oligocene miocene tectonic evolution of the south fiji basin ......oligocene–miocene tectonic...

(2007) 1–24www.elsevier.com/locate/margeo

Marine Geology 237

Oligocene–Miocene tectonic evolution of the South Fiji Basinand Northland Plateau, SW Pacific Ocean: Evidence from

petrology and dating of dredged rocks

N. Mortimer a,⁎, R.H. Herzer b, P.B. Gans c,C. Laporte-Magoni d, A.T. Calvert c,1, D. Bosch e

a GNS Science, Private Bag 1930, Dunedin, New Zealandb GNS Science, PO Box 30368, Lower Hutt, New Zealand

c Department of Geological Sciences, University of California, Santa Barbara, CA 93106, USAd LGPMC, EA 3325, Université de Nouvelle Caledonie, BP R4, 98851 Noumea, Nouvelle Calédonie, France

e Lab. de Tectonophysique, UMR-CNRS 5568, Univ. Montpellier II, cc 56, Montpellier 34095, France

Received 16 November 2005; received in revised form 5 September 2006; accepted 23 October 2006

Abstract

We present new analytical data from lavas and associated rocks dredged and/or drilled from the South Fiji Basin, NorthlandPlateau, Colville Ridge and Havre Trough. These results provide much-needed ground truth about the geology, age and tectonicevolution of the Cenozoic submarine ridges and basins between the active intraoceanic Tonga–Kermadec arc, and rifted continentalborderlands of New Zealand, the Norfolk Ridge and New Caledonia. Key results from this study include: (1) Ar–Ar dates onMinerva Abyssal Plain oceanic crust suggest that the ages of magnetic anomalies in the South Fiji Basin have been overestimatedby earlier workers; (2) subduction-related lavas are widespread across the region, are not presently organised into arc-like chains,and cluster in the age range 22–18 Ma (Early Miocene); (3) the oldest subduction-related lavas occur in the western part of theregion (32–26 Ma: Norfolk and Three Kings Ridge); (4) shoshonites, interpreted as rifted arc lavas, were erupted in a narrow 20–21 Ma interval over a wide area. Put together, these results indicate high magmatic flux and large and rapid horizontal tectonictranslations and basin opening from 18–23 Ma in the region immediately north of New Zealand. We explain the Miocenetectonomagmatic development of the region by a model of rapid rollback of a single, east-facing Pacific arc–trench system thatbecame established after Northland Allochthon emplacement. Critical testing of this, versus other, tectonic models must awaitdrilling and dating of thus-far unsampled Kupe Abyssal Plain crust.© 2006 Elsevier B.V. All rights reserved.

Keywords: petrology; geochemistry; geochronology; subduction; back-arc basins; shoshonite; South Fiji Basin; New Zealand

⁎ Corresponding author. Tel.: +64 3 479 9686; fax: +64 3 477 5232.E-mail address: [email protected] (N. Mortimer).

1 Present address: U.S. Geological Survey, 345 Middlefield Road,Menlo Park, CA 94025, USA.

0025-3227/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.margeo.2006.10.033

1. Introduction

The South Fiji Basin area (Fig. 1) is one of the leastinvestigated parts of the southwest Pacific Ocean. It is aregion of remnant volcanic arcs, plateaus and basins thatlie between the presently active Taupo–Kermadec–Tonga arc, and the edge of Gondwanaland continental

mailto:[email protected]

http://dx.doi.org/10.1016/j.margeo.2006.10.033

Fig. 1. Regional setting of the study area in the SWPacific Ocean, showing geographic names and dredge locations fromChurkin and Packham (1973),Packham and Terrill (1976), Davey (1982), Adams et al. (1994), Mortimer et al. (1998), Shipboard Party (1999), Ballance et al. (1999), Bernardel et al.(2002), Mortimer et al. (2003), Herzer et al. (2004a) and GNS Science (unpublished data). Bathymetry from CANZ (1997), basins are darker grey,ridges and plateaux lighter grey. Magnetic anomaly picks from Malahoff et al. (1982) and Davey (1982). JLin = Julia Lineament, VDLL = Van derLinden Lineament. Inset shows different geographic areas of the Northland Plateau mentioned in the text and appendices. Small text labels refer todredge numbers (new samples) and P numbers (old samples from Mortimer et al., 1998).

2 N. Mortimer et al. / Marine Geology 237 (2007) 1–24

3N. Mortimer et al. / Marine Geology 237 (2007) 1–24

crust which extends from New Zealand along theNorfolk Ridge to New Caledonia.

There is general agreement that from c. 85–55 Ma,the Pacific-Australian plate boundary was the TasmanSea spreading ridge, and that New Zealand, the LordHowe Rise and Norfolk Ridge formed a rifted continen-tal borderland – Zealandia – on the Pacific Plate. At c.45 Ma, convergence began between the Pacific andAustralian plates (Sutherland, 1999), and since at least5 Ma the convergent Pacific-Australian plate boundaryhas been close to its present position which is theHikurangi–Kermadec–Tonga Trench. However, there ismajor disagreement on the ages of the Norfolk and SouthFiji Basins and also the age, polarity and numbers ofvolcanic arcs and subduction zones that operated in theinterval 85–5 Ma (e.g. Malpas et al., 1992; Herzer andMortimer, 1997; Mortimer et al., 1998; Ballance, 1999;Crawford et al., 2003; Sdrolias et al., 2004; Schellartet al., 2006). In addition to these uncertainties in tectonicmodels, the nature and role of a newly defined tectonicelement in the southwest Pacific, the 45000 km2

Northland Plateau (Fig. 1, Herzer et al., 2000) isessentially unknown. Is the Northland Plateau a pieceof trapped Cretaceous Hikurangi Plateau (Mortimer andParkinson, 1996), foundered New Zealand continentalcrust, or a Cenozoic volcanic pile? It is important toaddress these issues in order to understand how typical oratypical is the New Zealand–NewCaledonia–Fiji area interms of progressive arc and backarc basin developmentand the applicability of standard trench rollback models.A knowledge of the timing, sequence and kinematics ofSW Pacific backarc opening and volcanism is importantfor petroleum prospectivity of the Reinga, Northland andEast Coast continental shelves, as such events providelong range driving forces for basin evolution.

In order to help test the various tectonic models wemade two cruises on the R/V Tangaroa to the NorthlandPlateau in 1999 and 2002, ONSIDE (Offshore North-land SeIsmic and Dredging Expedition) I and II, withthe specific aim of recovering, analysing and datingrocks in conjunction with reflection seismic profiling(Shipboard Party, 1999; Herzer et al., 2004a). We alsoobtained material from contract dredging of seamountsin the South Fiji Basin and Colville Ridge, andreanalysed and dated some material from our earlierwork in the Norfolk Basin (Mortimer et al., 1998) andfrom Deep Sea Drilling Project holes in the South FijiBasin (Churkin and Packham, 1973; Stoeser, 1976)(Fig. 1).

The purpose of this paper is to present a summary ofour new analytical results, previously published only asabstracts (Herzer and Mortimer, 1997; Herzer et al.,

2001; Bosch et al., 2002; Mortimer et al., 2002; Herzeret al., 2004b, 2005). The data comprise 81 X-rayfluorescence analyses, 41 ICP-MS element analyses and26 Ar–Ar dates of lavas from 36 different dredge sites.The results greatly expand our knowledge of this part ofthe SW Pacific seafloor and enable a more completeview of the Cenozoic volcanotectonic evolution of thelarge triangular area between New Zealand, NewCaledonia and Fiji. In particular they challenge existingtectonic models which are largely based on longstandingmagnetic anomaly interpretations (e.g. Malahoff et al.,1982; Davey, 1982). We present an integrated, data-constrained petro-tectonic model that explains theMiocene tectonic evolution of the area.

2. Regional setting

2.1. South Fiji Basin

The South Fiji Basin (Packham and Terrill, 1976) is agenerally smooth-floored, 3–4 km deep basin with up to500 m of sediment cover (Fig. 1). It is bounded to thewest by the Three Kings Ridge, Loyalty Ridge and CookFracture Zone, to the east by the Colville–Lau Ridge andto the south by the Northland Plateau. Reflection seismicprofiles across many of these margins (work in progress)show both rifted volcanics and volcaniclastic apronsmerging into, or slightly above, the top of the oceaniccrust of the deep basin (e.g. Herzer et al., 2000, 2004b).There is no evidence of a fossil trench or accretionaryprism on any side of the South Fiji Basin. The basin isdivided into two parts, a southern Kupe Abyssal Plainand a northern Minerva Abyssal Plain, separated by apoorly surveyed and defined Central Ridge regionbetween the Lau and Fantail Terraces (Fig. 1).

Several groups of seamounts (Sarah, Devonport,Margot, Marion, Coquille), some elongated, some flattopped, rise from the western part of the southern SouthFiji Basin, which also has a different satellite gravitytexture from the adjacent flatter, and slightly deeperKupe Abyssal Plain. Two elongated flat-topped sea-mounts, Matahourua and Mascarin, lie on the easternside of Kupe Plain near the Colville Ridge (Fig. 1).

The age and kinematics of the opening of the SouthFiji Basin are controversial. Oligocene magnetic anom-aly (7–12) patterns outlining a triple junction have longbeen interpreted for the Minerva Plain and westward-younging 7–12 anomalies have long been interpreted inthe Kupe Plain (Weissel and Watts, 1975; Davey, 1982;Malahoff et al., 1982). However, Sdrolias et al. (2003)reinterpreted the Kupe 7–12 anomalies as younging east,not west, and connected them to a Minerva triple

Table 1Selected whole rock analyses of lavas

GNS no. Location SiO2 TiO2 Al2O3 Fe2O3T MnO MgO CaO Na2O K2O P2O5 LOI Total As Ba

Minerva Abyssal PlainP63848w DSDP205 48.20 1.35 16.10 10.40 0.17 7.33 11.50 2.56 0.60 0.19 1.61 100.01 b1 20P63850w DSDP285 50.10 2.02 13.80 13.10 0.21 6.54 10.40 3.13 0.12 0.19 0.36 99.97 b1 11P63834w Julia 50.71 1.47 15.62 8.36 0.16 7.23 12.63 2.84 0.17 0.15 0.00 99.33 b1 20P63836w Alison 46.88 2.00 17.76 10.18 0.16 4.73 11.73 3.65 0.33 0.31 1.20 98.94 10 9

South Fiji Basin SeamountsPotassic suiteP59786w Devonpt C 55.45 0.93 15.48 6.24 0.08 5.70 6.25 2.16 5.88 0.53 0.48 99.18 3 1067P59788g Devonpt C 51.81 0.99 17.27 6.85 0.11 2.84 6.88 2.53 6.32 0.88 2.62 99.10 36 1144P61717g Devonpt E 59.41 0.53 18.43 3.87 0.06 1.33 2.60 2.27 10.30 0.34 0.72 99.90 7 244P61724g Margot 56.43 0.60 18.07 5.54 0.07 3.31 5.97 3.08 4.99 0.43 0.91 99.40 9 1382P59769g Sarah W 53.19 0.71 15.71 8.47 0.15 5.57 9.78 2.36 2.25 0.30 1.06 99.55 5 1366P63821w Sarah C 43.32 0.76 17.29 8.79 0.07 2.59 9.76 2.80 4.55 4.21 5.08 99.22 96 3211P63825w Sarah N 51.95 0.61 15.94 9.03 0.14 6.69 10.72 1.91 1.83 0.28 0.50 99.60 5 1532Ocean island basaltsP63830w Marion 50.82 2.38 18.29 9.22 0.23 2.15 7.73 4.45 2.63 1.01 0.93 99.83 9 277P63835w Coquille 47.83 2.00 18.44 9.16 0.17 5.29 7.05 4.61 2.32 1.10 1.66 99.62 2 423P67647w Mascarin 38.88 1.52 16.88 9.73 0.16 2.01 15.00 3.36 1.61 5.77 4.92 99.83 52 125

Northland PlateauPoor Knights Seamount ChainP61712g P70L 53.08 1.10 17.32 9.21 0.26 0.93 4.49 4.04 5.34 1.25 2.48 99.50 54 556P66794w D14B-3 50.61 1.12 20.33 8.65 0.16 2.47 5.87 4.23 1.65 0.20 4.68 99.97 9 90P66824w D21-3 56.10 1.27 14.56 12.10 0.13 3.79 4.26 4.03 1.85 0.20 1.69 99.98 4 266P66808w D18A-1 44.88 1.31 17.16 9.54 0.11 3.39 12.17 2.49 2.77 2.82 3.14 99.78 44 402P66825w D21-6A 47.30 0.73 23.69 8.70 0.08 3.22 12.02 2.33 0.61 0.12 1.18 99.98 3 49

