Exhumation of the Dayman dome metamorphic core complex, eastern Papua New Guinea

Transcription

1 J. metamorphic Geol., 2009, 27, doi: /j x Exhumation of the Dayman dome metamorphic core complex, eastern Papua New Guinea N. R. DACZKO, 1 P. CAFFI, 1 J. A. HALPIN 1,2 AND P. MANN 3 1 GEMOC ARC National Key Centre, Department of Earth and Planetary Sciences, Macquarie University, Sydney New South Wales 2109, Australia (ndaczko@els.mq.edu.au) 2 ARC Centre of Excellence in Ore Deposits, University of Tasmania, Private Bag 126, Hobart Tasmania 7001, Australia 3 Institute for Geophysics, Jackson School of Geosciences, University of Texas at Austin, Austin Texas , USA ABSTRACT The 750 km 2 Dayman dome of the Late Cretaceous Suckling-Dayman massif, eastern Papua New Guinea, is a domed landform that rises to an elevation of 2850 m. The northern edge of the dome is a fault scarp >1000 m high that is now part of an active microplate boundary separating continental crust of the New Guinea highlands from continental and oceanic crust of the Woodlark microplate. Previous work has shown that a parallel belt of eclogite-bearing core complexes north-east of the Dayman dome were exhumed from up to kbar in the last few millions of years. The remarkably fresh and lightly eroded scarp of the Dayman dome exposes shallowly-dipping mylonitic (S1) metabasite rocks (500 m thick) on the northern flank of Mount Dayman. Field relationships near the base of this scarp show a cross cutting suite of ductile and brittle meso-structures that includes: (i) rare ductile S2 folia with a shallowly ESE-plunging mineral elongation lineation defined by sodic-calcic blue amphibole; (ii) narrow steeply-dipping ductile D2 shear zones; and (iii) semi-brittle to brittle fault zones. Pumpellyite-actinolite facies assemblages reported by previous workers to contain local aragonite, lawsonite and or glaucophane are found in the core of the complex at elevations greater than 2000 m. These assemblages indicate peak metamorphic pressures of kbar, demonstrating exhumation of the core of the Dayman dome from depths of km. The S1 metamorphic mineral assemblage in metabasite includes actinolite-chlorite-epidote-albite-quartz-calcite-titanite, indicative of greenschist facies conditions for the main deformation. New mineral equilibria modelling suggests that this S1 assemblage evolved at kbar at 425 C. Modelling variable Fe 3+ indicates that the sodic-calcic blue amphibole (D2) formed under a higher oxidation state compared with the S1 assemblage, probably at <4.5 kbar. A SE-dipping, Mio-Pliocene sedimentary sequence (Gwoira Conglomerate) forms a hangingwall block juxtaposed by low-angle fault contact with the metabasite footwall. Prehnite-bearing D3 brittle fault zones separate the two blocks and likely accommodated the final exhumation of the S1 greenschist facies assemblage in the footwall. These results indicate that the extensive Mt Dayman fault surface coincides with a domed S1 greenschist facies foliation that was last active at >20 km depth. Exhumation of this foliation must therefore be controlled by brittle faults of the active microplate boundary that are largely not observed in the study area. The structural record of the final exhumation of the Dayman dome to the surface was likely lost as a result of erosion, poor exposure or wide spacing of semi-brittle to brittle fault zones. Key words: actinolite; ferrobarroisite; ferrowinchite; greenschist; metabasalt; metabasite; prehnite; THERMOCALC pseudosection. INTRODUCTION AND SIGNIFICANCE Low-angle normal faults and associated metamorphic core complexes are globally recognized along slowspreading mid-ocean ridges (e.g. Mutter & Karson, 1992; Cann et al., 1997; Blackman et al., 1998; Tucholke et al., 1998, 2008; Ohara et al., 2001; Okino et al., 2004; Drolia & DeMets, 2005) and in areas of distributed, but now inactive, continental extension such as the western Cordillera of North America (e.g. Davis & Coney, 1979) and active areas of continental extension that include the Gulf of Aden (Manighetti et al., 1997), the Gulf of California (Langenheim & Jachens, 2003) and the Laptev Sea (Drachev et al., 2003). Work over the past 30 years has shown two parallel bands of younger and possibly active core complexes in eastern Papua New Guinea (PNG) that are associated with broad extension related to propagation of the Woodlark basin spreading system and diffuse extension along an active microplate boundary known from GPS-based studies (Davies & Jaques, 1984; Davies & 405

