Excavation of buried Dun Mountain–Maitai terrane ophiolite by volcanoes of the Auckland Volcanic field, New Zealand

Abstract

We present a description of the crustal rocks that underlie the Auckland Volcanic Field (AVF) based on a diverse suite of country rock-derived lithic clasts in the phreatomagmatic tuff of Glover Park and Taylors Hill volcanoes. The clasts are dominated by mafic schistose and non-schistose, amphibolite grade, meta-igneous (metabasite) rocks. Structural and mineralogical studies of these rocks reveal a complex structural and metamorphic history, including metasomatism and retrogressive metamorphism, which can be linked to their deformation history. These metabasites have no genetic relationship with associated AVF basalts. Location, composition, deformation, metamorphism and metasomatism indicate that the metabasite clasts come from a melange along the eastern boundary of the ophiolite that causes the regional Junction Magnetic Anomaly (JMA), which passes beneath the Auckland region and connects up with exposures of the Dun Mountain–Maitai terrane of the South Island. This conclusion is supported by one specimen from a suite of lower-grade metamorphic, greywacke-type clastic ejecta that contains shell fragments which we interpret as equivalents of the Atomodesma fragments in the Dun Mountain–Maitai terrane. Other lithic ejecta are sandstones from the underlying Miocene Waitemata Group, Mesozoic greywacke with prehnite veins, and chert. Pervasive cataclasite networks in the lithic clasts indicate that the Glover Park and Taylors Hill volcanoes mined a crustal fault zone within the Mesozoic basement several hundred metres deep, at an unconformity between basement rocks and overlying Cenozoic sediments.

Keywords:

• amphibolite
• Atomodesma sandstone
• Dun Mountain–Maitai terrane
• Junction Magnetic Anomaly
• lithic clasts
• metabasites
• phreatomagmatic eruption deposits
• pyroxenite

Introduction

Xenoliths transported to the surface by volcanic eruptions can include samples of the mantle and crust beneath the volcano (e.g. Reay & Sipiera 1987; Saadat & Stern 2012; Bianchini et al. 2013). Such xenoliths provide important information about the origin of the magma and its physical properties and can reveal the ambient geothermal gradient (e.g. Cloetingh et al. 2013). In addition, they may give clues to the processes by which the magma rose to the surface (e.g. Yang et al. 2012; Howarth & Skinner 2013; Moore 2013). Xenoliths of subcrustal material in andesites can provide evidence of subduction processes: for example garnet- and amphibolite-bearing Miocene andesites in Northland, New Zealand, provide evidence for ponding of magmas near the base of the crust prior to eruption (Day et al. 1992; Bach et al. 2012). The time between xenolith entrainment and eruptive quenching can sometimes also be inferred from the extent of time-dependent chemical diffusion between the host melt and the xenolith (Blake et al. 2006; Demouchy et al. 2006).

Country rock-derived lithic clasts from closer to the surface preserved in pyroclastic deposits can provide insight into high-level subsurface processes: for example, Kaawa Formation-derived clasts found in the Barriball Road tuff ring in the South Auckland Volcanic Field revealed that depth of phreatomagmatic fragmentation occurred 170–190 m below the surface (Ilanko 2010; Pittari et al. 2012).

The active monogenetic Auckland Volcanic Field (AVF) presents a distinct risk for the large city of Auckland built upon it (Bebbington & Cronin 2011; Lindsay et al. 2011). However, because the volcanoes of the field were emplaced through a cover of weak Miocene and younger sediments, very little is known about the basement crust below them. Geophysical work (Hatherton & Sibson 1970; Eccles et al. 2005; Williams et al. 2006) indicates that the AVF lies on the Junction Magnetic Anomaly (JMA) that represents a major ophiolitic suture affecting the basement of both the North and South Islands of New Zealand (Fig. 1A). Understanding the effect of such a prominent tectonic feature and of the crust in general on the generation and ascent of AVF magmas is important, and any samples of the underlying crust or mantle brought up by volcanic activity are therefore of great value. The petrography of such samples and their correlation with outcropping units elsewhere in New Zealand can enhance recognition of tectonic units. Structural analysis, together with temperature and pressure determinations on constituent minerals and of veins cutting xenoliths, may shed light on their conditions of formation, deformation, dislodgment and transport. Analysis of xenolith shape and any coatings may provide information on eruption mechanisms. We present descriptions and a new interpretation of country rock-derived lithic clasts found in the phreatomagmatic deposits of two AVF volcanoes (Glover Park and Taylors Hill, both in the suburb of St Heliers; Fig. 2), with special emphasis on samples of metamorphosed basic igneous rocks. Our synthesis has important implications for crustal structure below the AVF and provides a ‘ground truth’ for tectonic units so far only detected by geophysical methods.

Figure 1 Tectonic setting. A, The Auckland Volcanic Field within northern New Zealand. Intraplate volcanic fields labelled in pink. Inset shows relationship to New Zealand tectonic setting. Green line: Dun Mountain–Maitai terrane (M) with the Dun Mountain Belt that generates the Junction Magnetic Anomaly. B, Gravity and magnetic anomalies of the Auckland area, dominated by the effects of the Dun Mountain Belt. Magnetic anomalies after Eccles et al. (2005). Takapuna gravity anomaly after Williams et al. (2006). C, Schematic block diagram of basement structure and terrane subdivision under the Auckland region (not to scale) after Williams et al. (2006). Blue line marks postulated position of in situ Maitai Group sediments. AVF: Auckland Volcanic Field.