Outer Volcanic PlateauP63154m D4B-4 47.90 1.43 20.89 11.04 0.43 2.02 8.95 3.50 1.20 0.14 2.48 99.98 23 178P63162m D5B-1,2 45.85 1.96 20.28 11.74 0.18 2.78 8.38 3.93 1.11 0.61 3.13 99.95 68 159P63167m D6 hi K 54.60 1.16 20.13 7.15 0.09 1.34 6.13 5.30 2.07 0.48 1.24 99.69 13 187P63165m D6 lo K 50.57 1.32 18.27 8.75 0.22 3.31 9.05 3.41 0.20 0.26 4.42 99.78 1 58P63179m D7B-32 46.25 1.64 19.74 11.54 0.13 4.33 9.02 3.24 1.00 0.23 2.87 99.99 32 75P66800w D20-1 44.67 1.68 18.64 13.50 0.12 3.90 10.76 3.21 0.79 0.22 2.48 99.97 40 70P66803w D19-1 48.17 0.93 18.25 9.49 0.12 6.31 9.27 2.56 1.99 0.30 2.40 99.85 2 700P66807w D19-9 55.33 0.69 20.34 6.07 0.12 1.69 3.87 4.44 4.24 0.26 2.49 99.58 6 1789

Havre TroughP63474m SO135-36 50.47 1.54 17.12 10.17 0.17 5.98 10.16 3.09 0.56 0.26 0.49 100.01 2 98

Colville RidgeP63206 OR99-1 57.82 0.65 22.86 10.86 bd 0.07 0.11 0.00 0.14 0.13 6.84 99.48 3 56

Continental shelfP66851w D24A-3 67.12 0.37 15.44 2.76 0.04 0.52 1.97 4.60 3.84 0.11 3.19 99.96 12 676P66854w D24B-1ii 60.16 0.75 18.96 5.03 0.02 0.48 3.38 5.36 4.42 0.29 0.97 99.82 29 643

Ce Cr Cs Cu Dy Er Eu Ga Gd Hf Ho La Lu Nb Nd NiSee text for analytical methods. ICP-MS analyses done at wWashington State University (analyst C. Knaack), gGrenoble University (analyst C.Laporte-Magoni), mMontpellier University (analyst D. Bosch).


junction via a spreading ridge along the Julia Lineament.In yet another model, Lawver et al. (2002) interpreted theJulia Lineament as a major transform and incorporatedthe suggestion of Herzer and Mortimer (1997) that the

Kupe magnetic anomalies could be reinterpreted asMiocene (e.g. 5AA-5E; c. 19–13 Ma). Our dating ofMinerva Plain tholeiites (Section 6.1) further questionsthe original Davey–Malahoff interpretations.

Table 1Selected whole rock analyses of lavas

Ce Cr Cs Cu Dy Er Eu Ga Gd Hf Ho La Lu Nb Nd Ni

12.1 203 0.51 72 5.23 2.95 1.21 14 4.31 2.28 1.09 4.64 0.41 3.24 9.65 8014.6 47 0.19 63 8.49 5.00 1.79 18 6.63 3.52 1.77 5.22 0.69 4.76 13.0 468.11 252 0.02 79 4.42 2.47 1.03 16 3.62 1.91 0.94 3.14 0.33 4.50 7.27 106

20.1 354 0.46 71 6.95 3.81 1.93 17 6.11 4.05 1.41 6.48 0.52 1.39 16.3 178

94.6 261 3.21 49 4.58 2.03 2.11 19 6.69 9.78 0.83 47.4 0.25 16.6 41.5 106105 182 3.44 37 4.19 2.00 2.77 19 7.57 9.74 0.78 48.7 0.24 20.6 46.7 75102.3 14 1.19 10 3.09 1.72 1.40 18 4.51 18.2 0.62 46.7 0.24 35.9 41.3 19111 49 6.52 28 4.28 2.19 2.91 18 7.17 7.61 0.82 55.1 0.30 17.6 45.4 3259.7 249 7.16 83 3.46 1.98 2.44 16 5.03 2.49 0.71 29.7 0.28 5.04 25.7 73163 118 0.3 63 7.45 4.71 2.97 11 9.32 6.01 1.61 100.7 0.72 12.7 60.9 9071.3 117 4.84 125 3.71 1.78 1.75 14 5.29 2.09 0.69 38.4 0.25 3.59 31.2 45

79.1 13 0.37 40 5.14 2.33 2.26 23 6.33 4.41 0.98 43.7 0.28 81.5 35.8 35107 99 0.5 39 5.81 2.96 2.47 19 6.54 6.96 1.13 64.4 0.43 93.3 40.0 9328.6 296 0.27 91 8.88 6.41 2.06 14 7.77 3.01 2.14 40.8 1.06 15.8 26.4 70

41.6 5 2.19 56 7.34 4.92 1.86 23 7.23 5.54 1.66 21.3 0.70 7.36 25.4 6718.4 5 0.13 25 4.12 2.18 1.33 18 3.50 4.91 0.81 5.14 0.30 2.07 8.80 4023.4 b1 2.24 111 5.85 3.42 1.37 20 5.19 3.32 1.26 10.2 0.46 4.49 14.2 1657.1 282 0.41 64 5.83 3.48 1.77 13 5.93 5.53 1.23 35.7 0.49 14.3 29.5 585.78 14 2.41 92 2.49 1.45 0.62 20 2.03 0.83 0.54 2.51 0.21 0.47 4.17 11

18.7 109 0.24 170 3.02 1.62 0.98 17 3.10 3.92 0.61 9.59 0.24 6.59 11.5 6227.9 419 0.16 118 8.02 4.99 2.02 18 7.88 4.23 1.74 30.1 0.71 15.5 28.5 10623.2 3 1.67 76 8.13 4.85 1.95 23 7.87 5.06 1.72 15.7 0.70 3.95 25.4 2016.8 17 0.12 98 4.38 2.70 1.19 18 4.04 2.80 0.93 6.88 0.43 2.03 12.0 2513.2 238 0.37 83 6.23 3.35 1.64 15 6.38 3.01 1.24 21.8 0.41 1.52 21.8 9813.1 538 0.11 118 7.19 3.72 2.10 18 7.45 2.77 1.45 22.9 0.48 2.44 23.7 9038.7 13 1.67 104 4.1 2.08 1.52 14 4.91 2.44 0.80 18.6 0.27 3.97 20.3 49103 0 0.74 48 3.7 1.66 2.23 15 5.12 10.7 0.66 35.5 0.21 17.5 29.8 43

16.6 161 0.21 60 4.68 2.75 1.40 18 4.47 2.57 0.98 6.14 0.40 2.85 12.9 71

20 10 – 5 – – – 41 – – – 10 – 1 – 7

64.7 b1 5.27 7 7.14 3.94 1.10 20 6.82 9.01 1.43 35.5 0.57 10.9 31.7 458.1 13 1.58 15 4.9 2.50 1.25 23 5.35 7.09 0.96 28.6 0.33 13.0 25.8 30

(continued on next page)


2.2. Northland Plateau

The extent and shape of the Northland Plateau (Fig. 1;Davey, 1982) has become much better known since the

bathymetric compilation of CANZ (1997). It lies betweenthe base of the clearly defined Northland continental slope(c. 1500mwater depth) and the KupeAbyssal Plain of theSouth Fiji Basin (c. 3000 m). The Northland Plateau

Table 1 (continued )

GNS no. Pb Pr Rb Sc Sm Sr Ta Tb Th Tm U V Y Yb Zn Zr

Minerva Abyssal PlainP63848w 1.54 1.86 7.3 44 3.42 165 0.39 0.78 0.30 0.42 0.13 256 28.2 2.64 79 82P63850w 1.50 2.34 1.4 46 5.02 112 0.80 1.26 0.32 0.71 0.12 326 47.2 4.45 87 120P63834w 1.03 1.24 1.9 50 2.72 148 0.42 0.69 0.29 0.35 0.74 262 32.6 2.13 96 76P63836w 1.81 3.17 6.6 38 5.26 357 0.44 1.09 0.09 0.54 0.20 241 36.6 3.36 85 175

South Fiji Basin SeamountsPotassic suiteP59786w 22.3 10.6 138 24 8.67 715 1.58 0.89 16.1 0.27 3.71 160 22.6 1.59 115 366P59788g 20.5 12.5 126 9 8.59 756 1.78 0.95 15.6 na 3.08 193 24.2 1.72 91 429P61717g 19.0 11.9 283 7 5.66 449 2.50 0.67 8.15 na 2.11 49 18.5 1.70 46 734P61724g 46.7 12.6 110 15 8.41 918 1.23 0.95 29.3 na 6.76 119 25.3 2.00 53 326P59769g 25.9 6.80 70.9 12 5.2 798 0.70 0.67 14.9 na 3.47 247 20.5 1.84 87 100P63821w 75.7 16.6 33.6 28 11.8 1453 1.33 1.30 71.7 0.68 13.4 184 74.9 4.21 120 185P63825w 29.2 7.85 52.0 39 6.85 967 0.61 0.72 16.1 0.26 3.17 228 18.5 1.58 71 74Ocean island basaltsP63830w 1.93 8.66 40.3 20 7.14 720 2.63 0.97 3.86 0.32 0.99 176 48.4 1.80 94 367P63835w 4.63 10.9 38.2 16 7.73 1535 6.14 1.00 9.17 0.43 2.52 126 30.7 2.71 75 335P67647w 2.68 5.80 19.3 35 6.24 570 1.09 1.29 1.64 0.96 2.52 245 93.5 6.17 179 121

Northland PlateauPoor Knights Seamount ChainP61712g 17.9 5.94 105 19 6.27 193 0.98 1.17 7.72 na 4.04 188 56.7 4.59 109 221P66794w 17.8 1.78 16.5 31 3.17 197 0.16 0.65 3.42 0.32 0.29 101 17.6 2.00 155 137P66824w 6.60 2.96 46.5 35 4.53 189 0.32 0.92 3.58 0.49 1.21 263 31.9 2.97 122 100P66808w 6.44 7.27 47.4 34 6.06 493 1.00 0.91 6.48 0.50 2.38 259 43.6 3.05 128 190P66825w 7.09 0.78 18.9 44 1.48 182 0.03 0.37 0.61 0.21 0.29 389 14.7 1.30 187 23

Outer Volcanic PlateauP63154m 14.0 2.47 13.0 25 2.68 279 0.65 0.47 2.13 0.24 0.62 246 15.4 1.54 194 168P63162m 11.4 6.15 13.0 35 6.08 271 0.90 1.19 1.46 0.70 1.73 291 65.1 4.40 155 191P63167m 2.97 5.10 34.7 22 6.19 263 0.41 1.21 1.16 0.69 0.27 134 58.5 4.37 111 201P63165m 3.99 2.43 1.7 30 3.17 250 0.39 0.65 1.09 0.40 0.42 304 27.6 2.59 561 108P63179m 2.16 4.54 15.6 40 4.92 350 0.39 0.95 0.17 0.45 0.67 197 36.2 2.72 175 118P66800w 7.65 5.02 7.3 42 6.51 325 0.15 1.18 0.44 0.52 0.65 315 38.7 3.11 183 122P66803w 47.3 4.63 51.9 31 5.18 456 0.27 0.72 6.19 0.29 1.41 226 20.8 1.77 144 85P66807w 44.7 7.72 60.5 11 6.35 531 1.66 0.69 51.2 0.23 0.72 73 17.0 1.41 85 276

Havre TroughP63474m 1.91 2.49 8.3 35 3.51 256 0.84 0.70 0.62 0.39 0.15 295 26.2 2.55 72 92

Colville RidgeP63206 10 – 6 45 – 38 – – b5 – 4 207 20 – 22 87

Continental shelfP66851w 17.8 8.00 98.4 11 7.47 171 0.95 1.16 15.6 0.58 2.50 17 38.4 3.71 64 334P66854w 16.7 6.43 116 8 5.84 265 0.97 0.84 12.7 0.35 2.30 81 25.0 2.18 56 259


varies in width from 100–200 km and was divided byHerzer et al. (2000) into the Inner Sedimentary Basins(Whangaroa and Knights Basins) the Central SeamountChain (Poor Knights Seamount Chain) and OuterVolcanic Plateau. A prominent magnetic and gravitylineament (Van der Linden “Fault” of Sutherland, 1999,

herein referred to as the Van der Linden Lineament orVDLL) marks a fundamental change in the crust betweenthe Inner Sedimentary Basins and the combined PoorKnights Seamount Chain–Outer Volcanic Plateau. TheVDLL is parallel to, but not continuous with, the betterknown Vening Meinesz Fracture Zone (VMFZ), a left-


stepping dextral strike–slip fault system that separates theNorfolk Basin from the Reinga Ridge (Herzer andMascle, 1996).