2 406 N. R. DACZKO ET AL. Warren, 1988; Lister & Davis, 1989; Hill, 1994; Martinez et al., 2001; Wallace et al., 2004; Little et al., 2007; Spencer & Ohara, 2008). The relationship between thermochronology and metamorphic geology in rapidly exhumed or exhuming metamorphic core complexes in eastern PNG is an area of current research both from a lithological and a tectonic point of view (Davies & Smith, 1971; Ollier & Pain, 1981; Davies & Warren, 1988; Hill et al., 1992; Baldwin et al., 1993, 2004; Hill, 1994; Baldwin & Ireland, 1995; Little et al., 2007; Monteleone et al., 2007; Webb et al., 2008). Previous studies of changing mineral paragenesis of metamorphic core complexes during exhumation are restricted because shallow-level exhumation and brittle deformation of metamorphic core complexes commonly outlasts thermally driven metamorphic recrystallization of the exhuming and cooling rock mass. The mineral record for many metamorphic core complexes may therefore be incomplete or inadequate for recording the youngest and shallowest phases of their unroofing. However, P T-sensitive minerals may form in late stage shears and veins that cut structures or metamorphic assemblages that persist metastably during exhumation. Linking a metamorphic P T path with the kinematic history recorded in these late-stage shears and veins may constrain the mechanisms and timing of exhumation of deep crustal rocks in metamorphic core complexes as seen in south-central Turkey (Whitney & Dilek, 2000). Minor blue amphibole-bearing foliations and epidote-prehnite-bearing veins cut the main greenschist facies mylonite in the Dayman dome metamorphic core complex of the Suckling-Dayman massif, eastern PNG (Fig. 1). In this contribution these critical mineral assemblages are used to provide pressure limits for three points along the exhumation path of the metamorphic core complex. The Dayman dome is situated on the Papuan Peninsula adjacent to the Woodlark basin (Fig. 1a). This region is one of only a few well-recognized regions worldwide that spans the transition from active continental rifting to active seafloor spreading; the others include the Red Sea (Martinez & Cochran, 1988), the Gulf of Aden (Manighetti et al., 1997), the Gulf of California (Langenheim & Jachens, 2003) and the Laptev Sea (Drachev et al., 2003). Within the Woodlark basin spreading system, magnetic anomalies indicate that seafloor rifting has propagated westward since 6 Ma (Taylor et al., 1999) and continental extension has produced ÔactiveÕ subaerial metamorphic core complexes in eastern PNG. High-grade lower crustal metamorphic core complexes have been studied in eastern PNG on the DÕEntrecasteaux Islands (Fig. 1b) where rocks have been uplifted that formed under a large range of depth temperature conditions (Davies & Smith, 1971; Ollier & Pain, 1981; Davies & Warren, 1988; Hill et al., 1992; Baldwin et al., 1993, 2004, 2008; Hill, 1994; Baldwin & Ireland, 1995; Little et al., 2007; Monteleone et al., 2007; Webb et al., 2008). However, recent studies of emergent, late Quaternary coral reefs and Neogene sedimentary sections in eastern PNG suggest that the DÕEntrecasteaux Islands have been stable or subsiding for the past 0.5 Myr (Mann & Taylor, 2002). The main locus of active rifting associated with rapid uplift of Holocene reefs and late Neogene sedimentary rocks is to the south-west of the islands at Goodenough Bay (Fig. 1b) near the Dayman dome (Mann & Taylor, 2002) where actively uplifting moderate- to high-p metabasite rocks are exposed below the Owen Stanley fault system (Davies & Smith, 1974; Davies, 1980; Worthing, 1988). Objectives of this study This contribution presents the first detailed metamorphic study of the actively exhuming Dayman dome. New data and field observations from metabasite rocks from the northern flank of the Dayman dome (Fig. 2) outline the structural and kinematic history of the shear zones and faults. These data are integrated with quantitative calculated phase diagrams (in a model close to a natural system Na 2 O-CaO-FeO-MgO-Al 2 O 3 -SiO 2 -H 2 O-TiO 2 -Fe 2 O 3 ) to determine the P T history recorded in metamorphic assemblages during exhumation of the Suckling- Dayman massif. These results are evaluated in terms of the exhumation of deep crust in metamorphic core complexes and the overall tectonic history of eastern PNG. Our results suggest that the metamorphic core complex bounding shear zone of the Dayman dome was last active at >20 km depth and therefore records an early part of the exhumation story. If this observation is common in other metamorphic core complexes then the relationship between early and late exhumation of deep crust may require reevaluation, especially in areas of rapidly changing plate boundary conditions. REGIONAL GEOLOGY OF METAMORPHIC CORE COMPLEXES IN EASTERN PNG The Oligocene to recent tectonics of eastern PNG continues to be controlled by the west south-west convergence between the Pacific and Australia plates (Fig. 1a). Regional GPS surveys by Wallace et al. (2004) have shown that the intervening zone of continental and arc convergence in central and eastern PNG is characterized by small microplates that include the North Bismarck, South Bismarck, New Guinea highlands, Solomon Sea and Woodlark microplates (Fig. 1a). Most of these microplates are rotating rapidly about nearby vertical axes as a result of localized collisions. Webb et al. (2008) have proposed that these rotations have led to a process of Ôsubduction inversionõ in which previously subducted material including the core complexes are now being exhumed along active microplate boundaries.

3 EXHUMATION OF THE DAYMAN DOME, PNG 407 Fig. 1. (a) Regional tectonic setting of eastern Papua New Guinea, indicating locations discussed in the text. Dashed arrows along the OSFZ are predicted motion across the fault of the Woodlark microplate relative to a fixed New Guinea highlands microplate from the best-fit model of Wallace et al. (2004). The longest arrow represents 22.5 mm relative motion per year. (b) Map of eastern PNG and the DÕEntrecasteaux Islands [Goodenough (GI), Fergusson (FI) and Normanby (NI)]. PP, prehnite-pumpellyite; PA, pumpellyiteactinolite; A,G&E, amphibolite, granulite and eclogite; A,EB&G, amphibolite, epidote blueschist and greenschist facies; TF, Trobriand Fault; GB, Goodenough Bay; MB, Milne Bay; MS, Morseby Seamount. The Gwoira Conglomerate (1 km thick; Davies & Smith, 1974) is deposited onto the exposed metabasite Dayman dome and Biman dip slopes (modified from Hill & Baldwin, 1993; Wallace et al., 2004; Little et al., 2007). In eastern PNG, a large active fault system called the Owen Stanley fault zone (OSFZ) separates the New Guinea highlands microplate from the Woodlark microplate to the north-east (Fig. 1a). Wallace et al. (2004) used GPS results to define a pole of rotation that describes rapidly changing kinematics along the OSFZ. This includes a northern area of large folds of Quaternary sedimentary rocks and large thrust earthquakes where their pole predicts about 20 mm year )1 of convergence perpendicular to the OSFZ north of 8 S. In a central area their pole predicts mm year )1 of left-lateral strike-slip faulting consistent with geological evidence (Davies & Smith, 1971). In the area of the Dayman dome, motion along the OSFZ microplate boundary is predicted to be oblique-left slip and changing at the tip of the Papuan

4 408 N. R. DACZKO ET AL. Fig. 2. Geological map and metamorphic zones of the Suckling-Dayman massif and surrounding areas (modified from Davies & Smith, 1974; Davies, 1980). Boundaries of metamorphic grade represent the limit of field documentation and are not isograds. GB, Goodenough Bay. All faults are inferred active. N-S schematic cross section is shown to illustrate the structural relationships.