Figure 2 The St Heliers Volcanoes. A, Location of the volcanoes (red triangles) within the Auckland Volcanic Field (blue triangles). Yellow rectangles: basement depths after Kenny et al. (2011, 2012) and Edbrooke et al. (1998). MRD: Mount Roskill drill hole, EP: Eden Park drill hole (E Shalev, IESE, University of Auckland, pers. comm. 2013). Green rectangles: basement depths after Davy (2008). t.c.: volcano with xenoliths of terrigenous clastics (greywacke). B, Map of the St Heliers volcanoes, based on figure 6.6 in Searle (1981).

Methods

Our study is based on suite of 275 lithic clasts collected from the pyroclastic deposits associated with Glover Park and Taylors Hill volcanoes of the AVF. The overall sample suite was collected in several phases. An initial suite was partially described by Searle (1959). Subsequent samples from Glover Park were collected by the authors during the period 1999–2004. Some of these were examined by Jones (2007) whose work provided bulk-rock and trace element geochemical data, and electron microprobe (EMP) analyses of minerals in selected mafic clasts.

To supplement these collections, we have carried out additional field work since 2007. All samples from Taylors Hill were collected in the tuff surrounding the volcano (Fig. 2B). Samples from Glover Park were collected primarily from eroded cliffs of tuff at the coast, with the richest concentration of fragments located on the shore platform north of Glover Park (see Fig. 2B, ‘main sampling area’). In this location, clasts eroded out from the tuff cliff have been concentrated by wave action. A full list of the samples with detailed descriptions is given in Spörli & Black (2013). The samples are lodged in the petrology collection of the Earth Science Programme, School of Environment, University of Auckland. For historical reasons, the earlier set of samples collected by Searle has different numbers for rock samples and corresponding thin-sections. In our paper they are prefaced by the word ‘Searle’, for example rock Searle 4997, corresponding to thin section Searle 5204. All the samples collected subsequently have one single AU number for both the rock sample and thin section, for example AU58775.

Collected samples were washed and representative samples selected for thin-sectioning, slabbing and detailed description. Since we have no information about the original orientations of the samples, orientations of structural and other features are described relative to each other in illustrations. Because most of the descriptions are based on more or less two-dimensional surfaces and the orientation of the true movement vectors cannot be determined, we use the two-dimensional term ‘separation’ (e.g. Davies 1984, p. 273; Twiss & Moores 1992, p. 66) for our descriptions of fault offsets.

In the schistose rocks, we label deformation phases recognised from cross-cutting patterns of structures in the standard manner used for metamorphic tectonites (e.g. Hobbs et al. 1976; Passchier & Trouw 1998): S₁– S_n for foliation surfaces, F₁–F_n for folds and L₁–L_n for lineations. Phases consisting of a number of such structural elements are labelled D₁–D_n. Any primary layering (e.g. bedding, flow layering) is labelled S₀. Note that the numbering of phases and structures is specific to each individual specimen.

After detailed thin-section analysis, polished thin sections were made of some critical samples and the mineral phases were analysed to supplement the data of Jones (2007); both sets of microprobe analyses were obtained from a JXA-5A microprobe fitted with a Link systems LZ – 5EDS detector. Some material was also analysed using X-ray diffraction (XRD) to facilitate identification of specific mineral phases. XRD analyses were carried out using a Philips X-ray powder diffractometer using Ni- filtered CuKα radiation and 40 kV and 20 mA operating conditions. Finally, we compared the rocks described in our study with known possible correlative rocks outcropping at the surface and described in the literature.

Geological framework

Within the tectonic framework of the SW Pacific and New Zealand, the development of the Auckland region can be summarised as follows.

1. Accretion of Palaeozoic–Mesozoic terranes (Murihiku, Maitai, Caples and Waipapa) onto eastern Gondwana (Spörli 1978; Adams et al. 2009). Crustal basement units, from west (Tasman Sea side) to east (Pacific Ocean side), are shown in Fig. 1A. The Waipapa greywacke terrane is the only exposed basement in the Auckland region. The Dun Mountain–Maitai terrane, the presumed source of the New Zealand-wide JMA (Eccles et al. 2005) which passes under the AVF (Fig. 1A, C), is only exposed in one locality in the North Island (at the Wairere serpentinite quarry 190 km south of Auckland) as a steeply dipping zone of serpentinite with partially metasomatised blocks of gabbro (O’Brien & Rodgers 1973, 1974).

2. Cessation of subduction, followed by uplift and erosion of the terranes with subsequent deposition of Cenozoic sedimentary sequences and extensional faulting during rifting of New Zealand away from Gondwana from 84 Ma to 52 Ma (Gaina et al. 1998) with consequent reduction of crustal thickness to 29 km.

3. Establishment of a new subduction system to the northeast of the area, initiated by obduction of Cenozoic shelf sequences and underlying/adjacent ocean floor in the form of the Northland Allochthon (Ballance & Spörli 1979; Hayward 1993) during the Late Oligocene and culminating in the formation of a Miocene volcanic arc and the Waitemata turbidite basin in the region.