Sillimanite schists, rapidly exhumed in the EarlyMiocene, have been dredged from Cavalli Seamountwhich lies near a jog in the continental shelf edgebetween the Whangaroa and Knights Basins (Mortimeret al., 2003). Either side of Cavalli, the borders of theInner Sedimentary Basins against the continental shelfand Poor Knights Seamount Chain (VDLL; Fig. 1 inset)are roughly linear and NW-trending; in contrast theborder of the Outer Volcanic Plateau with the South FijiBasin is very embayed. The western boundary of thePlateau (as drawn in Fig. 1) with the Three Kings Ridgeis somewhat arbitrary (magnetic and gravity trends ofthe two features appear continuous with each other). Theeastern part of the Northland Plateau is more obviouslytransected by the NNE structural and volcanic trends ofthe Colville Ridge and Havre Trough (Fig. 1).

2.3. Three Kings Ridge

The Three Kings Ridge has generally been treated asa simple, narrow bathymetric feature extending northfrom Northland, which separates the Norfolk and SouthFiji Basins (e.g. Davey, 1982; Ballance, 1999). Recentgravity, seismic and bathymetric surveys have shownthe ridge to be a much wider and more complexcomposite feature (Herzer and Mascle, 1996; Herzeret al., 2000; Bernardel et al., 2002; Sdrolias et al., 2004).Strong north–south gravity lineaments are present that,based on the identification of the fault-bounded CagouTrough and Weta Terrace, are of probable tectonic,rather than volcanic, origin. The main ridge is flankedby the Fantail Terrace and Sarah Seamount group to theeast, and the Bates Plateau and Weta and TuataraTerraces (new names for the informal upper and lowerThree Kings terraces of Herzer and Mascle, 1996) to thewest. These latter terraces are basically fault-boundedplatforms between the crest of the Three Kings Ridgeand the abyssal Norfolk Basin (see Mortimer et al., 1998Fig. 3). Rock samples have been dredged from sevensites on the Three Kings Ridge (Fig. 1) and consist ofserpentinite, Late Eocene boninite, Early Mioceneandesite and Early Miocene shoshonite (Mortimeret al., 1998; Bernardel et al., 2002; Sdrolias et al., 2004).

2.4. Colville Ridge

The Colville Ridge is regarded as a classic remnantarc, separated from the active Kermadec arc by theHavre Trough (Ballance et al., 1999 and references

therein). To the north, the Colville Ridge becomes theLau Ridge which, in turn, connects bathymetrically withFiji. The only dated lava from the entire Colville Ridgeis a 5.5 Ma (K–Ar whole rock age) arc basalt (Adamset al., 1994). At c. 32°S many features of the Colville-Lau Ridge change: to the north it becomes wider andshallower and has a more northerly trend, and the LauTerrace abuts the Minerva Plain (Fig. 1). Morphologicalchanges in the Kermadec Ridge and intervening Havre–Lau basins also occur at c. 32°S.

Ballance et al. (1999) report a 7.8 Ma K–Ar age for alava from the Kermadec Trench as well as Oligocene toMiocene microfossils from volcaniclastic sediments onthe Kermadec Ridge. They interpret that, prior to inferredcessation of Colville volcanism and opening of the HavreTrough at about 5Ma, the formerly conjoinedKermadec–Colville ridge was volcanically active.

3. Methods

Each rock dredge was sorted on board into broadlydifferent rock types. Further examination, sawing andsorting of rocks were done at GNS Science. About 130representative samples were selected for thin sectioning,micropaleontology and geochemical work. Prior tochemical analysis, manganese and weathering rindswere removed to leave as fresh a sample as possible, andthin rock slabs were soaked in deionised water for a weekto remove seawater salt. No acid leaching was done. Therocks were crushed in a tungsten carbide ring mill;analyses of quartz blanks crushed at GNS Science haverevealed no issues with Nb or Ta contamination fornormal sized samples (Appendix A, see also Roser et al.,2003). However, given the variable Nb/Ta ratios for theoften very small samples in this study, we have not usedTa in our geochemical interpretations. Major elements,loss on ignition, and As, Ba, Ce, Cr, Cu, Ga, La, Nb, Ni,Pb, Rb, Sc, Sr, Th, U, V, Y, Zn and Zr were determinedfor all samples by XRF (X-ray fluorescence) analysis atSpectrachem Analytical, Wellington. For the subset ofsamples shown in Table 1, Cs, Hf, Pb, Ta, Th, U, Zr andrare earth elements (REEs) were determined by ICP-MS(inductively coupled plasma mass spectrometry) meth-ods at Washington State University, Pullman, and atGrenoble and Montpellier Universities. Trace elementICP-MS analyses followmethods described in Knaack etal. (1994), Barrat et al. (1996) and Lapierre et al. (1997).

Samples were dated by conventional argon step heatingat the University of Santa Barbara using techniques similarto Faulds et al. (1996). Samples were irradiated in thecadmium-lined tube at Oregon State University in threeseparate packages for 3–8 MW-h. The irradiations were


monitored with Taylor Creek Sanidine (Dalrymple andDuffield, 1988) using an assigned age of 27.92 Ma.

Geochemical, electron microprobe and geochrono-logical data are stored on the PETLAB database (http://data.gns.cri.nz/pet) and listed in Appendix A. Analysesof rock standards, silica blanks and detection limits arealso presented in Appendix A.

4. Sample descriptions

Samples and their locations are described in theONSIDE I and II cruise reports (Shipboard Party, 1999;Herzer et al., 2004a) and in Appendix A. In general, theantiquity of the – mainly Miocene – rocks meant thatalmost all dredged samples were covered by Mn crusts,with Mn oxides also having penetrated joint planes andgrain boundaries. Zeolites, too, were almost alwayspresent (particularly in amygdules), as was orangesmectite clay as an alteration product of glass and olivine.

4.1. Minerva Abyssal Plain

The basaltic rocks penetrated at the bottom of DeepSea Drilling Project (DSDP) 205 and 285A have beendescribed by Churkin and Packham (1973) and Stoeser(1976). Plagioclase and olivine in the dolerites are freshbut glassy groundmass is somewhat altered to smectiteand celadonite. The Julia dredge site (Fig. 1) is an east-facing fault scarp from which a suite of altered basalticlavas and basaltic breccias was recovered. A range oftextural types was present, from dolerite to chilled basalt.Secondary alteration to chlorite+actinolite+epidoteassemblages is common. The Alison dredge site(Fig. 1; new name) is on a volcanic rise in the seaflooreast of the Loyalty Ridge. Samples from Alison areslightly altered olivine+plagioclase+pyroxene porphy-ritic pillow basalts. Some pillows have glassy rinds.

4.2. South Fiji Basin seamounts

Altered lavas, volcaniclastic sedimentary rocks, andsome limestones were recovered from three seamountsin the Sarah and Devonport seamount groups, and fromMargot, Marion and Coquille seamounts. All theseseamounts are in the western part of the South Fiji Basinand appear to be constructional volcanoes. Smallamounts of Mn-crusted lavas were also dredged fromMascarin and Matahourua, two large seamounts that risefrom the eastern Kupe Abyssal Plain.

Petrographically, the lavas can be divided into twogroups: (1) highly porphyritic, variably amygdaloidal cli-nopyroxene+plagioclase+FeTi oxide±olivine±biotite±

K-feldspar basalts, andesites and trachytes. Orange-brown smectite replaces groundmass and olivine, and aK–Na zeolite is common in amygdules but despite thealteration, original phenocrysts are commonly preserved;(2) less porphyritic, commonly amygdaloidal, olivine+plagioclase or aphyric basalts with chlorite+ calcitealteration assemblages. The distinction between these twopetrographic suites is developed in Section 5.2 below.


4.3.1. Lavas and volcanic brecciasThe majority of lavas from the Northland Plateau are

plagioclase+ pyroxene±olivine ±biotite ±hornblendeporphyritic basalts and/or andesites (Appendix A). Thelargest quantity of material was obtained from D21(Purerua Seamount). Small (b1 kg) pieces of Mncrusted and altered lavas were obtained from un-named volcanic peaks at D5, D6, D7 and D18 andapparent fault scarps at D19 and D20 (Fig. 1).

At sites D2 (Whangaroa Seamount), D4, D6, D14 andD25 analytical work was done on individual cm-sizedclasts in volcanic breccias. The polymict nature of thesebreccias, on slopes far from canyon systems, suggests tous they are fault scarp talus sampling a section through avolcanic pile, and are therefore representative of the localin situ geology. D2 material consists of palagonitebreccia with clasts of plagioclase+clinopyroxene basalt.Zeolite is present in amygdules. Petrographic examina-tion of seven lava samples from dredge D6, at the edge ofthe Northland Plateau indicated the presence of at leasttwo different suites of lavas. The dominant set (P63164,66, 67) was like most of the other Northland Plateaulavas in that they contain the phenocryst assemblageplagioclase+pyroxene+biotite and have minor smectiteand zeolite alteration. However P63165, 68, 72, from thesame dredge, contain just plagioclase phenocrysts andhave epidote+sericite+carbonate alteration. These lavaswere also slightly harder to crush. Carbonate alteration ina D6 rhyolite clast, P63169, suggests a correlation withthe latter basalt–andesite suite.

4.3.2. Polymict volcano-sedimentary brecciasAt sites P70, D9A and B, D11, D12, D16 and D17

significant proportions of breccias containing clasts otherthanbasaltic–andesitic lavaswere recovered in the dredges(Appendix A). In the case of D9, D11 and D12 extremelysoft and porous altered rhyolitic pumice breccias dominate.Breccia from D9B contains three sorts of clast: (1) cpx+plag+hbl lavas of the sort common on the NorthlandPlateau volcanic peaks (described above); (2) rarerepidotised lava and dolerite; (3) a single clast of quartz–

http://data.gns.cri.nz/pet

http://data.gns.cri.nz/pet

Fig. 2. Binary diagrams showing broad compositional ranges ofdredged rocks. A. Nb/Y versus Zr/Ti (Pearce, 1994); B. SiO2 versusK2O; C. SiO2 versus TiO2. All elements normalised to anhydrousvalues. Havre Trough and Northland Arc reference fields from data inMiddleton (1983), Ruddock (1990) and Gamble et al. (1993).


plagioclase granofels. Breccia from D9A additionallycontained greywacke, actinolitised gabbro and individualclastic grains of reddish garnet and biotite. In all cases,secondary quartz and epidote alteration preceeded (re-)deposition. Greywacke clasts and/or isolated pebbleshave also been found at dredge sites D16 (The Slab) and atP70. Large, well-rounded basalt and dolerite pebbles atD17 resemble those found in the D9 breccias. Thepolymict breccias are not considered further in this paper.

All dredges thus far made on Cavalli Seamount haveyielded exclusively plutonic and high grade metamor-phic rocks. Rocks from the ONSIDE I cruise have beendescribed in detail by Mortimer et al. (2003).

4.4. Other sites

A basalt from the central volcanic ridge of the HavreTrough (Sonne 135–36 dredge site; Haase et al., 2002) isa relatively fresh and unaltered black vesicular, plagio-clase–porphyritic basalt. The only newmaterial obtainedfrom the Colville Ridge for this study (OR99 dredge site)is hydrothermally altered sparsely porphyritic andesite.

Two dredges (D24A, B) on a spur of the continentalshelf near Cavalli Seamount yielded a variety ofporphyritic basalts, andesites and rhyolites.

5. Geochemistry

Following petrographic examination, representativelavas were chosen for geochemical analysis (see Table 1and Appendix A). The emphasis was on analysingpetrographically different assemblages to characterisedifferent igneous suites at each location. In some cases,lack of sample amount meant that only one lava fromeach dredge site could be analysed.