5 EXHUMATION OF THE DAYMAN DOME, PNG 409 peninsula to about 12 mm year )1 of roughly orthogonal opening near the eastern tip of the Woodlark basin spreading system (Fig. 1b; Wallace et al., 2004). Most previous workers have inferred that the large metamorphic core complexes in this area like the Dayman dome are produced by the extensional component of opening along the New Guinea highlands- Woodlark microplate boundary (Davies & Jaques, 1984; Davies & Warren, 1988; Hill, 1994; Martinez et al., 2001; Little et al., 2007; Spencer & Ohara, 2008). As in the case of the chain of metamorphic core complexes underlying the DÕEntrecasteaux Islands, the Dayman dome exhibits a broad topographic dome that is onlapped or faulted against the coastal plain of the Papuan Peninsula (Fig. 3). The Suckling-Dayman massif and Dayman dome consist primarily of metamorphosed basalt and minor limestone (Davies, 1980). A Late Cretaceous age is indicated by planktonic foraminifera recovered from pelagic limestone interbedded with basalt (Smith & Davies, 1976). The metabasite rocks are chemically homogeneous with a composition approximating average mid-ocean ridge Fig. 3. (a) Shuttle Radar Topography Mission (source: image view looking south of the Dayman dome. A dashed line marks the MaiÕiu Fault, part of the Owen-Stanley fault zone. GC, Gwoira Conglomerate. Sample sites discussed in text are labelled. (b) (d) Lower hemisphere equal area projection stereoplots of poles to foliation and lineation data from the east and north flank of the Dayman dome.

6 410 N. R. DACZKO ET AL. basalt (MORB; Smith & Davies, 1976; Davies, 1980). The unit names and stratigraphic framework used in this contribution follow the 1: Geological series map of the Tufi-Cape Nelson region (Davies & Smith, 1974). NEW FIELD RELATIONSHIPS AND PETROLOGY OF THE DAYMAN DOME The outcrops described here are within 1 5 km of the Pumani (AGD66 Grid Reference N E) and Biniguni ( N E) villages (Figs 2 & 3). The Goropu Metabasalt of the Suckling-Dayman massif at these localities is uniform metabasite mylonite (500 m thick; Davies, 1980) that displays welldeveloped outcrop scale extensional shear bands (S-C fabrics; Fig. 4a). A comparable actinolite-bearing greenschist facies shear zone foliation (S1) is observed in all metabasite outcrops near Pumani and Biniguni. S1 at each site is correlated on the basis of similar metamorphic mineralogy and by containing a similarly N to NE-trending mineral elongation lineation (L1, Fig. 3). However, the orientation of S1 is domed and therefore different at the two sites and lies parallel to the slope of the dome (Fig. 3). For example, S1 dips shallowly towards the ENE (average S1 = ENE) near Pumani (Fig. 3b) and shallowly towards the NNE (average S1 = NNE) near Biniguni (Fig. 3d). Cross cutting structures were not observed at most outcrops and therefore the metamorphic conditions that accompanied the pervasive deformation event (D1) are the focus of this contribution. The observed second-generation structures at Pumani are different to those at Biniguni. We therefore define the second-generation foliations as S2(P) for those near Pumani and S2(B) for those near Biniguni. The relative timing between S2(P) in the east and S2(B) in the west is unknown as no cross cutting relationship was observed. Therefore the two structures are Fig. 4. (a) View looking WNW of S1 folia (S-planes) that anastomose between extensional shear bands (C -planes) spaced 1 2 m apart, site 0630, Yatap Creek. Curvature of S-planes indicates top down to the NNE sense-of-shear. (b) View looking ESE of a narrow steeply-dipping S2(B) shear zone, site 0664, Biniguni River. The sense of curvature and transposition of S1 into S2(B) indicates a top down to the NNE sense-of-shear. There may be a continuum between D1 C shear bands and steeply-dipping D2 shear zones. D1 C shear bands are continuous with S1 and diffuse, whereas D2 steeply-dipping shear zones are wider and evolve to semi-brittle conditions. Lens cap is 60 mm across. (c) View looking down onto an outcrop with a dyke cutting S1, site 0664, Biniguni River. The dyke lacks chilled margins. Lens cap is 60 mm across. (d) View looking E of prehnite (prh) and epidote (ep) veins in the D3 fault zone, site 0604, Pumani River. Lens cap is 60 mm across.

