Rises is similar to that of the Papuan Peninsula, with Owen Stanley metamorphics in the south (Misima, Tagula, and Rossel Islands) and Milne Basic Complex outcrops on Woodlark Island (Davies and Smith, 1971; Ashley and Flood, 1981). Likewise, late Paleocene volcanics occur at the base of the Nubiam 1 well (Stewart et al., 1986), west of the Trobriand Islands, and the metamorphic core complexes on the D'Entrecasteaux Islands have a core of Owen Stanley metamorphics and a cover of unmetamorphosed PUB ultramafics (Davies and Warren, 1988).
Miocene-Quaternary Arc and Forearc
Superimposed on this Paleogene basement is widespread middle Miocene to Holocene calc
alkaline and shoshonitic magmatism (Smith and Milsom, 1984) associated with southwest subduction of the Solomon Sea Basin at the Trobriand Trough. Active arc volcanism and a deforming accretionary prism are compatible with present-day slow subduction at the Trobriand Trough (Hamilton, 1979; Davies and Jaques, 1984; Lock et al., 1987), though this remains controversial given the small number of intermediate-depth earthquakes beneath the region (Abers and Roecker, 1991) and the lack of 10Be in the arc lavas (Gill et al., 1993). The geochemistry of the volcanics reveals melting and mixing of at least three magma sources: (1) subduction-modified mantle supplied the calc-alkaline arc volcanism; (2) this mantle, with the addition of partial melts from upwelling aesthenosphere, produced the comenditic (transitional basalt-peralkaline rhyolites) series around Dawson Strait; and (3) contamination by lower crust of Australian affinity formed minor high-K trachytes (see Smith, 1976; Hegner and Smith, 1992; Stolz et al., 1993; and references therein).
The Cape Vogel (including Trobriand) Basin is a Neogene forearc basin, characterized by middle Miocene subsidence, volcanism, deep-marine sedimentation, late Miocene uplift and erosion (1-2 km) of the margins, and Pliocene coarse clastic (from uplift of the Papuan Peninsula and D'Entrecasteaux Islands to the south) and Quaternary carbonate, shallow-water sedimentation during broad subsidence (Tjhin, 1976; Stewart et al., 1986; Francis et al., 1987; Davies and Warren, 1988). The basin experienced late Miocene/Pliocene inversion in the northwest, but its center continues to subside in the southeast (Pinchin and Bembrick, 1985). The Lusancay Trobriand-Woodlark Islands sit atop an outer forearc structural high with an associated 150-200 mgal free-air gravity anomaly.
The Trobriand Trough, the outer forearc structural and gravity high, and the Cape Vogel forearc basin terminate near Woodlark Island. Farther east, the Woodlark and Pocklington Rises were not a Pliocene/Pleistocene arc-forearc system. Rather, the northern edge of the eastern Woodlark Rise was a transform margin, and seismicity and sidescan data indicate that it is still an active right lateral fault. Thus, the Woodlark Basin did not originate as a backarc basin, in that the eastern Woodlark and Pocklington Rises were not active island arcs (Weissel et al., 1982). Nevertheless, the locus of present rifting (see below) bisects an inherited crustal asymmetry, with the Neogene forearc basin to the north and the Paleogene accretionary/collision complex and Neogene backarc to the south.