As mentioned above, most rocks contain somesecondary zeolite, smectite clay, phosphate, manganeseoxide and/or calcite. Smectite alteration imparts red andyellow hues to the rocks. We selected the least alteredsamples for geochemical analysis. Relative to typicalNorthland Plateau lavas, a manganese crust from site D7is enriched in Fe, Mn, P, LOI, As, REE, Pb, Nb, Y, Znand Zr and depleted in Si, Cr, Rb, Sc, U and Th (P63180,Appendix A). Our previous experience with pre-Quaternary submarine lavas is that P, Ca, Rb, U and Kcontents can also be particularly unreliable in terms ofprimary values (Mortimer and Parkinson, 1996).

Even after careful sample preparation, some analysedsamples were found to have significant MnO, P2O5 andloss on ignition contents (N0.2, N1 and N2 wt.%respectively). Where other material was available fromthe same dredge, such samples were not used for geo-

chemical interpretations. In interpreting the dredgedlavas, we have placed most emphasis on elementsgenerally considered to be less mobile during secondaryalteration e.g. Ti, Y, Zr, Nb, V, Th, REE.

5.1. Minerva Abyssal Plain

All analysed samples from Minerva Abyssal Plainare subalkaline basalts with moderate TiO2 contents


(Fig. 2). Normalised multi-element patterns generallyshow flat shapes with downward slopes at the large ionlithophile element (LILE) end (Fig. 3A). None of thesamples show convincing Nb anomalies, as might beexpected if there had been significant subducted slabinvolvement in their petrogenesis. Likewise, low Ba/Nband moderate Ti/V are consistent with mid-ocean ridgebasalt (MORB)-like compositions (Fig. 4).

The glass and holocrystalline rock analyses from theFAUST-2 dredge site in the Cook Fracture Zonereported by Bernardel et al. (2002) are very similar to

Fig. 3. Multielement diagrams for selected representative samples, and somemantle (Sun and McDonough, 1989) values. Cs, Rb, U, Ta, K, Pb and P haveigneous patterns.

the abyssal samples from this study, as is P57025 fromthe east edge of the Norfolk Basin (Figs. 1 and 3A;Mortimer et al., 1998). These have been interpreted asbackarc basin basalts (BABBs).

5.2. South Fiji Basin seamounts

5.2.1. Less porphyritic (ocean island) basaltsLavas from Matahourua, Mascarin, Marion, Coquille

and dredge SF9801-D3 on Margot have the highestTiO2 and Nb contents of the entire sample set and are

P number samples from Mortimer et al. (1998), normalised to primitivebeen omitted in order to give a clearer visual representation of primary

Fig. 4. Binary diagrams showing variation in chemistry between differentareas. A. Ba vs Nb (Perfit et al., 1980); B. Ti vs V (Shervais, 1982).


interpreted as ocean island basalts (OIBs) (Figs. 2and 4). The lavas also show convex-up mantle-normal-ised patterns, with peaks at normalised Nb (Fig. 3B;Matahourua and Mascarin lavas are strongly contami-nated by phosphate which increases heavy REE). Binarydiagrams show the typical low Ba/Nb and high Ti/Vratios of OIB suites (Fig. 4).

5.2.2. More porphyritic (potassic) lavasLavas dredged from Devonport East and Central,

Sarah North, Sarah West, Sarah Central and dredgeSF9801-D2 on Margot, are trachybasalts and trachyan-desites and distinctly different in composition to theOIBs. As shown below they belong to shoshonitic rockseries (Morrison, 1980). These lavas have high toextreme total alkali contents, largely due to highpotassium, coupled with low TiO2 contents (Fig. 2).Given the almost ubiquitous smectite and zeolitealteration, all of the high K contents cannot be assumedto be primary igneous values, but the presence of biotitephenocrysts confirms a potassic suite. The normalised

incompatible element abundance patterns have straight-ish negative slopes, steeper than that of the OIBs exceptfor Nb and Ti troughs, and LILE peaks (Fig. 3C). Theselatter features are indeed similar to shoshonites andfurther suggest a partly subduction-related origin. Theshoshonitic nature is confirmed on the Ba vs Nb plot(Fig. 4). Electron microprobe analyses of clinopyrox-enes (Appendix A) also reveal the typically high Ca+Naand low Ti of shoshonites (Mortimer et al., 1998).

The most primitive lava in the shoshonitic suite is aSarah West lava which contains Cr spinel, lacks biotiteand has 5.6 wt.% MgO and 249 ppm Cr. The mostevolved lava is a biotite+sanidine porphyritic trachytefrom Devonport East seamount. The inferred primarypotassium content of these lavas (estimated by extrap-olation of normalised REE trends on Fig. 3C) farexceeds that of Clark Volcano in the modern Kermadecarc (Gamble et al., 1997).


Unlike for the South Fiji Basin there is no simplematchbetween geographic site, petrography and chemistry forNorthland Plateau lavas. Basalts and andesites from thePoor Knights Seamount Chain (Purerua, Whangaroa andP70) and the Outer Plateau (D4 lava, three clasts fromD6,D18-20) all plot in and/or parallel to the field of NorthlandArc lavas on an SiO2 vs TiO2 diagram (Fig. 2). On amulti-element normalised diagram many basalts and andesitesshow distinctive Nb and Ti troughs and relatively highconcentrations of LILEs typical of subduction-relatedlavas (Fig. 3D). Compositions range from a very depletedplagioclase-porphyritic (accumulative?) basalt from Pure-rua (D21) to very LILE-enriched hornblende trachyande-site at D19. Some trachyandesites (P70 and D19) areprobably high-K to shoshonitic but, because of LILEmobility we cannot be sure they are shoshonites.

Possibly the most LILE-depleted lavas on theNorthland Plateau are present at the eastern, outerregion at sites D6 and D7 (Fig. 3E). However the LILEconcentrations of these D6 clasts and D7 lava arecomparable to those reported for the active northernKermadec–Tonga arc (Haase et al., 2002), Nb and Tianomalies are still evident, and the light REE contentsare atypical of MORB and backarc basin basalts. Likemost of the rest of the Northland Plateau, we interpretthe D6 and D7 lavas as subduction related.

A single, highly altered lava from the edge of thePlateau at D5 is a subalkaline basalt (Fig. 2). It has LILEenrichment, and high Zr andHf concentrations but noNbdepletion. This is the only non-arc type lava dredgedfrom the Northland Plateau and is plotted with the


petrologically similar OIBs of Kupe Abyssal Plain inFig. 3B.

5.4. Other sites

The Havre Trough basalt is an important backarcbasin reference with which to compare our other lavas.A normalised multi-element plot of the SO135-36 basaltshows a negative Nb anomaly together with Ti to Ybnormalised values slightly lower than MORB (Fig. 3A).Overall, the lava is similar (except in age, see Section 6)to some of the less depleted abyssal plain basalts.

The porphyritic basalts, andesites and rhyolite fromD24 on the Northland continental shelf edge show arange of high-K arc-like compositions (Figs. 2, 3Fand 4). In this respect they bracket the entire com-

Table 2Ar–Ar ages (Ma) of lavas analysed at University of California Santa Barbar

GNS # Location Rock Mat. UCSB #

Minerva Abyssal PlainP63847 DSDP205 Basalt Plag SB50-07P63849 DSDP285A Dolerite Plag SB50-05P63850 DSDP285A Dolerite Plag SB36-45P68211 Julia Basalt clast Plag SB52-25P68215 Alison Basalt Plag SB50-04

South Fiji Basin seamountsPotassic suiteP59772 Devonport Central Trachyandesite Biot SB18-72P59788 Devonport Central Trachyandesite Biot SB18-73P61717 Devonport East Trachyte Biot SB24-15P61724 Margot Trachyandesite Biot SB24-17P59769 Sarah West Basaltic andesite wr SB18-74P63825 Sarah North Basaltic andesite Plag SB36-53

Ocean island basaltsP63830 Marion Trachybasalt Plag SB36-49P67646 Matahourua Basalt wr SB50-12

Northland PlateauP61712 Poor Knights Chain P70 Trachyandesite Plag SB24-16P66825 Poor Knights Chain D21 Basalt Plag SB50-16P66794 Poor Knights Chain D14 Basaltic andesite Plag SB50-22P63153 Outer Plateau D4 Basalt clast Plag SB31-04P63158 Outer Plateau D4 Andesite clast hbl SB31-05P63165 Outer Plateau D6 Bas andesite clast Plag SB31-06P63166 Outer Plateau D6 bas trachyand clast Plag SB31-07P63179 Outer Plateau D7 Basalt Plag SB31-08P66802 Outer Plateau D20 Andesite hbl SB50-84P66803 Outer Plateau D19 Basalt Plag SB52-23P66807 Outer Plateau D19 Trachyandesite Biot SB50-20P66798 Outer Plateau D25 Andesite Plag SB50-09

Havre TroughP63474 Havre Trough SO135-36 Basalt wr SB31-09

Errors are 2 sigma. Mat = material; bas = basalt; plag = plagioclase; biot =

positional range of the Northland Arc and Three KingsRidge (Fig. 3F).

6. 40Ar/39Ar geochronology

Where present, fresh biotite and hornblende weredated from porphyritic lavas, fresh plagioclase wasdated from other porphyritic lavas, and groundmassconcentrates were dated from unaltered aphyric basalts.Plagioclase separates typically had very low K/Ca ratiosand particular attention was paid to monitoring thetailing correction in order to give accurate ages. Resultsare summarised in Table 2 and gas release spectra fromselected samples are shown in Fig. 5. In the text andTable 2 2σ errors are reported (note only 1σ error barsare plotted in Fig. 5). For almost all plagioclase ages, our

a (analysts P.B. Gans and A.T. Calvert)

K/Ca TF age Plateau age Isochron age Preferred age

0.004–0.006 26.4 25.9±0.6 26.2±2.9 26.0±1.00.011–0.013 23.1 22.7±0.3 22.7±1.6 22.8±0.40.004–0.005 22.0 21.9±0.8 21.5±2 21.9±0.70.005–0.007 21.8 22.1±1.8 22.2±2.4 22.1±1.80.002–0.003 19.2 19.3±1.3 19.4±3.3 19.3±1.5

40–209 20.7 20.7±0.1 20.8±0.2 20.7±0.0553–148 20.8 20.7±0.1 20.6±0.1 20.7±0.0556–195 20.0 19.9±0.1 19.9±0.1 19.9±0.134–78 20.5 20.5±0.1 20.5±0.1 20.5±0.10.46–0.48 21.0 21±2 15.5±1.2 21±20.025–0.040 20.2 20.4±0.22 21.8±0.6 20.6±0.5

0.017–0.023 16.1 16.2±0.2 16.3±0.3 16.2±0.20.014–0.075 14.7 14.1±0.2 15.1±0.4 15±2

0.023–0.024 19.5 19.7±0.2 19.5±0.3 19.7±0.20.001 19.2 19.2±2.1 18.6±6 18.5±40.005 14.9 14.7±0.6 14.6±1.5 14.7±1.00.005–0.007 18.5 17.9±0.8 18.0±1.4 17.9±0.80.028–0.030 21.1 21.9±0.3 22.6±0.3 21.9±0.30.004 19.9 20.1±1.9 19.5±3.6 20.1±1.90.007 16.1 17.6±0.9 – 17.6±0.90.008 36.6 22±9.3 – 22±90.015–0.030 31.6 31.7±0.2 31.5±0.8 31.7±0.20.003–0.006 20.0 20.4±0.8 21.1±1.8 21.1±1.8100–400 20.4 20.4±0.04 20.4±0.07 20.4±0.050.008–0.012 2.5 1.04±0.8 1.4±1.4 1.2±0.8

0.021–0.068 1.3 1.12±0.38 0.76±0.64 1.1±0.4

biotite; wr = whole rock; hbl = hornblende; TF = total fusion.

Fig. 5. Selected Ar–Ar spectra. Error bars are ±1 sigma without error in the J parameter.


cited preferred age error is greater than analytical un-certainty, a conservative approach that we considerappropriate given the ultra low K/Ca of the samples.