7 EXHUMATION OF THE DAYMAN DOME, PNG 411 grouped together in D2 as they formed under broadly similar metamorphic conditions. Their ductile nature distinguishes them from rare semi-brittle thirdgeneration structures (D3). S1 near Pumani and S2(P) are similarly shallowly dipping (Fig. 3). However, S2(P) is easily distinguished from S1 as the second-generation foliation contains blue amphibole instead of green actinolite. The orientation of the associated mineral elongation lineation is also distinct for the two foliations. L1 at Pumani trends NNE whereas L2 at Pumani trends ESE (Fig. 3). The blue amphibole mineral elongation lineation was only observed at three outcrops on the eastern flank of the dome and therefore represents a minor feature in the structural development of the dome, possibly of local P T X significance. The shallowly-dipping S2(P), ESE-plunging L2 and blue amphibole were not observed in the field on the northern flank of the dome near Biniguni. Later thin section analysis identified minor blue amphibole in one sample from the northern flank of the dome, suggesting one minor occurrence of S2(P) near Biniguni. Instead, ductile recrystallization in narrow steeply-dipping shear zones was observed as the second-generation structure at Biniguni (Fig. 4b). These D2 shear zones are similarly oriented to D1 C shear bands and the foliation (S2(B)) in these narrow shear zones contains a similar greenschist facies mineral assemblage to S1. Therefore, the steeplydipping D2 shear zones may be further developed versions of the D1 C shear bands and there may be a continuum between the two structures (i.e. one evolved into the other). The wider zones of deformation and a ductile overprinted by semi-brittle character distinguish the D2 shear zones from D1 C shear bands in the field. However, their distinction is not critical for the metamorphic interpretations of this contribution. Rare mafic dykes that cut S1 and lack chilled margins (Fig. 4c) show evidence of minor recrystallization, suggesting they intruded during or after ductile D1 deformation and before the onset of brittle deformation. D3 structures are distinguished on the basis of their semi-brittle character. These comprise semi-brittle to brittle fault zones that may be up to a few metres wide with epidote-prehnite veins (Fig. 4d). We highlight three key metamorphic assemblages in these outcrops. Greenschist facies metamorphic minerals define the first assemblage (S1) and are common. This assemblage was observed at every metabasite outcrop of the dome examined. A rare blue amphibole-bearing assemblage defines the second key assemblage mainly observed on the eastern flank of the dome. The epidote-prehnite association in veins forms a third key metamorphic assemblage that was only observed at the contact between the metabasite footwall and Gwoira Conglomerate hangingwall. All metabasite samples of the Dayman dome examined in this study (excluding the mafic dykes) contain a well-developed S1 foliation and L1 lineation. S1 is defined by aligned actinolite, albite and epidote (e.g. sample 0668A; Fig. 5a). Chlorite, titanite, calcite and quartz are commonly also part of the S1 assemblage. Accessory minerals include apatite, pyrite, muscovite, potassium feldspar, chalcopyrite, rutile and zircon. The metabasite samples are fine-grained (commonly mm but up to 1.5 mm). Actinolite and epidote vary in grain size from 0.1 to 0.2 mm (e.g. samples 0629, 0676, majority of site 0677) up to 0.7 mm across (e.g. samples 0668A, 0672 & 0677H). The coarsest grained samples are located near Biniguni where albite ribbons are up to 1 mm long (Fig. 5a) and large round to elliptical single grains or mineral clusters of epidote (up to 1.5 mm across) are enveloped by the fine-grained S1 (Fig. 5b). The single grains of epidote are optically zoned (e.g. sample 0676; Fig. 5b) and the mineral clusters of epidote (e.g. sample 0677G) are zoned from a pale green coarse-grained core to a pale brown fine-grained rim. Chlorite and albite are common minerals in the strain shadows of the coarse epidote grains or clusters. Chlorite may partially pseudomorph epidote or actinolite and the S1 foliation commonly envelops the larger actinolite and epidote grains (e.g. sample 0605). Minor calcite and quartz veins (up to 2 mm across) are deformed and aligned with S1 in some places. Chlorite may pseudomorph epidote or actinolite adjacent to veins. Samples commonly show multiple generations of veins. For example, in sample 0630, pre-s1 quartz veins are recrystallized and cut by post-s1 calcite veins. S2(P) blue amphibole is restricted and concentrated in narrow bands (up to 3 mm wide) of S2(P) folia (e.g. sample 0605; Fig. 5c). The blue amphibole is sodic-calcic ferrowinchite or ferrobarroisite (see Mineral chemistry) and is aligned with S2(P) in the interior of the bands but may also form random splays of elongate diamond-shaped minerals that pseudomorph S1 minerals adjacent to the bands (Fig. 5d). S1 chlorite and actinolite are pseudomorphed by the S2(P) blue amphibole in some places indicating that S2(P) postdates S1. Accessory muscovite, calcite and very rare zircon (<10 lm across) are associated with the blue amphibole textures (Fig. 5d). Titanite partially replaces rutile grains (200 lm) where the rutile is left as many small remnants enveloped by titanite. An oxidized iron staining is associated with S2(P) folia. Samples 0677A and 0677I are representative of the narrow (<250 mm) moderately to steeply-dipping D2 shear zones at Biniguni. The deformation mode within the narrow shear zones spans ductile to semi-brittle. The ductile S2(B) foliation is defined by very finegrained chlorite, epidote and albite (<0.02 mm; e.g. sample 0677A; Fig. 5e). However, these structures evolved to semi-brittle conditions. The semi-brittle component is defined largely by very fine-grained