4. Transfer of the subduction zone to its present location about 400 km to the east, leaving the Auckland/Northland area in an intraplate setting (e.g. Smith 1989) and leading to formation of the basaltic intraplate AVF.

A combination of low seismicity in the Auckland region (which hinders tomography studies) and high urban noise (which hinders analysis of those seismic signals that are recorded) means embarrassingly little is known about the nature of the lithosphere below New Zealand's largest city. The Cenozoic cover sediments generally have low-angle dips and overlie the basement rocks on a sharp unconformity at a depth averaging about 500 m (Williams et al. 2006). They consist of the Late Eocene–Oligocene Te Kūiti Group; the Miocene Waitemata Group; and a complex, irregular assemblage of Late Miocene–Holocene marine, coastal sand and terrestrial sediments (Hayward 1993; Edbrooke et al. 1998; Edbrooke 2001).

The Auckland Volcanic Field

The active AVF consists of c. 50 alkali olivine basalt volcanoes and is very young, with the age of the oldest eruptions dated at c. 250 ka (Molloy et al. 2009; Lindsay et al. 2011). The c. 250 ka span of activity in the AVF covers the last two major ice ages (e.g. Denton et al. 2010). Sea levels were high in the 250–200 ka interval and in the last interglacial (c. 130 ka). A 30 ka flare-up in the AVF took place during a very low sea- level interval just before the termination of the last ice age. A number of volcanoes in the AVF, including Glover Park and Taylors Hill, therefore erupted when sea level was low and the area was well inland from the coast. The present sea coast exposure in the vicinity of the volcanoes in this study (Fig. 2) is entirely due to postglacial sea-level rise.

The AVF is one of a group of intraplate volcanic fields (Fig. 1A) distant from the subduction zone presently active further to the east (Huang et al. 1997). The volcanoes are mostly monogenetic (Smith 1989, 1992; Smith et al. 1993; Huang et al. 1997), although a few appear to have erupted in the same, relatively brief time interval (Cassidy & Locke 2004, 2010; Cassidy 2006). Rangitoto (Fig. 2A), the youngest, most productive volcano, had more than one episode of eruption (McGee et al. 2011; Needham et al. 2011) and may even have been active as far back as 1500 years ago (Shane et al. 2013).

The source of the magma for these volcanoes lies in the aesthenospheric mantle at depths of around 100 km (Huang et al. 1997; Horspool et al. 2006; McGee et al. 2013). The field has an elliptical shape (Spörli & Eastwood 1997), possibly corresponding to a dome-shaped mantle structure located at the northwards-propagating tip of a major lithospheric fracture. Within this ellipse, groups of three or more volcanoes form distinct alignments, indicating smaller scale control by fractures. Most volcanoes in the AVF were produced by phreatomagmatic activity with or without subsequent Hawaiian and/or Strombolian eruptions (Allen & Smith 1994; Kereszturi et al. 2014). Phreatomagmatic eruptions reflect varying degrees of magma–water interaction and produced tuff rings. Shallow-seated controls on explosive basaltic eruptions in the AVF have been described by Houghton et al. (1999) at Crater Hill, where simultaneous eruptions from vents along a NE-striking, 600 m long dyke produced contrasting eruption styles due to controls within the uppermost 80 m of the conduit; by Agustín-Flores et al. (2014) at Maungataketake, where base-surge run- out distance was controlled by eruption through soft sediment substrate of the Tauranga Group; and by Németh et al. (2012) at Ōrākei, where a small-volume magma body nevertheless produced a violent explosion due to subsurface interaction with water.

Xenoliths are generally rare in AVF deposits, although some centres do contain distinct populations. Distinctive white quartz xenoliths with pyroxene reaction rims found in lavas from Mount Eden, Mount Wellington and Māngere Mountain all record high-temperature interaction with the magma. They were interpreted by Searle (1962a, b) as being due to partial assimilation of quartz veins from greywacke-type basement rocks. A xenolith population derived from even greater depths and preserved in tuffs at Pupuke volcano (Fig. 2A) consists only of olivine-rich ultramafic inclusions; these were interpreted to have been formed at 1250 °C and 11 kb pressure in the upper mantle (Brothers 1960; Brothers & Rodgers 1969; Rodgers & Brothers 1969; Rodgers et al. 1975). Very rare greywacke-type xenoliths were also collected by Searle at Mount Wellington and Māngere Mountain (Spörli & Black 2013; for locations see Fig. 2A).

The assemblage of lithic clasts that are the subject of this paper are found in pyroclastic deposits of two geographically closely associated volcanoes, Glover Park and Taylors Hill, both located in St Heliers, one of Auckland's eastern suburbs. The lithic clasts differ significantly from xenoliths found in other AVF volcanoes, both in terms of their nature and abundance and in that, although some may be coated with lava, they show no signs of high-temperature metamorphism at their contacts with the lava. They are also different from populations of non-juvenile lithic clasts in other described phreatomagmatic tuffs of the AVF, which are dominated by Cenozoic sedimentary rocks (e.g. Németh et al. 2012; Agustín-Flores et al. 2014).