6.1. Minerva Abyssal Plain and South Fiji Basinseamounts

Plagioclase separates from the DSDP drillcores andJulia dredge site give good flattish, spectra that become

irregular only in the high temperatures of gas release(Fig. 5). Different samples from DSDP285A are withinerror of each other and also within error of the (lessprecise) Julia age (Table 2). Plagioclase from Marionseamount also gives a good, flat, spectrum that isdistinctly younger than the abyssal DSDP samples. Thehigher K/Ca from the Marion OIB gives a more preciseage than from the abyssal tholeiites (Fig. 5). There is nocompelling evidence to suggest that the crystallisation


age of the plagioclases is anything older than the agescited in Table 2.

Biotite from the Devonport East trachyte and Margottrachyandesite (Fig. 5) both yielded plateaus in slightlyU-shaped argon release spectra. For both these shosho-nitic samples, highly precise inverse isochron ages of19.9±0.1 and 20.5±0.1 Ma respectively were obtainedusing 100% of the gas.


Lavas from the Poor Knights Seamount Chain gaveAr–Ar plagioclase ages of 19.7±0.2 Ma (P70; possiblyshoshonitic) and 18.5±4 Ma (D21, Purerua; Table 2).These are essentially coeval with the time of rapidcooling and exhumation of the nearby Cavalli Seamountschists (Mortimer et al., 2003).

Hornblende from an andesite dredged from OuterPlateau site D20 gave the oldest age from the NorthlandPlateau (31.7±0.2 Ma; Fig. 5). Ages of plagioclasefrom nearby D19 basalt give 21.1±1.8 Ma and ofbiotite from a D19 trachyandesite (possibly shoshoni-tic), 20.4± 0.05 Ma. Two separate breccia clasts fromOuter Plateau volcanic peak D4, gave ages of 17.9±0.8 Ma (plagioclase) and 21.9±0.3 Ma (hornblende).Two clasts from the D6 breccia were dated: plagioclasefrom an epidote-altered basaltic andesite gave 20.1±1.9 Ma and plagioclase from a zeolite-altered por-phyritic trachyandesite 17.6±0.9 Ma. These agesoverlap but the more altered rock has a permissiblyolder age. The arc basalt from Outer Plateau D7 gavea very imprecise plagioclase age of 22±9 Ma. Theyoungest Miocene age was given by plagioclase fromD14 from the eastern Northland Plateau: although thespectrum is quite irregular, the age of 14.7±1.0 Ma isdistinctly younger than the other Northland Plateaulavas (Fig. 5).

6.3. Other sites

Havre Trough SO135-36 basalt is essentially unal-tered. A whole rock (groundmass) separate gave a lowprecision but distinctly young age of 1.12±0.38 Ma age.A low seamount on the easternmost Northland Plateau,D25, gave a plagioclase age of 1.2±0.8 Ma. Although itis physically part of the Northland Plateau, the youngage must be regarded as an expression of westernmostColville volcanism.

We have reinterpreted an age reported in Mortimeret al. (1998). This is a whole rock age from P57142, anOIB from a seamount in the South Norfolk Basin. Wenow judge the sample to possibly have excess argon

which means it could be younger, and less precise, thanthe plateau age reported by Mortimer et al. (1998). Wereinterpret the crystallisation age for the basalt, as theisochron age of 15.8±3.4 Ma.

7. Discussion

7.1. Age of South Fiji Basin ocean crust

7.1.1. Minerva Abyssal PlainDSDP 205 was drilled in crust interpreted to be of

anomaly 12 (c. 31 Ma) age. Our Ar–Ar age of 26±1 Mafor the DSDP205 basalt suggests that this magneticinterpretation requires modification. Our basalt age is,however, in agreement with Churkin and Packham's(1973) interpretation of pillow basalt eruption contem-poraneous with, and/or injection of basalt into, unlithi-fied late mid Oligocene nanno ooze (c. anomaly 10 age).DSDP 285 was drilled in crust of interpreted anomaly8 (c. 26 Ma) age (Davey, 1982; Fig. 1). As with DSDP205, our Ar–Ar ages of 22.8±0.4 and 21.9±0.7 Ma forsamples from DSDP 285 are significantly younger thanthe inferred magnetic anomaly picks (Figs. 1 and 6). InDSDP 285A, the actual basalt-sediment contact was notobserved but was inferred to possibly be intrusive(Stoeser, 1976). We re-interpret the contact relations asearly Middle Miocene sediments deposited on a thickEarly Miocene basalt with a chilled flow top. The basaltwould be proper oceanic crust basement not a later, off-axis sill. Thus although we have not re-examined thecore, our new Ar dates are consistent with the reportedfossil ages from both DSDP sites.

Further evidence that the Minerva spreading centrespersisted into the Miocene and did not cease atanomaly 7 (c. 25 Ma) comes from the 23 Ma age of abasalt from the Cook Fracture Zone (Fig. 6; Meffreet al., 2002; Sdrolias et al., 2004), 22.1±1.8 Ma agefrom the Julia Lineament basalt and 19.3±1.5 Ma agefrom the Alison dredge (this paper). An MORB fromthe east edge of the Norfolk Basin abyssal plain has anage of 19.8±0.8 Ma (Mortimer et al., 1998). Thus fiveout of the six sites at which MORB and/or BABBlavas have been sampled in the South Fiji and NorfolkBasins have given Early Miocene ages, and DSDP 205has given a Late Oligocene, not an Early Oligocene,age (Fig. 6).

7.1.2. Kupe Abyssal PlainThere are no dredge or drill samples of abyssal ocean

crust from Kupe Abyssal Plain (Fig. 6). Local minimumages for various parts of the Kupe Plain are provided bythe OIB and shoshonite ages from Devonport Central,

Fig. 6. Summary of the age and composition of lavas in the Norfolk Ridge–New Zealand–Colville Ridge area. New ages are in bold type; other datafrom Mortimer et al. (1998), Ballance et al., 1999, Hayward et al. (2001), Meffre et al. (2002) and Sdrolias et al. (2004). See Fig. 1 for geographicnames and abbreviations.


Margot, Sarah West and Sarah North (21 Ma), Devon-port East (20 Ma), Marion (16 Ma), and Matahourua(15 Ma) seamounts. Devonport East sits astride, andamplifies the westernmost interpreted linear magneticanomaly in Kupe Plain (Fig. 1) which has beeninterpreted as either anomaly 7 (c. 25 Ma; Davey,1982; Malahoff et al., 1982) or anomaly 12 (c. 31 Ma,Sdrolias et al., 2003). With the more accurate delinea-

tion of the Northland Plateau by CANZ (1997), Herzeret al. (2000) and Stagpoole (2002), the linear magneticanomalies from Kupe Abyssal Plain appear to penetrateup to 100 km into the Northland Plateau (Fig. 1),although their sizes increase and their shapes changeradically; Sdrolias et al. (2003), perhaps justifiably, stopthem at the foot of the Northland Plateau. The acousticbasement and volcaniclastic aprons of the 18–22 Ma


volcanoes at Outer Plateau sites D4 and D7 merge withthe southernmost Kupe oceanic crust (Herzer et al.,2000). Collectively, these points lead us to believe thatthe matching of the Kupe anomalies to simple abyssalspreading patterns of Oligocene age is highly suspect.A critique of existing Kupe magnetic anomalyinterpretations is beyond the scope of this paper butour new radiometric ages and seismic stratigraphicinterpretations suggest that at least some of the KupePlain opened as the Northland Plateau grew, i.e. in theEarly Miocene.

If Sdrolias et al. (2003) are correct that the JuliaLineament is a fossil spreading centre then our 22.1±1.8 Ma basalt from there could approximate the age ofcessation of a Kupe spreading centre arm. However inthe sparse seismic and bathymetric data (Davey, 1982),updated by several recent multibeam echo sounding andseismic crossings of the feature (Herzer et al., 2005), theJulia Lineament more closely resembles a fault scarp(see also Fig. 7). The true origin of the lineament willonly be revealed with further surveys.

7.2. Significance of shoshonitic lavas

Shoshonitic lavas have now been dredged from eightof the sites in Fig. 6 (inverted triangle symbols) and

Fig. 7. Vector calculation of Vening Meinesz Fracture Zone (VMFZ)slip rates at c. 20 and 16 Ma based on possible hotspot trails in theSouth Fiji Basin. CFZ = Cook Fracture Zone. Same area as Fig. 1 withshaded predicted bathymetry background (Stagpoole, 2002). 67±5 mm/yr Early–Middle Miocene Australian-hotspot plate motion fromMcDougall and Duncan (1988).

possible high-K to shoshonitic lavas are present at D19,D24 and P70. Prior to the discovery of the South FijiBasin seamount shoshonites reported here, Mortimeret al. (1998) interpreted the two shoshonites from theThree Kings and Norfolk Ridge to have been emplacedabove the backmost part of a single, west-facing, LateOligocene–Early Miocene arc that subsequently mi-grated rapidly east in response to Pacific trench retreat.Our discovery of shoshonites east of the Three KingsRidge potentially complicates this interpretation.

Shoshonitic volcanism has been reported from anumber of tectonic settings (Morrison, 1980). In thecase of the intraoceanic South Fiji Basin seamounts wecan probably discount origins involving melting of, orsignificant contamination by, continental crust. TheHFSE-depleted nature of the lavas points to a partlysubduction-related origin either: (1) in the backmost partof a mature arc system (Morrison, 1980; Kepezhinskas,1995); (2) associated with arc deformation in thetermination of a subduction phase, including subductionflip or slab breakoff (Morrison, 1980; Davies and vonBlanckenburg, 1995); (3) rifting of a volcanic arc (Sternet al., 1988; Gill and Whelan, 1989; Kepezhinskas,1995).

The presence of high-K to shoshonitic rocks bothwest and east of the Three Kings Ridge means that, untiltectonic features are better defined by seismic work, it isinadvisable to use any of the shoshonites as evidence forlocal subduction polarity as was done by Mortimer et al.(1998). Their apparent short 19.9–20.7 Ma time rangeof eruption, and association with both OIBs and low-medium K suites (Fig. 6), leads us to invoke arc-riftingmodels for their petrogenesis. In the Marianas and Fiji,like in the South Fiji Basin-Northland Plateau region,shoshonites erupt shortly after the start of arc rifting and/or basin opening and extend across the entire arc andinto bordering basins.

7.3. Nature of the Northland Plateau

Samples recovered from the Northland Plateau in thisstudy consist mainly of inferred flows and breccias ofsubduction-related basalts, andesites and rhyolites thatrange in age from 32–14 Ma (with most in the 22–19 Ma age range). The 32 Ma lava is from a riftedterrace at the junction of the Outer Northland Plateauand eastern Three Kings Ridge (Herzer et al., 2004b). Atno site were exclusively fine-medium grained sedimen-tary rocks dredged, as might be expected for anaccretionary prism. Greywacke and dolerite clasts arepresent in breccias at the D9 and P70 dredge sites on thePoor Knights Seamount Chain, and high grade schists


have been obtained from Cavalli Seamount (Mortimeret al., 2003).

Thus the acoustic basement of the Outer NorthlandPlateau and much of the Poor Knights Seamount Chainappears to be fundamentally a constructional volcanicfeature with Late Oligocene and Early Miocenesubduction-related basalts, andesites and rarer rhyolitesexposed in several places. Magnetic and gravityanomalies, along with morphostructural features, indi-cate that the Plateau northeast of the Van der LindenLineament (Fig. 6) is a probable along-strike continua-tion of the Three Kings Ridge, and that Colville Ridgestructural and magmatic features have been super-imposed on its eastern part. The continent–ocean crustboundary could lie close to the Van Der Linden Line-ament, but its precise position has been obscured byOligocene–Miocene volcanism.

There is thus no compelling evidence, at present, tosuggest that the Poor Knights Seamount Chain andOuter Plateau comprise (a) thickened or overthrust KupeAbyssal Plain backarc crust without arc-related volca-nics; (b) a sedimentary accretionary prism; (c) anoffshore continuation of the Late Cretaceous–OligoceneNorthland Allochthon (see also Herzer et al., 2000) or(d) a trapped piece of Cretaceous Hikurangi Plateau(Mortimer and Parkinson, 1996).