8 412 N. R. DACZKO ET AL. epidote and fine-grained gouge (sub-microscopic; e.g. sample 0664; Fig. 5f). The gouge contains common angular lithic and monomineralic clasts derived from the host rock (Fig. 5f). Abundant calcite and quartz veins ( mm wide) are deformed and aligned with the D2 shears faults. The dykes preserve a doleritic texture defined by elongate and randomly oriented augite grains (e.g. sample 0661C; Fig. 5g). The augite grains are euhedral and mineral chemistry data (see Mineral chemistry) indicate that the augite grains are the igneous minerals least affected by the thermal metamorphism of the dykes. Plagioclase is everywhere pseudomorphed by albite with randomly oriented chlorite, muscovite and fine-grained acicular epidote (Fig. 5g). Fine-grained domains of the dykes show the same texture and mineralogy as the domains with larger grain size (e.g. sample 0661C). A 2 m wide D3 semi-brittle to brittle fault zone was observed at the contact between the Gwoira Conglomerate and Goropu Metabasalt (site 0604; Fig. 4d). Narrow D3 fault zones most likely overlap in time with the D2 semi-brittle structures as these contain very fine-grained gouge and define a rare semi-brittle to brittle overprint of the ductile S1 and S2(B) foliations. The gouge is associated with an oxidized iron staining and contains common angular lithic and monomineralic clasts derived from the host rock. Rare D3 epidote and prehnite veins are associated with and or cut the semi-brittle to brittle faults (e.g. sample 0604; Fig. 5h). The prehnite veins are commonly in the centre of epidote veins (e.g. sample 0604; Fig. 5h). MINERAL CHEMISTRY In this section the major features of mineral chemistry in these rocks are outlined to characterize the metamorphic evolution evident in the textures (Table 1). Mineral compositions were determined using a Cameca SX100 electron microprobe in the GEMOC laboratories, Macquarie University, Sydney, New South Wales, Australia. Plagioclase feldspar displays no variation in composition and is albite with XAn [Ca (Ca + Na)] < 0.02; very rare potassium feldspar included within plagioclase has XOr [K (K + Ca + Na)] = 0.95 (Fig. 6a), where K, Ca and Na are the number of potassium, calcium and sodium cations, respectively, per formula unit. S1 chlorite classifies mainly as pycnochlorite with minor brunsvigite (Hey, 1954) with XFe [Fe (Fe + Mg)] = for all samples except 0676 which has slightly lower XFe = , where Fe and Mg are iron and magnesium cations, respectively, per formula unit (Fig. 6b). Chlorite grains within the mafic dyke (sample 0661C) classify as talc-chlorite (Hey, 1954) with XFe = 0.29 (Fig. 6b). The clinopyroxene classification of grains within the mafic dyke (sample 0661C) was determined using the spreadsheet PX-NOM (Sturm, 2002). The clinopyroxene grains classify as aluminian or aluminian ferrian diopside, or aluminian or chromian augite, have XFe = and plot as either diopside or augite (Fig. 6c). Epidote shows appreciable substitution of Al by Fe 3+. S1 epidote is Fe 3+ -rich pistacite with Fe 3+ = cations on the basis of 25 oxygen (Fig. 6d). However, the least Fe 3+ -rich grains are from a coarse-grained S1 foliation (sample 0672) and the most Fe 3+ -rich grains include both samples analysed from the eastern flank of the Dayman dome (samples 0605 and 0628) and finer-grained samples from the northern flank. Prehnite may also show appreciable substitution of Al by Fe 3+. D3 prehnite shows variable Fe 3+ = cations on the basis of 22 oxygen (Table 1). All samples contain S1 calcic amphibole that classifies as actinolite using the scheme of Leake et al. (1997, 2004) with XFe = (Fig. 6e). Rare blue S2(P) sodic-calcic amphibole grains in sample 0605 classify as ferrowinchite and ferrobarroisite with XFe = (Fig. 6f). Sample 0676 contains mostly actinolite with rare sodic-calcic grains that classify as winchite with XFe = (Fig. 6f). Fe 3+ is calculated for all S1 actinolite grains at commonly <0.4 wt% oxide (Fe 2 O 3 ) using the International Mineralogical Association favoured procedure (dependent upon composition) of adjusting the sum Fig. 5. (a) Cross-polarized light photomicrograph of coarse-grained S1 (white line) running diagonally from top right to bottom left, sample 0668A, Gwariu River. S1 is defined by aligned actinolite (act), albite (ab) ribbons and epidote (ep). Field of view is 1.75 mm. (b) Cross-polarized light photomicrograph of a large optically zoned epidote grain, sample 0676, Yutmae Creek. Field of view is 1.75 mm. (c) Plane-polarized light photomicrograph of S2(P) bands of blue sodic-calcic amphibole (outlined by dashed lines) running subparallel to S1 (white line) diagonally from top left to bottom right, sample 0605, Biyawap Creek. Field of view is 1.75 mm. (d) Backscattered electron (BSE) image showing random splays of sodic-calcic blue amphibole (ferrowinchite-ferrobarroisite; outlined by dashed line) that partly pseudomorph S1 minerals, very rare zircon (zrc), very minor S2(P) phengitic muscovite (ms) and titanite (ttn) pseudomorphs of rutile (rt), sample 0605, Biyawap Creek. Scale is shown on the image. (e) Cross-polarized light photomicrograph of S2(B) defined by chlorite (chl), epidote (ep) and albite (ab), running top to bottom (white line), sample 0677A, Ampae Creek. Quartz (qtz) and calcite (cal) veins are oriented sub-parallel to S2(B). Field of view is 1.75 mm. (f) Plane-polarized light photomicrograph of fine-grained gouge in semi-brittle D2 fault zone, sample 0664, Biniguni River. Field of view is 3.5 mm. (g) Plane-polarized light photomicrograph of partially recrystallized dyke with randomly oriented augite (aug) grains. Plagioclase domains are pseudomorphed by albite (ab), chlorite (chl), muscovite (ms) and epidote (ep), sample 0661C, Biniguni River. Field of view is 1.75 mm. (h) Planepolarized light photomicrograph of epidote (ep) and prehnite (prh) veins, sample 0604, Pumani River. Field of view is 0.9 mm.

9 EXHUMATION OF THE DAYMAN DOME, PNG 413

10 414 N. R. DACZKO ET AL. Table 1. Representative chemical analyses of minerals. Sample mineral 0672 act core 0672 act rim 0676 act 0677G act 0605 fbarro 0676 winch 0605 chl 0628 chl 0661C chl 0672 chl 0676 chl 06777H chl 0661C cpx 0672 ep core 0672 ep rim 0677G ep core 0677G ep rim 0677E ab 0677B ttn 0604 prh 0604 prh SiO TiO Al2O Cr 2 O Fe 2 O FeO MnO MgO CaO Na 2 O K2O Total O Si Ti Al Cr Fe Fe Mn Mg Ca Na K Total X An 0.00 XFe a(act) f(act) y(act)