Pliocene/Pleistocene Rifting
Shallow seismicity, with extensional and strike-slip focal mechanisms having northerly T-axes, is concentrated east of the D'Entrecasteaux Islands and extends westward into the Papuan Peninsula at 9°-10°S to about 148°E (Fig. 2; Weissel et al., 1982; Abers, 1991; Abers et al., 1997). Pliocene/Pleistocene extension has produced three flooded grabens (Mullins Harbor, Milne Bay and Goodenough Bay) on the eastern extremity of the Papuan Peninsula with associated rift-flank subaerial uplift to over 500 m (Smith and Simpson, 1972). Metamorphic core complexes on the D'Entrecasteaux Islands and in the Suckling-Dayman massif on the Papuan Peninsula were also exhumed in the Pliocene/Pleistocene (Davies, 1980; Davies and Warren, 1988; Hill, 1990). The best structural studies, geothermometry, geobarometry, and age dating of these complexes have been done on Goodenough and Fergusson Islands (Davies and Warren, 1992; Hill et al., 1992, 1995; Hill and Baldwin, 1993; Baldwin et al., 1993; Lister and Baldwin, 1993; Hill, 1994; Baldwin and Ireland, 1995). There, normal movement along a 0.3- to 1.5-km-thick ductile mylonitic shear zone resulted in the uplift of deep metamorphic rocks and the juxtaposition of unmetamorphosed cover rocks. Granodioritic intrusion then focused uplift on several domes, offset by strike-slip faults. The ductile shear zones were brecciated and truncated by brittle faults late in their history. These domal structures are juxtaposed along strike with regions of significantly less unroofing, such as the low-grade (greenschist) eastern halves of both Normanby and Misima Islands.
Woodlark Basin Evolution
The oldest magnetic anomalies (An.3R), in the extreme east of the basin, indicate that seafloor spreading began by 6 Ma (Taylor, 1987; Taylor and Exon, 1987). Spreading has sequentially transgressed westward, stepping across Simbo Transform (156.5°E) about 4 Ma and Moresby Transform (154.2°E) about 1.9 Ma to reach its current tip at 151.7°E (Figs. 1, 2). Bruhnes Chron spreading rates decrease from 67 mm/yr at 156.2°E to 36 mm/yr at 152°E (Fig. 1). The 500-km
long spreading axis reoriented synchronously about 80-100 ka (Goodliffe et al., 1997, and unpubl. data).
The sidescan and geophysical data show that the rifting-to-spreading transition involves both nucleation of discrete spreading cells and organized ridge propagation (Taylor et al., 1995). Two ridge propagation events into the margin at 153°E formed continental slivers surrounded by oceanic crust. Spreading is about to propagate into this margin again. Rifting of the conjugate margins continues for ~1 m.y. after spreading has separated them. Extension does not immediately localize to the ridge axis, as shown by the present overlap between spreading and seismogenic margin faulting, and by inwardly curved seafloor fabric and magnetic anomalies that require nonrigid margin reconstructions. The initial spreading system lacks transform faults and has both overlapping and orthogonally offset segments (Figs. 1, 2). The 50-km-offset Moresby Transform Fault formed by cutting through rifted crust to join overlapping spreading segments of initial oceanic crust. It is not contiguous with transfer faults in the rifted margins. The initial spreading system evolves by ridge propagation, transform development, and ridge jumping/rotation.
Seismic reflection data indicate a very sharp (1 to 2 km wide at the surface) transition from rifted crust to oceanic crust. There are no dipping reflector sequences indicative of excessive lava production and high degrees of mantle partial melting, but there are small volcanoes a few kilometers in diameter that are often erupted along margin faults. Indeed, the initial seafloor spreading lavas indicate low degrees of partial melting: basalts from the youngest spreading segment, just east of Moresby Seamount, have Na8 = 3.1 (Binns and Whitford, 1987). Other young axial lavas include FeTi basalts and low- and high-Si andesites with evidence for both mantle heterogeneity and crustal contamination.
Secondary convection in the mantle, induced by higher horizontal temperature gradients created during rifting of the thicker continental margins in the west, may explain the contrast in geophysical characteristics of the oceanic basins to either side of Moresby Tranform (Martinez et al., submitted 1998). With respect to the eastern basin, the western basin (1) is ~500 m shallower; (2) has Bouguer gravity anomalies that are >30 mgals lower; (3) has magnetic anomaly and modeled seafloor magnetization amplitudes that are respectively 100% and 50% higher; (4) has spreading centers with rifted axial highs as against axial valleys; (5) has smoother seafloor fabric; and (6) has exclusively nontransform spreading center offsets in contrast to transform faults and fracture zones that extend to the basin edges, all despite having lower spreading rates.