7.4. Widespread Early Miocene volcanism

Prior to the present study and that of Mortimer et al.(1998), known Early Miocene subduction-related vol-canism in the area of Fig. 6 was restricted to onlandnorthern New Zealand (the “Northland Arc”; Ballanceet al., 1982; Brothers, 1986; Spörli, 1989). Mortimer etal. (1998) showed that at least parts of the Three KingsRidge and Norfolk Basin consisted of Early Mioceneandesites, shoshonites, abyssal tholeiites and alkalibasalts. The present study and that of Bernardel et al.(2002) has still further expanded the known area ofEarly Miocene lavas to include Cook Fracture Zone,Minerva Abyssal Plain, Julia Lineament, all seamountsthus far sampled rising from Kupe Plain, and much ofthe Northland Plateau. Ballance et al. (1999) from fossilevidence show that the Kermadec and Colville Ridgeswere sites of volcaniclastic deposition in the EarlyMiocene. Fiji also has an Early Miocene arc volcanicrecord (Wharton et al., 1995).

The presence of Early Miocene subduction-relatedlavas in a vast, non-linear 400×300 km region enclosingthe Northland Plateau and onland Northland (Fig. 6)demands an explanation. This distribution has beenachieved either by unusual plate tectonic processes and/

or tectonic juxtaposition of formerly separate arcs (seeSection 8 below).

7.5. Hotspot migration trails and basin opening rates

The present day half spreading rate for the northernLau Basin is c. 50 mm/yr (Taylor et al., 1996). This iscomparable to the calculated half spreading rate forMinerva Plain between DSDP 205 and 285 of c. 50 mm/yr, measured perpendicular to magnetic anomalies usingour new argon dates (see Section 7.1.1 above). It isinstructive to compare these values with bracketedspreading rates for the Norfolk Basin and Kupe Plain,based on various assumptions. Excluding the riftedterraces east and west of the Three Kings Ridge, thewidths of both the eastern Kupe Plain and South NorfolkBasin are c. 300 km. The opening of each could thus beachieved with a (not unreasonable) half spreading rate of50 mm/yr in only 3 m.y.

The E to ENE elongation of Devonport–Margot,Mascarin, Matahourua and Marion seamounts isreminiscent of backarc basin migration trails of thekind described for the Havre Trough by Wright et al.(1996). OIB and potassic lavas with possible astheno-spheric sources have been sampled from these KupePlain seamounts which raises the possibility that theycan be treated as very short-lived, scattered hotspottrails. Fig. 7 is a simple analysis of these putative shorthotspot tracks assuming that a rigid Kupe microplatemoves solely in response to northward motion of theAustralian Plate and to ESE motion along the VMFZ,both relative to a fixed mantle hotspot frame. Decreas-ing VMFZ slip rates of c. 100–67 mm/yr are suggestedin the interval 21–15 Ma (Fig. 7). If the Kupe Plain wasfully open during this time (Malahoff et al., 1982;Sdrolias et al., 2004) then the 15–16 Ma figures needonly match the opening rate, and age, of the NorfolkBasin. However if Kupe Plain was also opening in theMiocene then a more complicated analysis is required.

Some Norfolk Basin spreading took place at 20–23 Ma (the ages of backarc tholeiites reported byMortimer et al., 1998 and Sdrolias et al., 2004), thoughthere was probably extension and asymmetric rifting ofthe Three Kings and southern Loyalty Ridges to formtheir western flanking terraces and plateaus, prior to this.There are few direct constraints on when Norfolk Basinspreading ceased but Herzer et al. (1997) note thatsynsedimentary deformation along the VMFZ continuedthrough into the Middle Miocene (16–11Ma). Availableevidence is thus consistent with our kinematic analysisconnecting the ages and shapes of the Kupe Basinseamounts with hotspot tracks related to VMFZ


movement and Norfolk (and possibly Kupe) Basinopening.

8. Tectonic models

There exists a variety of tectonic models of theCenozoic evolution of the New Zealand–New Caledo-nia–Fiji area. Many of these (e.g. Ballance et al., 1982;Brothers, 1986; Spörli, 1989; Malpas et al., 1992) predateany significant offshore work. Those that do use offshoredata are largely based on swath bathymetry and potentialfield datasets, and also use the New Caledonia obductionmodel of Aitchison et al. (1995) as a key part of offshoreinterpretations near New Zealand (e.g. Crawford et al.,2003; Sdrolias et al., 2003; Schellart et al., 2006).

8.1. New Caledonia vs New Zealand

Both New Caledonia and the Northland area of NewZealand can be regarded as emergent parts of theNorfolk and Reinga Ridges. From the Late Cretaceousto Eocene, much of this strip of Gondwanaland accretedterranes was a submerged, continental borderland.Along and outboard of this borderland, the Poya andTangihua basalts were erupted in submarine basins andwere subsequently thrust over continental crust in bothonland areas. It could be argued, given the still-submerged nature of much of the Zealandia continentalcrust, that the thrusting/obduction events were confinedto these two geographic areas, i.e. because it is stillsubmerged, none of the intervening Norfolk Ridge hasbeen overthrust.

Significant differences in the geological history ofNew Zealand and New Caledonia include: (1) theobduction event in New Caledonia took place in theLate Eocene (37–34 Ma) (Cluzel et al., 1994, 2001;Aitchison et al., 1995; Lagabrielle et al., 2005) but inNorthland the allochthon was emplaced during theWaitakian stage (25–22 Ma) (Spörli, 1989; Isaac et al.,1994); (2) the uppermost New Caledonia nappecomprises a thick section of mantle peridotites whereasthe Northland Allochthon comprises mainly basalts andsedimentary rocks (Spörli, 1989; Malpas et al., 1992;Isaac et al., 1994). The only significant peridotite massis at North Cape (Fig. 6); (3) the post-Eocene history ofNew Caledonia is one of very minor magmatism andgreat distance from a plate boundary (Cluzel et al., 2001,2005; Lagabrielle et al., 2005) whereas the NorthlandAllochthon emplacement coincided with inception ofthe Alpine Fault as the major Australia-Pacific plateboundary (Sutherland, 1999) and, during but largelyfollowing thrusting, abundant Miocene arc volcanics

were erupted onto the Northland Allochthon (Spörli,1989; Hayward et al., 2001); (4) throughout theCenozoic, New Caledonia was much further from thePacific-Australia pole of rotation than New Zealandmeaning that the total amount of trench rollback andwidth of backarc opening would have been significantlyless near the New Zealand pivot point (Sutherland,1999; Ballance, 1999).

8.2. Northeast-dipping subduction?

New Caledonia models use an inferred NE-dippingsubduction zone to explain the overthrust geometry ofthe ultramafic nappe (Cluzel et al., 1994, 2001;Aitchison et al., 1995; Mauffret et al., 2002; Lagabrielleet al., 2005 and references therein). The arc generated bythis east-dipping subduction is believed to be theLoyalty Ridge which lies c. 150 km NE of theallochthon front (Fig. 8A). Crawford et al. (2003) andSchellart et al. (2006) have continued this model southinto the New Zealand area and regard the NorthlandAllochthon emplacement and Three Kings Ridge–Northland Plateau volcanism as being due to the finalshutting down of this NE-dipping subduction systemnear New Zealand at the end of the Oligocene.

The alternative model for Northland Allochthonemplacement involves crustal delamination of a west-dipping Pacific slab (e.g. Spörli, 1989; Rait et al., 1991;Malpas et al., 1992) possibly involving a triple junction(Bradshaw, 2004). A west-dipping slab model readilyexplains the presence of the 26 Ma shoshonitic brecciason the Norfolk Ridge (Mortimer et al., 1998) and thelinear 23–15 Ma Northland arc chain that erupted on thecontinent and allochthon. The absence of thick ultramaficsheets in the Northland Allochthon does not necessarilydemand a lithosphere-penetrating megathrust.

Our present dating and petrological study providesno new information on the origin or tectonic mode of the25–22 Ma emplacement of the Northland Allochthon(see Bradshaw, 2004 for a recent discussion), nor on theage of the Northland Allochthon basalts (Whattam et al.,2005). We accept that both NE-subduction and flaketectonics models are possible and the first two panels ofFig. 8 are a summary of the two end-member options.Note that in both models the Norfolk Basin is closed, ornearly closed, at 27 Ma.

8.3. Preferred model

The main themes in our 26–15 Ma tectonic model(Fig. 8C–F) are: (1) driving force of a single, west-dipping, retreating Pacific slab for post-Northland

Fig. 8. Schematic model showing possible stages in development of the South Fiji Basin and surrounding ridges and plateaux.



Allochthon Miocene volcanism and basin opening nearNew Zealand; (2) progressive southward propagation ofbackarc basin opening in the Oligocene–Miocene from(in order) the North Loyalty, Minerva, Kupe and finallyNorfolk Basins; (3) rapid, tandem Early Mioceneopening of the Norfolk Basin and Kupe Plain, andsignificant co-eval opening of the Minerva Plain; (4)local doubling up of the width of the Early Miocene arcas the Three Kings–Northland Plateau segment wastectonically juxtaposed outboard of the Northlandsegment along the VMFZ and VDLL.

In the 25–22 Ma interval the Northland Allochthonwas emplaced onto Northland continental crust atbathyal depths. Spreading from the North LoyaltyBasin propagated south into Minerva Plain (DSDPsites) and into western Kupe Plain (beneath the slightlyyounger Sarah seamounts). 23 Ma tholeiites at theeastern end of the Cook Fracture Zone and exhumationof serpentinites in the Cagou Trough (Bernardel et al.,2002; Sdrolias et al., 2004) may indicate incipientNorfolk Basin spreading and/or Three Kings Ridgerifting. Features such as the Cagou Trough, Tuatara andWeta Terraces and related prominent north–southlineaments (Fig. 1) are plausibly extensional rift featuresrather than aligned volcanoes (Mauffret et al., 2002).

21–20 Ma was a period of rapid tectonic changeaccompanied by voluminous magmatism. The Minervaspreading centres propagated into the Norfolk Basin andeastern Kupe Plain. Rhomboidal opening of the NorfolkBasin took place along the Cook and Vening MeineszFracture Zones: movement on the latter, cutting themagmatic arc at a shallow angle, started to shuffle theThree Kings–Northland Plateau segment of the ThreeKings–Northland Plateau–Northland Arc outboard ofthe Northland segment. Rift related shoshonites inwestern Kupe Basin and Weta Terrace erupted in thisshort time period, concomitant with fast slip on theVMFZ. Arc volcanism waned on the Three Kings Ridgeand it became a remnant arc. In what is now onlandNorthland the active arc magmatism started to shift eastwith the retreating trench. At c. 20Ma, rapid exhumationfrom 10 km depth of the schists of the probablemetamorphic core complex of Cavalli Seamount (Fig.1; Mortimer et al., 2003) was a response to rapid KupeBasin opening and/or VMFZ/VDLL movement.

The 19–15 Ma interval saw final spreading in theNorfolk Basin and seamount-free parts of eastern KupePlain. By this time, active arc volcanism had migratedsouth and east of Auckland. From about 15 Ma, rapidtrench retreat ceased, the Australia-Pacific plate marginstabilised and the NNE–SSW Colville arc trend wasestablished through the North Island (Herzer, 1995).

8.4. Comparison with other models

Mortimer et al. (1998) ignored the South Fiji Basin intheir simple model of Miocene Norfolk Basin and ThreeKings Ridge development. The present work, based onnew data from the South Fiji Basin and environs,remedies that. We agree with Ballance et al. (1982),Brothers (1986), Spörli (1989), Malpas et al. (1992) andBallance (1999) that a west-dipping Pacific slab model isfeasible for the Miocene volcanism, albeit accompaniedby rapid tectonic and magmatic change offshore in the23–19 Ma interval. Crawford et al. (2003) and Schellartet al. (2006) have proposed NE-dipping subduction as amechanism, not just for Northland Allochthon emplace-ment, but also for Three Kings–Northland Plateauvolcanism. In this scenario, the Kupe Plain shoshoniteswould probably be interpreted as backmost-arc, collisionand/or slab-breakoff lavas (Morrison, 1980; Davies andvon Blanckenburg, 1995). However, unlike at NewCaledonia latitudes, the Poya–Tangihua backarc basinformed near New Zealand may not have been wideenough to make enough subducted slab to generate 12 m.y. of volcanism on the Northland Plateau; what Eocene–Oligocene subduction there was near New Zealand waslikely “amagmatic” (Ballance, 1999). If aNE-dipping slabexisted (e.g. to emplace the Northland Allochthon), itcould possibly have been undercut and replaced by adominant Pacific slab from 23 Ma. The volume andcontinuum of onland Northland volcanism from 23–15Ma seemsmore consistent with a gradual SEmigrationof one west-dipping, east-retreating subduction system(Fig. 8B–F) than a subduction flip during this time.