11 EXHUMATION OF THE DAYMAN DOME, PNG 415 (Si + Al + Cr + Ti + Fe + Mg + Mn) to 13 by varying the Fe 3+ and Fe 2+ appropriately in the amphibole recalculation on the basis of 23 oxygen (Leake et al., 1997, 2004). Fe 3+ in S1 actinolite may be subtly zoned from <0.01 wt% oxide (Fe 2 O 3 ) in the core to 0.06 wt% oxide (Fe 2 O 3 ) at the rim (Fig. 6g). Winchite grains in sample 0676 contain wt% oxide (Fe 2 O 3 ). S2(P) ferrowinchite and ferrobarroisite grains in sample 0605 are calculated at up to 4 7 wt% oxide (Fe 2 O 3 ). CALCULATED PHASE DIAGRAMS P T and P X pseudosections were constructed in the model system NCFMASHTO (Na 2 O-CaO-FeO- MgO-Al 2 O 3 -SiO 2 -H 2 O-TiO 2 -Fe 2 O 3 ), using THERMO- CALC (version 3.31; Powell et al., 1998) and the internally consistent thermodynamic data set 5.5 of Holland & Powell (1990; November 2003 update). Using this phase diagram method, we aim to: (i) construct the first metamorphic model for the Goropu Metabasalt, and (ii) constrain the P T conditions that accompanied the development of the S1 assemblage. Minerals (references refer to the mineral model used) included in the construction of the pseudosections are diopside (di; Green et al., 2007), omphacite (o; Green et al., 2007), hornblende (hb; Diener et al., 2007), actinolite (act; Diener et al., 2007), glaucophane (gl; Diener et al., 2007), epidote (ep; Holland & Powell, 1998), chlorite (chl; Holland et al., 1998), plagioclase (pl; albite = ab; Holland & Powell, 2003), ilmenite (lm; White et al., 2000), magnetite (mt; White et al., 2000), hematite (hem; White et al., 2000), quartz (q), titanite sphene (sph), rutile (ru), lawsonite (law) and water (H 2 O). The bulk composition used to model the assemblages in metabasite samples is a typical MORB composition (from Sun & McDonough, 1989) following Diener et al. (2007). Whole rock data indicate the Goropu Metabasalt is MORB composition (Smith & Davies, 1976, pp ). The amount of Fe 2 O 3 (recast as ÔOÕ in the NCFMASHTO system) is likely to have been different during D1 and D2 as indicated by variable Fe 3+ in S1 and S2(P) amphibole (see Mineral chemistry). Therefore, variations in Fe 3+ content are explored in our models. Due to ambiguity in interpreting equilibrium in very low-t rocks and the current absence of suitable models for prehnite and pumpellyite, the low-t fields in the following pseudosections are likely to be metastable with respect to prehnite- and or pumpellyite-bearing assemblages. In the discussion below the low-t petrogenetic grid of Beiersdorfer & Day (1995) is used to argue for the P T conditions of prehniteand or pumpellyite-bearing assemblages at temperatures below the lower limit of the pseudosections (Fig. 7a). A lower temperature limit for the S1 assemblage of 375 C is indicated by a lack of pumpellyite (Fig. 7a; Beiersdorfer & Day, 1995) with an upper temperature limit of 475 C indicated by a lack of hornblende or plagioclase (Diener et al., 2007). We therefore contend that the S1 assemblage formed at 425 ± 50 C. In order to estimate the amount of Fe 2 O 3 in our model bulk rock composition and to test the sensitivity of the results to our choice of Fe 2 O 3, we first constructed an H 2 O-saturated P XFe 3+ pseudosection in NCFMASHTO at 425 C (Fig. 7b). The diagram was contoured for three variables in the composition of actinolite: (i) a(act) = XNa a = number of Na cations in the amphibole A site on the basis of 23 oxygen; (ii) f(act) = XFe 3+ M2; and (iii) y(act) = XAl M2. The variables f(act) and y(act) equal the number of Fe 3+ and Al cations, respectively, in the amphibole M2 site on the basis of 23 oxygen (note that this is half the total Fe 3+ and Al cations, respectively, as there are two M2 sites in amphibole). Our mineral chemistry data indicate that actinolite compositions generally have a(act) > 0.05 (dashed fill pattern, Fig. 7b), f(act) = (vertical line fill pattern, Fig. 7b) and y(act) = (stippled fill pattern, Fig. 7b). The dashed bold line on Fig. 7b highlights the intersection of these three actinolite compositional variables. Therefore, an XFe 3+ = 0.26 is selected from the centre of this field as the choice for Fe 2 O 3 (wt% oxide) in the model bulk rock composition (Fe 2 O 3 = ÔOÕ = 0.86; Fig. 7c). Note that the XFe 3+ value modelled by Diener et al. (2007) was at XFe 3+ = 0.12 (labelled D07 on Fig. 7b; Fe 2 O 3 = ÔOÕ = 0.50), indicating that our samples evolved at higher oxidation state compared with average MORB (Diener et al., 2007). An H 2 O-saturated P T pseudosection in NCFMASHTO is shown in Fig. 7c. Glaucophaneabsent assemblages are stable below 8 kbar at 425 C and hornblende-plagioclase-absent assemblages occur below C (Fig. 7c). The quadrivariant field involving actinolite, chlorite, epidote, albite, titanite, quartz and H 2 O (highlighted by a bold line; Fig. 7a) best defines the peak conditions for the S1 assemblage. Actinolite compositional contours for y(act) are shown in Fig. 7c and suggest peak pressures are likely to be towards the higher-p range of the peak assemblage field at kbar at 425 C. The S2(P) assemblage involves sodic-calcic blue amphibole, chlorite, epidote, albite, titanite, quartz and H 2 O, though the P T conditions that accompanied the development of this assemblage should not be interpreted from the constructed pseudosections. The composition of the sodic-calcic blue amphibole suggests the development of the S2(P) textures occurred at a higher oxidation state than that which accompanied the development of S1. The composition of our sodiccalcic blue amphibole has very high Fe 3+ (Table 1), yet does not match the glaucophane composition anywhere in the P XFe 3+ model. The S2(P) amphibole best fits an actinolite-bearing assemblage

12 416 N. R. DACZKO ET AL.