Drilling Area
The rifting region just ahead of the apex of spreading has been imaged by several seismic reflection surveys (Figs. 3, 4, 5; Mutter et al., 1993, 1996; Goodliffe et al., 1993; Taylor et al., 1996). North of Moresby Seamount, a low-angle (25°-30°) normal fault dips north beneath a down
flexed prerift sedimentary basin and basement sequence, unconformably onlapped by synrift sediments that are cut by higher angle normal faults with a zigzag pattern in plan view (Figs. 2, 3, 4, 5). The seismic stratigraphy can be reasonably jump-correlated to that in the Trobriand Basin and is interpreted to be a Pliocene to Quaternary synrift sequence lying unconformably above a Miocene forearc basin sequence on Paleogene volcanic and metamorphic basement (Fig. 5). To the south of Moresby Seamount, high-relief rotated fault blocks are commonly overlain by only minor ponded sediments.
Shallow (2-10 km) normal and strike-slip faults, with northerly T-axes, bound the north side of the rifting-spreading transition (Fig. 2; Abers, 1991; Taylor et al., 1995; Abers et al., 1997). All of the earthquake hypocenters occur within or north of the rift graben, and there are no major extensional structures north of the graben-bounding antithetic fault. Without local seismometers, the teleseismicity cannot be definitively associated with the low-angle reflector, but there is no more likely candidate structure. Furthermore, the seismic stratigraphy of the profiles in Figure 3 cannot be matched without recent faulting on the low-angle reflector.
Two dredges of the northern flank of Moresby Seamount recovered metabasic greenschists, metagabbro, pelitic schist, and minor siliceous phyllite and microgranite that is material similar to the low-grade (greenschist) metamorphics on eastern Normanby and Misima Islands, not the core complex amphibolite metamorphics on Goodenough, Fergusson, and northwest Normanby (H. Craig, 1986, unpubl. SIO cruise report; Binns et al., 1987; J. Hill, pers. comm., 1996). In contrast, a dredge from 541 to 1211 m (~0.7-1.6 s two-way traveltime [TWT]) on the upper southern flank of the seamount recovered late Pliocene (N21 = 1.9-3.1 Ma) clastic sedimentary rocks of equivalent facies to the Awaitapu Formation of the Trobriand region in the Cape Vogel Basin (Francis et al., 1987). Benthic foraminifers indicate sediment deposition in water depths of 340 800 m (J. Resig, pers. comm., 1995). These rocks are equivalent to those that we infer lie near the base of the synrift cover sequence in the rift basin and on the northern margin.
A cross section consistent with the available seismic and dredge data is shown in Figure 5 (Taylor et al., 1996). At the end of the Miocene, the Paleogene basement and a forearc basin filled with Miocene sediment were being eroded at or near sea level. Pliocene rifting formed sediment-filled grabens in the southern orogenically thickened arc province, accompanied by gradual subsidence of the thinner, colder (and therefore stronger) forearc to the north (inferred Pliocene sediments are dotted in Fig. 5). Quaternary stretching localized on a low-angle normal fault (the antithetic hanging-wall fault accommodated little additional extension). The northern margin flexed down southward and was onlapped by sediments delivered via submarine channels incising northward. Recently, continued extension on the low-angle fault variably collapsed the hanging-wall graben, into which sediments are now prograding from the north.
This interpretation predicts about 12 km of heave on the low-angle fault. This compares with at least 130 km of total extension in the 4 m.y. prior to spreading at this longitude, calculated from the pole of opening derived from seafloor spreading magnetic anomalies, and probably greater amounts back to the beginning of spreading at >6 Ma (Taylor et al., 1996). Given only minor extension of the northern, flexed margin, we infer that the locus of current extension must be the northernmost of a series of similar structures that extended weak crust to the south, forming the block-faulted Pocklington Rise. The regional estimates predict that this rugged province of mainly inactive faults accommodated >120 km of total strain as it collapsed from heights comparable to the 3-km-high Owen Stanley Ranges that form the backbone of the Papuan Peninsula. We do not know where to locate so much extension given a current total width of 200 km for this province.