With west-dipping slab models, there has beendisagreement as to the orientation of the Pacific trenchthat generated the onland Miocene Northland arc.Ballance et al. (1982) proposed a NW-striking trenchwith a switch to a NNE trend in the Late Miocene. Incontrats, Brothers (1986) advocated a retreating trenchessentially striking subparallel to the present NNEKermadec trend. Our model of Norfolk and Kupe Basinopening in Fig. 8 provides a possible compromiseexplanation in that inception of onland Northland arcvolcanism at 23 Ma was formed by Pacific Platesubduction along a NW-striking trench whereas by15 Ma, Pacific Plate subduction was along a NNE-striking trench because of eastward movement of theColville arc and the Northland Plateau past onlandNorthland. In a more complex model, Ballance (1999)postulated three separate Miocene trenches to generatethe Three Kings, Colville and Northland arcs. As shownin Fig. 8, we regard these as possibly rapidly rifted anddispersed segments of the same, single, east-facing arc.


The Norfolk Basin appears to be one of the youngestof the Oligocene–Miocene backarc basins and it isunusual that the Norfolk Basin opened so far behind thePacific trench. Sdrolias et al. (2004) have invoked amantle plume mechanism as a possible driving force forEarly Miocene Norfolk Basin opening, additional tolithospheric tensional stresses due to trench rollback.Our study lends some support to this idea as we haveshown that, at Norfolk Basin latitudes, normal subduc-tion-related volcanism was occurring at least at 32 Ma,shoshonitic volcanism was largely confined to the 20–21 Ma period, and OIBs were widespread at 15–16 Ma.Thus there may be a trend to tap deeper mantle sourceswith time. We go one step further than Sdrolias et al.(2004) and suggest that many of the elevated features ofthe Norfolk Basin (e.g. Kingston Plateau, NepeanSaddle) could be thick piles of Miocene OIB eruptives,as seems to be indicated by existing dredges (Fig. 6;Mortimer et al., 1998).

8.5. Further tests

The biggest data gap in the region is knowledge ofthe age of crust and spreading directions in KupeAbyssal Plain. Longstanding and recent magneticanomaly interpretations suggest an Oligocene (anomaly7–12) age but with controversy about youngingdirections and geometry (Malahoff et al., 1982, Davey,1982; Sdrolias et al., 2003). Our radiometric dating oflavas (this paper) and seismic stratigraphic constraints(Herzer et al., 2000) instead suggest an earliest Mioceneage but are based on features not directly associated withabyssal ocean crust. The geometric–kinematic relation-ship between the non-parallel Cook and Julia linea-ments, and the seamount elongation (Fig. 7) remainsenigmatic. If an Oligocene age for Kupe Plain can beindependently confirmed, then our single Early Miocenewest-dipping Pacific slab model (Fig. 8) must beabandoned as the Three Kings arc to Pacific trenchgap of N500 km would be unreasonably large. Weregard establishing the age of different parts of KupePlain as a critical test for discriminating between ourmodel and that of Crawford et al. (2003).

Knowing whether there is, or is not, Cretaceousoceanic crust in the Norfolk Basin is also important forthe subduction polarity issue. The presence of a Poya–Tangihua Cretaceous basin is a key test of the Crawfordet al. (2003) model (Fig. 8A), whereas no Cretaceouscrust is predicted by the combined Kermadec–Norfolk(“Nordec”) Arc model (Fig. 8B). Dredging and directdating of lavas or sedimentary rocks in the deepest partsof the Norfolk Basin is sorely needed.

Swath bathymetric and detailed seismic surveys ofthe region are desirable in order to fully assess the rift,transform or trench significance of the structuresindicated by the Fantail, Weta and Tuatara Terraces,the Cagou Trough, the Julia and Van der LindenLineaments and Cavalli Seamount.

9. Conclusions

New analytical data from altered volcanic rocksrocks dredged or drilled from 38 sites between the HavreTrough, Three Kings Ridge and Northland Plateauprovide useful information on the tectonic evolution ofCenozoic arcs and basins in the SW Pacific Ocean.Early Miocene (18–22 Ma) volcanism is widespreadand includes a variety of different petrological suites.New Ar–Ar dates on abyssal tholeiites suggest that thecrust of much of the Minerva Abyssal Plain of thenorthern South Fiji Basin may be Early Miocene in age,not Oligocene as previously interpreted using magneticanomalies. Subduction-related lavas, identified by theirLILE enrichment and HFSE depletion are present on theNorthland Plateau. Shoshonitic lavas occur on bothsides of the Three Kings Ridge and are interpreted tohave formed during ultra-rapid arc rifting at 20–21 Ma.

A range of models can be used to explain our results,including microplate arrangements with NE-dippingslabs. However, we believe that from the Early Miocene(c. 23 Ma), the simplest model consist with the onland-offshore data is one of eastward rollback of a single arc–trench system controlled principally by a west-dippingPacific slab.

Acknowledgements

We thank the Captain and crew of the R/V Tangaroa,and our colleagues Jean Mascle, Bryan Davy, EtienneRuellan, Dan Barker, Kim Rose, Anya Duxfield andSteve Wilcox for assistance with dredging on theONSIDE I and II cruises. Cornel de Ronde, the scientificstaff and crews from the Sonne 135 and 167 cruises,Orange RoughyManagement Ltd., the National Instituteof Water and Atmospheric Research and the staff of theODP archive in San Diego supplied additional samplesfrom the Havre Trough, Colville Ridge, NorthlandPlateau and Minerva Plain. Valuable technical supportwas provided by Neville Orr, John Simes, John Hunt,Charles Knaack, Diane Johnson, Martin Wong, BelindaSmith Lyttle and Lorraine Paterson. We have had usefuldiscussions with Rupert Sutherland, Peter Ballance,Philippa Black, Sebastian Meffre, Bryan Davy, DanBarker,Mike Isaac, Tony Crawford, Bernhard Spörli and


Larry Lawver. Two anonymous referees provided usefulreviews (one of them extremely thorough) of an earlierversion of the manuscript. The work was mainly fundedby the New Zealand Public Good Science Fund. We aregrateful to the French Ministry of Foreign Affairs forsupport through the France–New Zealand CulturalAgreement. Land Information New Zealand gavepermission to publish some data gathered as part of theNew Zealand Continental Shelf UNCLOS project.

Appendix A. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.margeo.2006.10.033.

References

Adams, C.J., Graham, I.J., Seward, D., Skinner, D.N.B., 1994.Geochronological and geochemical evolution of late Cenozoicvolcanism in the Coromandel Peninsula, New Zealand. N.Z.J. Geol. Geophys. 37, 359–379.

Aitchison, J., Clarke, G., Cluzel, D., Meffre, S., 1995. Eocene arc-continent collision in New Caledonia and implications for regionalsouthwest Pacific tectonic evolution. Geology 23, 161–164.

Ballance, P.F., 1999. Simplification of the southwest Pacific Neogenearcs: inherited complexity and control by a retreating pole ofrotation. Geol. Soc. Lond. Spec. Publ. 164, 7–19.

Ballance, P.F., Pettinga, J.R., Webb, C., 1982. A model of theCenozoic evolution of northern New Zealand and adjacent areas ofthe southwest Pacific. Tectonophysics 87, 37–48.

Ballance, P.F., Ablaev, A.G., Pushchin, I.K., Pletnev, S.P., Birylina, M.G.,Itaya, T., Follas, H.A., Gibson, G.W., 1999. Morphology and historyof the Kermadec trench–arc–backarc basin-remnant arc system at 30to 32°S: geophysical profile, microfossil and K–Ar data. Mar. Geol.159, 35–62.

Barrat, J.A., Keller, F., Amossé, J., Taylor, R.N., Nesbitt, R.W., Hirata, T.,1996.Determination of rare earth elements in sixteen silicate referencesamples by ICP-MS using a Tm addition and an ion-exchangechromatography procedure. Geostand. Newsl. 20, 133–139.

Bernardel, G., Carson, L., Meffre, S., Symonds, P., Mauffret, A., 2002.Geological and morphological framework of the Norfolk Ridge toThree Kings Ridge region. Geoscience Australia Record 2002/08.

Bosch, D., Laporte-Magoni, C., Mortimer, N., Calvert, A., Herzer, R.H.,2002. Isotopic and geochemical study of the Cenozoic volcanicrocks dredged from the South Fiji Basin and Northland Plateau.Western Pacific Geophysics Meeting Supplement AbstractSE42D-07. Eos, Transactions American Geophysical Union,vol. 83 (22), p. 104.

Bradshaw, J.D., 2004. Northland Allochthon: an alternative hypothesisof origin. N.Z. J. Geol. Geophys. 47, 375–382.

Brothers, R.N., 1986. Upper Tertiary and Quaternary volcanism andsubduction zone regression, North Island, New Zealand. J. R. Soc.N.Z. 16, 275–298.

CANZ (Charting Around New Zealand Group), 1997. Undersea NewZealand (New Zealand Region Physiography) 1:4 000 000, 2nd ed.NewZealandOceanographic Institute ChartMiscellaneous Series 74.National Institute of Water and Atmospheric Research. Wellington.

Churkin, M., Packham, G.H., 1973. Volcanic rocks and volcanicconstituents in sediments. Init. Rep. Deep Sea Drill. Proj. 21,481–493.

Cluzel, D., Aitchison, J.C., Clarke, G., Meffre, S., Picard, C., 1994.Point de vue sur l'évolution tectonique et géodynamique de laNouvelle Calédonie (Pacifique, France). C. R. Acad. Sci. Paris 319(II), 683–690.

Cluzel, D., Aitchison, J.C., Picard, C., 2001. Tectonic accretion andunderplating of mafic terranes in the late Eocene intraoceanic fore-arc of New Caledonia (Southwest Pacific): geodynamic implica-tions. Tectonophysics 340, 23–59.

Cluzel, D., Bosch, D., Paquette, J.L., Lemennicier, Y., Montjoie, P.,Ménot, R.P., 2005. Late Oligocene post-obduction granitoids ofNew Caledonia: a case for reactivated subduction and slab break-off. Isl. Arc 14, 254–271.

Crawford, A.J., Meffre, S., Symonds, P.A., 2003. 120 to 0 Ma tectonicevolution of the southwest Pacific and analogous geologicalevolution of the 600 to 220 Ma Tasman Fold Belt System. Geol.Soc. Am. Spec. Pap. 372, 383–403.

Dalrymple, G.B., Duffield, W.A., 1988. High precision 40Ar/39Ardating of Oligocene rhyolites from the Mogollon-Datil volcanicfield using a continuous laser system. Geophys. Res. Lett. 15,463–466.

Davey, F.J., 1982. The structure of the South Fiji basin. Tectono-physics 87, 185–241.

Davies, J.H., von Blanckenburg, F., 1995. Slab breakoff: a model oflithospheric detachment and its test in the magmatism anddeformation of collisional orogens. Earth Planet. Sci. Lett. 129,85–102.

Faulds, J.E., Feuerbach, D.L., Reagan,M.K., Metcalf, R.V., Gans, P.B.,Walker, J.D., 1996. The Mount Perkins Block, northwesternArizona: an exposed cross section of an evolving, preexten-sional to synextensional magmatic system. J. Geophys. Res.100, 15249–15266.

Gamble, J.A., Wright, I.C., Baker, J.A., 1993. Seafloor geology andpetrology in the oceanic to continental transition zone of theKermadec–Havre–Taupo Volcanic Zone arc system, New Zeal-and. N.Z. J. Geol. Geophys. 36, 417–435.

Gamble, J.A., Christie, R.H.K., Wright, I.C., Wysoczanski, R.J., 1997.Primitive K-rich magmas from Clark Volcano, southern KermadecArc: a paradox in the K-depth relationship. Can. Mineral. 35,275–290.

Gill, J.B., Whelan, P., 1989. Early rifting of an oceanic island arc (Fiji)produced shoshonitic to tholeiitic basalts. J. Geophys. Res. 94,4561–4578.

Haase, K.M., Worthington, T.J., Stoffers, P., Garbe-Schönberg, D.,Wright, I., 2002. Mantle dynamics, element recycling, and magmagenesis beneath the Kermadec Arc–Havre Trough. Geochem.Geophys. Geosyst. 3 (11), 1071, doi:10.1029/2002GC000335.