13 EXHUMATION OF THE DAYMAN DOME, PNG 417 (Fig. 7b), suggesting that S2(P) evolved at < 4.5 kbar at high XFe 3+ and magnetite stable conditions (Fig. 7b). However, magnetite was not observed in the samples, the temperature for the S2(P) assemblage is not well constrained and the temperature was perhaps well below that of the S1 assemblage. Therefore, the pressure estimates for the development of S2(P) determined from Fig. 7b must be used with caution. D3 epidote-prehnite-bearing veins at site 0604 are stable at <3 kbar and below C using the low-t petrogenetic grid of Beiersdorfer & Day (1995; Fig. 7a). These results combine to give a P T D path for the Dayman dome that evolved from kbar at 425 C (S1) to <4.5 kbar (S2(P)) to <3 kbar below C (D3). DISCUSSION Pre-S1 conditions Davies (1980) defined metamorphic zones across the Suckling-Dayman massif in a reconnaissance study, including large areas of pumpellyite-actinolite facies metabasite with local occurrences of glaucophane, lawsonite and or aragonite (G, L and A, respectively, on Fig. 2). These rocks commonly preserve the primary texture of massive basalt (Davies, 1980). Greenschist facies rocks are restricted to S1 in a 500 m thick low-angle shear zone that conforms to the topography of the Dayman dome. The data presented here confirm the interpretation of greenschist facies conditions for the development of S1 on the northern and eastern flanks of the Dayman dome and provides the first quantitative estimates of the P T conditions. S1 cuts the footwall pumpellyite-actinolite facies metabasite rocks with minor pods of prehnitepumpellyite facies rocks (Davies, 1980; Fig. 2). Although these units were not sampled in this study, pumpellyite-actinolite assemblages are stable at <9 kbar, whereas glaucophane-lawsonite assemblages are stable above 6.5 kbar and a bold black line on Fig. 7a highlights the overlapping fields of pumpellyite-lawsonite-glaucophane stability (Beiersdorfer & Day, 1995). These assemblages suggest that the Suckling-Dayman massif has been exhumed from a depth of km. However, caution should be exercised in this interpretation, as no mineral chemistry data have been presented for the pumpellyiteactinolite facies core and the Fe 3+ content of the ÔglaucophaneÕ was never documented. The reported ÔglaucophaneÕ may therefore be similar to the blue S2(P) sodic-calcic amphibole documented here. On the other hand, observations of aragonite mainly in pumpellyite-actinolite facies metabasite rocks (Davies, 1980) and calcite in slightly higher-t greenschist facies rocks of this study support placing the start of the P T path for the Dayman dome at the light grey filled star shown on Fig. 7a. P T conditions for S1-D3 Important phases in rocks of mafic compositions have not been calibrated for very low-t conditions in the NCFMASHTO chemical system. In addition, the amphibole model of Diener et al. (2007) does not incorporate K or Ti. However, these are minor components at these P T conditions and K was not considered, as it is a very minor component of MORB. The advent of prehnite and pumpellyite models for basic rocks will allow for the calculation of more robust lower-t limits for these rocks. Notwithstanding these limitations, we consider that the NCFMASHTO pseudosections presented provide valid and constructive first constraints of the P T conditions of the S1 and S2(P) assemblages in metabasite of the Dayman dome. Based on the pseudosection models, we suggest that S1 formed at kbar at 425 C (Fig. 7c). The P XFe 3+ model shows that our pressure estimate is largely based on the amount of Al in actinolite [y(act), Fig. 7b]. This result is essentially independent of the choice of XFe 3+, as the y(act) compositional isopleths are nearly horizontal on the P XFe 3+ section (Fig. 7b). We therefore consider the P estimate for S1 to be robust. The Al content of S1 actinolite documented here is therefore consistent with minor exhumation from peak pressures of the metamorphic core during the development of S1 (see the first part of the P T path shown on Fig. 7a). However, our estimate of pressure would be lower at temperatures of C, as the y(act) isopleths curve down-p at higher-t approaching hornblendeplagioclase stability (Fig. 7c). We consider 425 C as suitable for the S1 assemblage as neither hornblende nor plagioclase are observed in our samples and XAn < 0.02 for all plagioclase grains analysed (see Mineral chemistry). Fig. 6. Mineral chemistry classification plots for (a) plagioclase [An = anorthite, Ab = albite, Or = orthoclase]; (b) chlorite [XFe = Fe (Fe + Mg), Si = number of silicon cations recalculated for 28 oxygen]; (c) clinopyroxene; (d) epidote [Fe 3+ and Al are number of ferric iron and aluminium cations, respectively, recalculated for 25 oxygen, Ps, pistacite (2 Fe 3+ ); Cz, clinozoisite (0 Fe 3+ )]; (e) calcic and (f) sodic-calcic amphibole. [Mg, Fe 2+ and Si are number of magnesium, ferrous iron and silica cations, respectively, recalculated for 23 oxygen]; and (g) mineral chemical zoning in S1 actinolite (sample 0672). The three variables in the composition of actinolite shown are: (i) a(act) = XNa a = number of Na cations in the amphibole A site on the basis of 23 oxygen; (ii) f(act) = XFe 3+ M2; and (iii) y(act) = XAl M2. The variables f(act) and y(act) equal the number of Fe 3+ and Al cations, respectively, in the amphibole M2 site on the basis of 23 oxygen (note that this is half the total Fe 3+ and Al cations, respectively, as there are two M2 sites in amphibole).