The amphibolitic metabasite clasts have geochemical compositions distinct from those of the AVF basalts: a limited amount of geochemical (major and trace element, including rare Earth element) data available (Jones 2007) suggest a range of island-arc tholeiite to mid-ocean ridge basalt (MORB)-type associations with a distinct Nb depletion more characteristic of an ophiolite-type sequence than standard oceanic crust or intraplate associations.

Glover Park volcano (Whakamuhu)

This volcano produced a simple inland phreatomagmatic tuff-forming eruption without any accompanying lava extrusion (Searle 1981; Hayward et al. 2011). The 200– 300 m diameter, slightly east–west elongated explosion crater today underlies the Glover Park sport facilities. Cliff retreat during post-eruption sea-level rise has left the northern rim of the tuff cone perched about 30 m up in a cliff (Fig. 2B) from which the blocks of lithic-bearing tuff have dropped down to the shore platform in the underlying Waitemata Group sediments, where they can be most easily sampled (Searle 1959). Large blocks of ‘basement material’ were first noted by Fox (1902) and attributed to the ‘Matai slates’. The age of the eruption that produced Glover Park volcano is estimated at >45 ka with a large uncertainty, based on Rotoehu tephra (>45 ka) present within 2 m of the surface in drill core from Glover Park (Lindsay et al. 2011).

The Waitemata Group underlying the volcano consists of interbedded turbiditic sandstones and mudstones which have been affected by eastwards-verging folds/thrusts and contain localised submarine slide deposits similar to those described elsewhere in the Auckland area (Spörli & Rowland 2007). The rocks are cut by later steeply dipping extensional faults.

Taylors Hill volcano (Taurere)

In contrast to Glover Park volcano, Taylors Hill first produced a tuff ring which was then occupied by a scoria cone with a number of spatter cones and two late-stage small lava flows (Searle 1981). The best age estimate is 32–34 ka because of an anomalous palaeomagnetic signature (Cassidy & Locke 2004, 2010) similar to that of the better dated Wiri Mountain and Crater Hill volcanoes (Fig. 2A), indicating that they all erupted during the same geomagnetic excursion (Cassidy 2006; Cassata et al. 2008; Lindsay et al. 2011). This would indicate that Taylors Hill is c. 10 ka younger than Glover Park volcano and was perhaps part of the 30 ka volcanic flare-up in the AVF.

Occurrence and nature of lithic clasts

Mode of occurrence

The largest lithic fragments are found in the ‘main sampling area’ (Fig. 2B). One angular block measured 70 cm in largest diameter, confirming the report by Fox (1902) of blocks ‘several feet in diameter’. The inclusions are mostly angular, but there are also some with rounded shapes. The collection of 275 specimens is dominated by around 200 metabasites, at least 148 of which are schistose (Table 1). A total of 260 of the specimens are from Glover Park and 13 are from Taylors Hill (there are also two from Māngere Mountain and Mount Wellington). Most of the metabasite samples are fine grained, but 38 are medium to coarse grained. There are at least three rodingites. Among the 22 non-schistose mafic rocks there are 16 gabbro/diorites, three pyroxenites and several examples of porphyritic volcanic rocks not from the present-day volcanic field. Nineteen samples are entirely cataclasite. Eight samples are of basement metamorphic sedimentary rocks (‘greywackes’), and include one with fossil fragments. Two additional specimens are red chert. Fourteen samples represent Cenozoic sedimentary rocks. A summary of the suite is presented in Table 1.

Table 1 Brief summary of lithology and mineralogy of the country rock-derived lithic clasts ejected with phreatomagmatic tuff at Glover Park and Taylors Hill volcanoes. For more detailed descriptions see Spörli & Black (2013).

Type of non-juvenile lithic clast	Mineralogical and lithological characteristics
Foliated (schistose) metabasites
Amphibolites with varying degrees of foliation	Greenish-grey to black clasts dominated by hornblende; less common calcic pyroxene and albite. Rocks range from finely foliated, almost phyllonitic hornblende schist to coarsely foliated types. Multiple sets of epidote veins common. At least one sample has macroscopic garnet. Veined and cut by cataclastic faults.
Severely metasomatised rocks
Rodingite	Yellowish light-coloured aggregates of quartz and fibrous talc with spectacularly zoned quartz veins. Contains hydrogrossular garnet, in some cases concentrated in layers and seams. Protolith was foliated.
Non-foliated metabasites
Gabbro/diorites	Coarse-grained rocks with little or no development of tectonic foliation. Strongly altered, with acicular actinolite replacing clinopyroxene and amphibole. Veined and cut by cataclastic faults.
Pyroxenites	Coarsely crystalline rocks with clinopyroxene crystals up to c. 3 mm; some replaced by metamorphic actinolite. Pervasive cataclasites.
Porphyritic volcanic rocks	Fine-grained mafic rocks, most porphyritic but some grade to aphanitic textures; veined and cut by cataclastic faults.
Metamorphosed sedimentary rocks
Meta-sandstones (greywackes)	Mostly fine-grained finely bedded volcaniclastic meta-sandstones, some veined with prehnite.
Calcareous meta-sandstone	Calcareous meta-sandstone with numerous (20–50% of rock) angular calcite platelets identified as Atemodesma.
Red chert	Massive with quartz veining.
Fault rocks
Cataclasites	Very conspicuous, homogenous to multi-modal rocks, some heavily charged with opaque material, others more translucent in thin section. Present as individual lithic clasts and as components of other clasts.
Non-metamorphosed sedimentary rocks
Sandstone and mudstone (including Waitemata Group)	Weak lithic sandstones and mudstones. No veining observed.