Hayward, B.W., Black, P.M., Smith, I.E.M., Ballance, P.F., Itaya, T.,Doi, M., Takagi, M., Bergman, S., Adams, C.J., Herzer, R.H.,Robertson, D.J., 2001. K–Ar ages of Early Miocene arc-typevolcanoes in northern New Zealand. N.Z. J. Geol. Geophys. 44,285–311.

Herzer, R.H., 1995. Seismic stratigraphy of a buried volcanic arc,offshore Northland, New Zealand and implications for Neogenesubduction. Mar. Pet. Geol. 12, 511–531.

Herzer, R.H., Mascle, J., 1996. Anatomy of a continent-backarctransform— the Vening Meinesz Fracture Zone northwest of NewZealand. Mar. Geophys. Res. 18, 401–427.

Herzer, R.H., Mortimer, N., 1997. A case for major Miocene spreadingin the South Fiji Basin. Eos 78 (46), F697.

http://dx.doi.org/doi:10.1016/j.margeo.2006.10.033

http://dx.doi.org/doi:10.1016/j.margeo.2006.10.033

http://dx.doi.org/10.1029/2002GC000335


Herzer, R.H., Chaproniere, G.C.H., Edwards, A.R., Hollis, C.J.,Pelletier, B., Raine, J.I., Scott, E.H., Stagpoole, V., Strong, C.P.,Symonds, P., Wilson, J.G., Zhu, H., Cobbly, T., 1997. Seismicstratigraphy and structural history of the Reinga Basin and itsmargins, Southern Norfolk Ridge System. N.Z. J. Geol. Geophys.40, 425–451.

Herzer, R.H., Mascle, J., Davy, B., Ruellan, E., Mortimer, N., Laporte,C., Duxfield, A., 2000. New constraints on the New Zealand–South Fiji Basin continent-back-arc margin. C. R. Acad. Sci. Paris,Sci. Terre Planet. 330, 701–708.

Herzer, R.H., Mortimer, N., Davy, B., Mascle, J., Laporte, C., Ruellan,E., 2001. Contrasting back-arc basins north of New Zealand:progression from active margin to intra-oceanic subduction, orseparate subduction systems? European Geophysical SocietyXXVI General Assembly, Nice, France, 25–30 march 2001.Geophysical Research Abstracts, vol. 3, p. 650.

Herzer, R.H., Davy, B., Mortimer, N., Laporte-Magoni, C., Barker, D.,2004a. Cruise report — GNS Cruise SF0202 “ONSIDE II”(Offshore Northland Seismic and Dredging Expedition II). Instituteof Geological and Nuclear Sciences File Report 2004-01.

Herzer, R.H., Mortimer, N., Davy, B., Barker, D., Quilty, P., 2004b.Geology and structure of the South Fiji Basin, southwest pacificOcean. Eos, Trans. - Am. Geophys. Union 85 (47) (Fall MeetingSupplement, Abstract T41A-1158).

Herzer, R.H., Roest, W., Barker, D., Mortimer, N., Mauffret, A., Lafoy,Y., 2005. Tectonic evolution of the South Fiji Basin: UNCLOShelps tackle regional tectonics. Eos, Trans. - Am. Geophys. Union86 (52) (Fall Meeting Supplement, Abstract T13D-0503).

Isaac, M.J., Herzer, R.H., Brook, F.J., Hayward, B.W., 1994.Cretaceous and Cenozoic sedimentary basins of Northland, NewZealand. Inst. Geol. Nucl. Sci. Monogr. 8.

Kepezhinskas, P., 1995. Diverse shoshonite magma series in theKamchatka Arc: relationships between intra-arc extension andcomposition of alkaline magmas. Geol. Soc. Lond. Spec. Publ. 81,249–264.

Knaack, C., Cornelius, S., Hooper, P.R., 1994. Trace element analysesof rocks and minerals by ICP-MS. Open File Report. Departmentof Geology, Washington State University, Pullman, USA.

Lagabrielle, Y., Maurizot, P., Lafoy, Y., Cabioch, G., Pelletier, B.,Régnier, M., Wabete, I., Calmant, S., 2005. Post-Eoceneextensional tectonics in southern New Caledonia (SW Pacific):insights from onshore fault analysis and offshore seismic data.Tectonophysics 403, 1–28.

Lapierre, H., Dupuis, V., Mercier de Lépinay, B., Tardy, M., Ruiz, J.,Maury, R.C., Hernandez, J., Loubet, M., 1997. Is the Lower DuarteIgneous Complex (Hispaniola) a remnant of the Caribbean plume-generated oceanic plateau? J. Geol. 105, 111–120.

Lawver, L.A., Gahagan, L.M., Campbell, D., 2002. Animation of thetectonic evolution of the southwest Pacific. Eos 83, T51D-03.

Malahoff, A., Feden, R.H., Fleming, S., 1982. Magnetic anomalies andtectonic fabric of marginal basins north of New Zealand.J. Geophys. Res. 87, 4109–4125.

Malpas, J.G., Spörli, K.B., Black, P.M., Smith, I.E.M., 1992.Northland ophiolite, New Zealand, and implications for plate-tectonic evolution of the southwest Pacific. Geology 20, 149–152.

Mauffret, A., Symonds, P., Bailleul, J., Gorini, C., van de Beuque, S.,Lafoy, Y., Bernardel, G., 2002. Structural evolution of theNorfolk Basin. Western Pacific Geophysics Meeting SupplementAbstract SE41D-08. Eos, Transactions American GeophysicalUnion, vol. 83 (22), p. 92.

McDougall, I., Duncan, R.A., 1988. Age progressive volcanism in theTasmantid Seamounts. Earth Planet. Sci. Lett. 89, 207–226.

Meffre, S., Symonds, P., Bernardel, G., Carson, L., Crawford, A.J.,2002. Oligocene collision of the Three Kings Ridge and initiationof the Tonga–Kermadec island arc system. Western PacificGeophysics Meeting Supplement Abstract SE41D-07. Eos,Transactions American Geophysical Union, vol. 83 (22), p. 91.

Middleton, L.M.H., 1983. TheWhangarei Heads calc-alkaline Tertiaryvolcanic complex, Northland New Zealand. Ph.D. Thesis,University of Auckland, New Zealand, unpublished.

Morrison, G.W., 1980. Characteristics and tectonic setting of theshoshonite rock association. Lithos 13, 97–108.

Mortimer, N., Parkinson, D.L., 1996. Hikurangi Plateau: a Cretaceouslarge igneous province in the southwest Pacific Ocean. J. Geophys.Res. 101, 687–696.

Mortimer, N., Herzer, R.H., Gans, P.B., Parkinson, D.L., Seward, D.,1998. Basement geology from Three Kings Ridge to West NorfolkRidge, southwest Pacific Ocean: evidence from petrology,geochemistry and isotopic dating of dredge samples. Mar. Geol.148, 135–162.

Mortimer, N., Laporte-Magoni, C., Calvert, A.T., Gans, P.B., Herzer,R.H., 2002. Early Miocene shoshonites and arc-related basalts inthe southern South Fiji Basin region, SW Pacific Ocean. WesternPacific Geophysics Meeting Supplement Abstract SE42D-08. Eos,Transactions American Geophysical Union, vol. 83 (22), p. 105.

Mortimer, N., Herzer, R.H., Walker, N.W., Calvert, A.T., Seward, D.,Chaproniere, G.C.H., 2003. Cavalli Seamount, Northland Plateau,SW Pacific Ocean: a Miocene metamorphic core complex? J. Geol.Soc. Lond. 160, 971–983.

Packham, G.H., Terrill, A., 1976. Submarine geology of the South Fijibasin. Init. Rep. Deep Sea Drill. Proj. 30, 617–633.

Pearce, J.A., 1994. A user's guide to basalt discrimination diagrams.In: Wyman, D.A. (Ed.), Trace Element Geochemistry of VolcanicRocks: Applications for Massive Sulphide Exploration. GeologicalAssociation of Canada Short Course Notes, vol. 12, pp. 79–113.

Perfit, M.R., Gust, D.A., Bence, A.E., Arculus, R.J., Taylor, S.R.,1980. Chemical characteristics of island arc basalts: implicationsfor mantle sources. Chem. Geol. 30, 227–256.

Rait, G., Chanier, F., Waters, D.W., 1991. Landward- and seaward-directed thrusting accompanying the onset of subduction beneathNew Zealand. Geology 19, 230–233.

Roser, B.P., Kimura, J.I., Sifeta, K., 2003. Tantalum and niobiumcontamination from tungsten carbide ring mills: much ado aboutnothing. Geosci. Rep. Shimane Univ. 22, 107–110.

Ruddock, R.S., 1990. The Karikari Plutonics of Northland, NewZealand: the petrology of an arc-type intrusion and its envelope.Ph.D. Thesis, University of Auckland, New Zealand, unpublished.

Schellart, W.P., Lister, G.S., Toy, V.G., 2006. A Late Cretaceous andCenozoic reconstruction of the Southwest Pacific region: tectonicscontrolled by subduction and slab rollback processes. Earth-Sci.Rev. 76, 191–233.

Sdrolias, M., Muller, R.D., Gaina, C., 2003. Tectonic evolution of thesouthwest Pacific using constraints from backarc basins. Geol.Soc. Am. Spec. Pap. 372, 343–359.

Sdrolias, M., Mueller, R.D., Mauffret, A., Bernardel, G., 2004.Enigmatic formation of the Norfolk Basin, SW Pacific: a plumeinfluence on back-arc extension. Geochem., Geophys., Geosyst. 5(6), Q06005, doi:10.1029/2003GC000643.

Shervais, J.W., 1982. Ti–V plots and the petrogenesis of modern andophiolitic lavas. Earth Planet. Sci. Lett. 59, 101–118.

Shipboard Party, 1999. Cruise report: GNS Cruise SF9901 “ONSIDE”(Offshore Northland Seismic and Dredging Expedition), 19–29March 1999. Institute of Geological and Nuclear Sciencesunpublished report. Wellington, New Zealand.

http://dx.doi.org/10.1029/2003GC000643


Spörli, K.B., 1989. Tectonic framework of Northland, New Zealand.R. Soc. N.Z. Bull. 26, 3–14.

Stagpoole, V., 2002. The New Zealand Continent, 1:7 500 000, version1.0. Institute of Geological and Nuclear Sciences Geophysical MapGPM15. Institute of Geological and Nuclear Sciences, Lower Hutt,New Zealand.

Stern, R.J., Bloomer, S.H., Lin, P.-N., Ito, E., Morris, J., 1988.Shoshonitic magmas in nascent arcs: new evidence fromsubmarine volcanoes in the northern Marianas. Geology 16,426–430.

Stoeser, D.B., 1976. Igneous rocks from Leg 30 of the Deep SeaDrilling Project. Init. Rep. Deep Sea Drill. Proj. 30, 401–414.

Sun, S.S., McDonough, W.F., 1989. Chemical and isotopic systematicsof oceanic basalts: implications for mantle composition andprocesses. Geol. Soc. Lond. Spec. Pap. 42, 313–345.

Sutherland, R., 1999. Basement geology and tectonic development ofthe greater New Zealand region: an interpretation from regionalmagnetic data. Tectonophysics 308, 341–362.

Taylor, B., Zellmer, K., Martinez, F., Goodliffe, A., 1996. Sea floorspreading in the Lau backarc basin. Earth Planet. Sci. Lett. 144,35–40.

Weissel, J.K., Watts, A.B., 1975. Tectonic complexities in the SouthFiji marginal basin. Earth Planet. Sci. Lett. 28, 121–126.

Whattam, S.A., Malpas, J.G., Ali, J.R., Lo, C.H., Smith, I.E.M., 2005.Formation and emplacement of the Northland ophiolite, northernNew Zealand: SW Pacific tectonic implications. J. Geol. Soc.Lond. 162, 225–241.

Wharton, M.R., Hathway, B., Colley, H., 1995. Volcanism associatedwith extension in an Oligocene–Miocene arc, southwestern VitiLevu, Fiji. Geol. Soc. Lond. Spec. Publ. 81, 95–114.

Wright, I.C., Parson, L.M., Gamble, J.A., 1996. Evolution andinteraction of migrating cross-arc volcanism and backarc rifting: anexample from the southern Havre Trough (35°20'–37°S).J. Geophys. Res. 101, 22071–22086.

oligocene miocene tectonic evolution of the south fiji basin ......oligocene–miocene tectonic...

Documents