14 418 N. R. DACZKO ET AL. The mafic dykes cut S1, lack chilled margins and are partially recrystallized. These observations are consistent with intrusion of the dykes during or shortly following the cessation of D1 shear, but prior to significant cooling of the greenschist facies rocks. This time frame is consistent with the orientation of the dykes reflecting a similar N to NE extension direction as for the L1 mineral elongation lineation (Fig. 3). An elevated oxidation state during the development of S2(P) is indicated by high Fe 3+ values in S2(P)

15 EXHUMATION OF THE DAYMAN DOME, PNG 419 amphibole and epidote (see Mineral chemistry). A lack of sodic amphibole (gl in the model, Fig. 7b) limits the development of S2(P) to <4.5 kbar at high oxidation state (Fig. 7b). The S2(P) temperature is likely to have been less than that which accompanied the development of S1 as the rocks cooled during exhumation and were probably in the range C. The prehnite-epidote assemblage identified within the D3 semi-brittle to brittle fault zone at site 0604 is consistent with exhumation of the Goropu Metabasalt to <3 kbar (10 km) prior to prehnite neocrystallization. Comparison with low- and high-grade metamorphism in the immediate region Prehnite-pumpellyite facies rocks include a large area south of the Onuam Fault (Fig. 2) and rare pods within the pumpellyite-actinolite facies metabasite rocks mapped by Davies (1980). Coexisting prehnite and pumpellyite suggest these metabasites likely recrystallized at 3 kbar and 250 C (Fig. 7a). This may represent a maximum burial depth of 10 km for the large area south of the Onuam Fault as these rocks grade into unmetamorphosed basalt of the Kutu Volcanics (Davies, 1980; Fig. 2). The rare pods of prehnite-pumpellyite facies rocks within the pumpellyite-actinolite facies metabasite rocks north of the Onuam Fault (Fig. 2) may represent low strain zones that did not recrystallize at the higher-p pumpellyite-actinolite facies metamorphic conditions experienced by most of the Goropu Metabasalt. Davies (1980) also mapped a contact aureole hornfels facies in the Goropu Metabasalt around felsic plutonic rocks near the summit of Mt Suckling (Fig. 2). Metamorphic mineral assemblages reported by Davies (1980) include garnet-mica schists with andalusite and staurolite. These assemblages suggest a depth of recrystallization within the stability field of andalusite (i.e. <4 kbar; depth <12 km). This observation suggests that the felsic pluton intruded following significant uplift of the pumpellyite-actinolite facies metabasite rocks. Metamorphic core complexes in the DÕEntrecasteaux Islands directly east of the Dayman dome (Fig. 1) have exhumed metamorphic rocks that vary in grade. These include peak and retrograde assemblages that lie in the blueschist, greenschist, amphibolite, granulite and eclogite facies (Davies & Warren, 1988, 1992; Hill & Baldwin, 1993; Baldwin et al., 2004, 2008; Little et al., 2007; Monteleone et al., 2007). Research on the islands has largely focused on the eclogite facies rocks in the core of the complexes. Peak conditions for the eclogite facies rocks are in the range of C and kbar (Davies & Warren, 1992), C and minimum 21 kbar (Hill & Baldwin, 1993), C and kbar (Baldwin et al., 2004), C and minimum kbar (Baldwin et al., 2005; Monteleone et al., 2007), and C and kbar (Baldwin et al., 2008). Eclogite facies metamorphism has been dated using U Pb zircon from one sample from Ferguson Island at 4.3 ± 0.4 Ma (Baldwin et al., 2004) and from five variably retrogressed samples from Ferguson (7.9 ± 1.9, 7.0 ± 1.0 Ma) and Goodenough (2.94 ± 0.41, 2.82 ± 0.27, 2.09 ± 0.49 Ma) islands (Monteleone et al., 2007; Baldwin et al., 2008). Mineral assemblages associated with retrogression of high-grade rocks in core complex-bounding shear zones on Fergusson and Goodenough islands suggest these shear zones evolved at C and 7 11 kbar (Hill & Baldwin, 1993). Therefore, the core complexes on Fergusson and Goodenough islands experienced much higher-p metamorphic conditions compared with the pumpellyite-actinolite facies core of the Suckling-Dayman massif. The core complex-bounding shear zones on these two islands also record higher-p T metamorphism compared with the moderate-p greenschist facies mylonite rocks of this study. Normanby Island has been less studied. Western Normanby Island contains quartzofeldspathic gneiss of amphibolite or relict eclogite facies, whereas epidote-blueschist facies rocks occur east of the Trobriand fault (Little et al., 2007). Greenschist facies assemblages variably overprint the epidote-blueschist facies rocks (Little et al., 2007). Schists within the core, east of the Trobriand fault, contain garnet-ferroglacophane- Fig. 7. (a) Schematic P T diagram for low-temperature metabasites. Facies abbreviations: Z, zeolite; Prh Act, prehnite-actinolite; Pmp Act, pumpellyite-actinolite; GS, greenschist; BS, blueschist; EBS, epidote blueschist. From Beiersdorfer & Day (1995). The filled regions represent the stability of prehnite, pumpellyite, glaucophane and lawsonite. The bold black lines highlight the P T region where pumpellyite, glaucophane and lawsonite are all stable. The argonite to calcite transition is shown (white dashed line). The stars represent our best estimate of the P T conditions for the metamorphic core (grey filled star) and the D1-D3 events (white filled stars). Arrows between events represent the P T path followed during exhumation of the metamorphic core. (b) An H 2 O-saturated P XFe 3+ pseudosection calculated for the Goropu Metabasalt MORB bulk composition at 425 C. The inferred peak S1 assemblage field is outlined in bold with compositional domains highlighted for actinolite (see text for discussion and Fig. 6 caption for definition of actinolite compositional variables). The short dashed black line outlines the field of overlapping actinolite compositional variables. The long dashed black line outlines the maximum pressure for S2(P). D07 = XFe 3+ used by Diener et al. (2007). (c) An H 2 O-saturated P T pseudosection calculated at XFe 3+ = 0.26 for the Goropu Metabasalt MORB bulk composition. The inferred peak S1 assemblage field is outlined in bold. Dashed lines represent the pressure-sensitive y(act) compositional variable with the common range of values in S1 actinolite highlighted by a stippled pattern. The star represents our best estimate of the conditions of metamorphism that accompanies the development of S1. The dark grey field at low-t and reactions 4 and 5 are the pumpellyite-actinolite facies field from (a) which provide a lower limit for temperature.