While the lithic clasts are usually in sharp contact with the surrounding tuff (Fig. 3A–B) some are encased in lava (Fig. 3C–D), fitting a classification as cored bombs or composite clasts after White & Houghton (2006). Fifteen samples in our collection (Spörli & Black 2013) still have a tuff coating preserved, 30 samples are lava-coated and one specimen has both an inner lava coating and an outer tuff coating. Some of the tuff layers from which the clasts were derived also contain abundant juvenile clasts (Fig. 3B). Swarms of very small country rock clasts, down to 1 mm or less in diameter, can occur within the juvenile fragments (Fig. 3D), indicating that the latter are recycled juvenile clasts in the sense of White & Houghton (2006). Some of the micro-clasts are associated with fine bubble layers (Fig. 3D), possibly indicating that they induced incipient magma rupture due to a transition from a viscous to a brittle state (e.g. Cordonnier et al. 2012). In some specimens, magma appears to have infiltrated the country rocks along joints, leading to jigsaw-puzzle patterns of breccia fragments (Fig. 3C). The non-coated fragments were most probably directly blasted from in-place country rock during the explosive phase.

Figure 3 Modes of occurrence of the inclusions; all examples are from the main sampling area below Glover Park volcano (Fig. 2B). A, Fallen block of bedded tuff with dark metabasic rock clasts of two different sizes. B, Fallen block of coarser tuff on the beach. Light-coloured fragments are Waitemata Group sandstones (W); note rounding of some clasts. Dark fragment on lower left is juvenile lava. Hand lens for scale. C, Lava clast with angular fragments of Waitemata Group rocks making a jigsaw pattern. D, Very small lithic clasts associated with bubble trains (arrowed) in a recycled composite lava fragment. Paired yellow arrows show sense of lava flow-shearing along bubble/fracture surfaces. Paper clip is 28 mm long.

Descriptions of key rock types

Here we restrict our descriptions to clasts that have particular significance for the interpretation of the substrate and the volcanic processes of the AVF. Textural and structural analysis is important for assessing the environment of formation, especially of the metabasic rocks. More comprehensive descriptions of samples can be found in the supplementary file (SF) and in Spörli & Black (2013).

Amphibolitic greenish-grey to black metabasic clasts are the most common. The dominant mineral is hornblende; most of these clasts also contain grains of calcic pyroxene. Feldspar, which has a variety of habits (occurring in veins, polygonal aggregates and crystals clouded with tiny epidote grains), is always albite. There is a complete gradation between finely foliated, almost phyllonitic hornblende schist and coarsely foliated types. Crenulations or stretching lineations are visible only on a few samples. Multiple sets of veins are common, with early veins often folded and the younger veins interacting with or postdating cataclastic faults.

Fine-grained amphibolite

A typical thin section of a fine-grained amphibolitic schist (AU58775, Fig. 4A–B) reveals the following main sequence of deformations: (D₁/D₂) formation of the igneous/metamorphic fabric (schist); (D₃) emplacement of epidote veins; (D₄) kink folding; (D₅) cataclastic faulting; and (D₆) further veining, mainly albite.

The main part of the thin section is made up of green, thinly foliated hornblende schist (Fig. 4A–B) with some feldspar-rich laminae. Folds of these laminae (Fig. 4B) indicate that initial S₁ surfaces were isoclinally folded during D₂ so that the axial planes (S₂) of the resulting F₂ folds ended up parallel to S₁. Very locally, S₂ also occurs as a fanning pattern of actinolite crystals in some hinges of isoclinal F₂ folds (Fig. 4B), indicating that D₂ is an important metamorphic event.

Figure 4 Finely foliated metabasite. A, Tracing of a whole thin-section scan. The sequence of structural features is indicated in red. An albite vein cross-cutting cataclasite is shown at a/c. Letter in rectangle locates the micro-photo. B, Micro-photo (plane-polarised light) showing the S₁/S₂ foliation and an F₂ fold. Fsp: feldspar; CC: cataclasite. Green lines and ‘amph’ indicate fanning S₂ amphiboles. C, Photograph of the specimen. ep: epidote; FR?: possible fault repetition of the thick epidote vein. The thin-section face is on the opposite side of the specimen.

Other important features are: thick D₃ epidote veins, some of them parallel, others at high angle to the foliation (Fig. 4A, C); conjugate F₄ kink folds (Fig. 4A, labelled folds); and an intricate network of very thin cataclasite faults (Fig. 4A). Discontinuous chlorite veins either pre-date or (most likely) formed during formation of the cataclasites. D₆ albite veins clearly postdate the cataclasites (e.g. at a/c in Fig. 4A). The rock specimen (Fig. 4C) displays the typical dark colour of these metabasites, the S₁/S₂ foliation and the prominent yellowish, folded epidote veins.

Coarse-grained samples of metabasites display similar features; detailed examples of these are given in Spörli & Black (2013).

Coarse-grained garnet-bearing amphibolite

AU58784 is the only sample in the collection with macroscopically recognisable garnet (Fig. 5), determined by EMP analysis to be andradite. The following sequence of events can be recognised: (D₂–D₄) formation of the igneous/metamorphic rock (for further details see SF Fig. 1); (D₅) feldspar polygonisation; (D₆) formation of talc veins; (D₇) cataclastic faulting; and (D₈) further veining.

Figure 5 Garnet-bearing, coarser metabasite with polygonised feldspar, sample AU58784. The sequence of structural features is indexed in red. A, Tracing of a whole thin-section scan. Location (1) shows reactivation of D₄ top-to-the-left ductile shear by D₇ top-to-the-right cataclastic faulting. B, Micro-photo, plane-polarised light. G: andradite garnet. Note contrast between dusty (saussuritised?) metamorphic feldspars (Fsp) and clear, polygonised (metasomatic?) feldspars (pFsp).

Irregular patches of yellowish-white material (Fig. 5A, SF Fig. 1A) consist of clear polygonised albite feldspar (Fig. 5B). They represent a metasomatic event, postdating the main metamorphic deformations (D₂–D₄).

Rodingite

Except for the red chert (as mentioned in the ‘Low-grade metasedimentary rocks from the basement’ section), the rodingite clasts are the only strongly quartzose samples in the collection (Spörli & Black 2013) and are similar to the metasomatic rodingites that occur in the serpentinites of the Patuki melange of the northern South Island (Coleman 1966). They are distinguished by their splotchy yellowish light-grey colours (SF Fig. 2A). A thin section of sample Searle 4992 (Searle 5134, Fig. 6A) shows that the rock consists of an aggregate of quartz and fibrous talc. Other thin sections of this material display spectacularly zoned quartz veins. Multiple talc/quartz seams in part follow fold-like structures and produce patterns reminiscent of metamorphic transposition of an earlier fabric by a later one (SF Fig. 2B).

Figure 6 Photomicrographs of various rock types. A, Rodingite: seam of garnets (black) cross-cut by quartz veins. Thin section Searle 5253, crossed polarisers. Note talc/quartz material with schistosity-like fabric (red lines), paralleled by quartz veins, in the upper left-hand corner. The rest of the section is more quartz-rich. B, Pyroxenite: thin section Searle 5233 from Glover Park volcano. Photomicrograph (plane -polarised light) showing transverse section of clinopyroxene crystal (Py) disturbed by cataclastic seams. Am: amphibole-rich interstitial material. The location of this photograph is shown in SF Fig. 3. Note that the photograph is rotated c. 90° counter-clockwise relative to SF Fig. 3. C, Portion of thin section (Searle 5275a) of a fossil-bearing basement sandstone from Glover Park volcano in plane-polarised light, showing shell fragments, clinopyroxene heavy minerals and calcitic veins. See the discussion for the important significance of this specimen.

The thin section (Fig. 6A) contains minute dispersed semi-opaque spots of a mineral identified by XRD as hydrogrossular garnet. In some cases these are concentrated in layers and seams. In Fig. 6A such a layer is cross-cut by a clear quartz vein.

Non- or little-deformed metabasites

These rock types range from pyroxenite to gabbro-diorite, but also include porphyritic volcanics (Spörli & Black 2013). Gabbros are some of the coarsest rocks in the collection, with visible crystals up to several millimetres in diameter. Although the minerals can define some layering (e.g. AU62910) there is little development of tectonic foliations. All samples show strong alteration, with the clinopyroxenes and amphiboles replaced by often acicular actinolite. Searle (1959, fig. 4) describes an example of tremolite schist (thin section Searle 5200). Sample AU63059 (Spörli & Black 2013) has well-developed fault striations, but without development of fibrous minerals. Cataclastic faulting is also visible in thin sections. The cataclasites in some cases are postdated by talc veins. It is unclear whether the porphyritic rocks are derived from one or more igneous sources. Some are non-schistose but strongly altered and faulted; others show incipient development of a foliation.

A typical pyroxenite (Fig. 6B) has a coarsely crystalline fabric, dominantly consisting of clinopyroxene crystals up to c. 3 mm in size (also see SF Fig. 3). Metamorphic actinolite occupies the interstices between the crystals but also partially replaces some clinopyroxenes. Talc occurs in veins and irregular masses. The whole rock is intensively disrupted by a rhomboidal network of cataclasites (SF Fig. 3), which at the smallest scale has exploited the cleavage planes of the clinopyroxenes (Fig. 6B).

Low-grade metasedimentary rocks from the basement

Seven specimens of this group of samples are fine-grained, non-schistose, feldspathic, finely bedded ‘greywacke’-type sandstones with or without prehnite veining. However, another sample (rock Searle 5155/thin section Searle 5275a) from Glover Park volcano stands out and is a key specimen in that it contains numerous angular calcite platelets (Fig. 6C). Calcite crystals in the platelets form a prismatic fabric at right angles to the long edges of the fragments, indicating that these are prismatic layers of bivalves. The platelets make up 20–50% of the sandstone and, although very angular, appear to be graded together with the other grains which are mainly feldspar and abundant clinopyroxene. A thin-section tracing (SF Fig. 4) illustrates the structural development of this specimen and maps shell-rich versus shell-poor clastic material. The significance of this specimen is considered in the ‘Discussion’ section. Two further samples are of red chert, and are similar to the red cherts in the zones of ocean-floor rocks that delineate accretionary thrusts in the Mesozoic basement terranes of the northern North Island (Spörli et al. 1989; Spörli & Black 2013).

Interpretation of the Glover Park and Taylors Hill clast suite

Eruptions at Glover Park and Taylors Hill volcanoes have ejected a heterogeneous suite of country rock clasts dominated by mostly foliated (schistose) metabasic rocks. These range from peridotites, gabbros and diorites to amphibolitic schists, and none contain primary quartz. There are also some rodingites. The non- (or less-) foliated metabasites include coarse ultramafics, gabbros and various finer grained volcanic lithologies. All of these display extensive replacement by amphiboles and are veined and cut by cataclastic faults. Additional lithologies are low-grade ‘greywacke’-type metamorphic sedimentary rocks, including one fossil-bearing sandstone; fragments of cataclasites; less deformed volcanic rocks; and Cenozoic non-metamorphic sedimentary fragments.

How many rock suites?

A fundamental question regarding the assemblage of basement clasts is whether they represent a contiguous, single tectonic unit or come from separate locations along the path of the magma ascending to the vents of each of the two volcanoes. Monogenetic volcanoes are generally fed by dykes or plugs (Kiyosugi et al. 2012). In both cases (dyke or plug) it is likely that the horizontal extent of the area (and therefore the geological units) sampled by the ascending magma will be limited in size. In a horizontally layered lithosphere, the magma may nevertheless vertically sample a number of lithological units. However, the tectonic model of the region under the AVF suggests that the tectonic units in the basement are steeply dipping (Fig. 1C), which is likely to restrict the geological variety sampled.

That these two volcanoes ejected amphibolite-grade, part- schistose metabasites together with non-metamorphic sedimentary rocks, many of which can be correlated with the Miocene Waitemata Group, indicates the presence of an unconformity between a deeper, high-grade basement and an upper non-metamorphic sedimentary cover at the source of the clasts.

Towards a common structural history for the foliated metabasites

On the basis of the argument put forward in the previous section, we feel justified in attempting to distill a common history from the deformation phases detected in the individual samples. Foliation development can be subdivided into at least two phases (S₁, S₂) associated with isoclinal folding, development of ductile shears and porphyroclasts. This is followed first by open folding on hinges that are often typically rounded in shape (SF Fig. 1), without generation of schistosity or cleavage, and then by kink folding (Fig. 4A) that in some cases appears to be transitional into the next phase of deformation: pervasive multiphase cataclastic faulting that affects all the samples down to the thin-section scale. Fault rocks include fine-grained, often anastomosing seams of various thicknesses and breccia-textured zones.

Formation of veins started during foliation development and significantly postdates it. It had its peak after kink folding and before cataclastic faulting, but a few veins also postdate the faults. As well as albite veins, which contribute to formation of the schistosity in a few samples (e.g. AU 63001), epidote veins are some of the earliest (e.g. Fig. 4). Prehnite veins then follow (Spörli & Black 2013) and may indicate a step down to lower pressure or a change in fluid chemistry as the rock rises to the surface. Talc veins mostly pre-date, but may also be contemporaneous with, cataclastic faulting. Albite/chlorite veins postdate the cataclastic faulting (e.g. Fig. 4A).

An important result of the micro-structural analysis of the metabasic specimens is that their earliest foliations show textural evidence of having been formed at very high temperatures. Unlike those in lower temperature schists, they do not display segregation differentiation of quartz and/or feldspar along pressure solution cleavage planes during early folding (e.g. Hobbs et al. 1976; Craw 1998). The foliation surfaces in our samples may either represent a primary flow foliation (S₀) formed in a partially molten stage and/or a S₁ foliation due to very high-temperature deformation. This result has important implications for deducing the site of formation of these rocks (see section on ‘Metamorphism and metasomatism’).

Metamorphism and metasomatism

While some of the specimens, for example AU58775 (Fig. 4) and AU58777 (Spörli & Black 2013), only display obvious evidence of dynamic metamorphism others, for example AU58784 (Fig. 5), have been affected by an often complicated combination of dynamic and static recrystallisation. The assemblage clinopyroxene + hornblende + plagioclase is indicative of amphibolite facies; the crystals of albite clouded with epidote common in the metabasites described here are considered to be pseudomorphs of what was once a calcic plagioclase. Metamorphism is therefore mainly amphibolite facies, but can range down to prehnite/pumpellyite facies or lower. On the one hand, clinopyroxenes are irregularly replaced by large amphiboles; on the other, there is tectonically controlled actinolite occupying axial planes and fanning cleavages in F₂ folds (Fig. 4B).

The metabasites show various intensities of metasomatic recrystallisation. In sample AU58784 (Fig. 5) andraditic garnet was probably formed in an early phase of metasomatism predating F₂ folding, whereas the polygonisation of the feldspar postdates all the high-grade deformation but not the cataclastic faulting. Replacement of the mafic minerals in some dark metabasites and in the semi-opaque cataclasites is probably in part a manifestation of metasomatism, as are the grid-patterned feldspar zones and the irregular replacement by granular augite in other specimens (Spörli & Black 2013). In the rodingite of specimen Searle 4992 (Fig. 6A, SF Fig. 2) a schistose metamorphic tectonite has been completely overprinted by metasomatic talc, quartz and hydrogrossular.

The recognition of metasomatism in the metabasite suite leads us to infer that these probably underwent metamorphism in an environment similar to that postulated for the Patuki volcanic rocks of the northern South Island (Sivell & Waterhouse 1984b), with a dominance of hydrothermal alteration under a steep (hundreds of degrees per kilometre) geothermal gradient.

Except for the specimens of red chert, all other metasedimentary rocks in our study are terrigenous volcaniclastic greywackes. Generally, the sandstones are non-calcareous except the shell fragment-bearing specimen from Glover Park volcano (Fig. 6C). Some sandstones are cut by cataclastic fault zones and veins. Pre-fragmentation metamorphism up to prehnite-pumpellyite grade is indicated by the veins and/or recrystallisation of the rock matrix. None show evidence of any cleavage. They did not experience the extensive metasomatic processes seen in the metabasites. We interpret these rocks as being derived from sedimentary units either from within the Dun Mountain–Maitai terrane and/or the terranes to the east. Some of the non-calcareous rocks may be from the Waipapa terrane (Fig. 1A).

Discussion

Correlation with the Dun Mountain–Maitai terrane

The position of the two St Heliers volcanoes in relation to the geophysical anomalies in Auckland (Fig. 1B) and the nature of the assemblage of the ejected metamorphic lithic clasts (pyroxenites, amphibolitic meta-igneous rocks, rodingites plus meta-sediments including one shell fragment-bearing sandstone) leads us to conclude that these clasts were predominantly derived from a source near the eastern border of the ophiolitic part (Dun Mountain Belt and adjacent melanges) of the Dun Mountain–Maitai terrane in a situation similar to that described from the South Island by Landis & Blake (1987).

The discovery of a terrigenous sandstone specimen with abundant shell fragments (Fig. 6C, SF Fig. 4) in Searle's collection was serendipitous. While these fragments do not lend themselves directly to a palaeontological determination, the total aspects of the rock allow a reasonable conclusion about its provenance. There are two major occurrences of shell fragment horizons in the Palaeozoic–Mesozoic stratigraphy of New Zealand: (1) Cretaceous Inoceramus fragments; and (2) Palaeozoic Atomodesma fragments.

We have eliminated the possibility that they are derived from Inoceramus group fossils because these are usually much larger than the 0.2–0. 5 mm range in our specimen (Spörli & Black 2013). Furthermore, there are no Cretaceous sedimentary rocks known in the area of the AVF. Rich accumulations of Atomodesma shell fragments typical for the Permian rocks of some basement terranes in the South Island are a much better match, firstly because they cover the size range in our sample and secondly because they occur in terranes closely associated with the Junction Magnetic Anomaly (Fig. 1A). In the South Island, Atomodesma shell layers occur in the Permian Maitai Group overlying the Dun Mountain ophiolite (Landis 1980; Johnston 1981; Cawood 1986, 1987), that is, to the west and south of the ophiolite (see Fig. 1A); in sedimentary slivers within melanges adjacent to the Dun Mountain ophiolite (Dickins et al. 1986); and in the Caples/Pelorus terrane (Bishop et al. 1976; Turnbull 1980; Dickins et al. 1986) to the east and north of the ophiolites. In the Maitai Group, the Tramway Formation (and equivalents) is the main unit that contains prominent horizons of Atomodesma shell fragments (e.g. Landis 1980; Cawood 1986), but they also occur in the other formations, for example Wooded Peak Formation (Landis 1980) near the base of the group.

The clinopyroxene-dominated mafic mineral suite in our sample (see Fig. 6C) may indicate erosion off a mafic or ultramafic body and would therefore support a derivation from the base of the Maitai Group sediments overlying the Dun Mountain Ophiolite although we note there are also detrital clinopyroxenes in Caples terrane rocks of the South Island (Mortimer 1993).

The Dun Mountain–Maitai terrane therefore appears a good match for our rock suite. However, a remaining problem is that, despite an extensive literature review (Spörli & Black 2013), we have not found any examples where the Dun Mountain Ophiolite displays development of any schistosity in the style and intensity as seen in the clasts described here, although there are localised shear zones with planar fabrics (e.g. Webber et al. 2008) and small slivers of foliated amphibolite in melanges (e.g. Kawachi 1974). However, bodies up to several hundred metres in diameter of schistose epidote amphibolites occur in the Dun Mountain–Maitai terrane melanges of the northern South Island (Sivell & Waterhouse 1984a, fig. 1, 1984 b). Their formation is attributed to submarine hydrothermal metamorphism and deformation at a spreading ridge, in the transition between greenschist and amphibolite facies conditions.

The situation at presently active spreading ridges should therefore provide additional hints about the significance of the deformed and altered metabasites in our rock suite. While fast-spreading ridges are dominated by basic igneous rocks with plutonic textures (MacLeod & Manning 1996), slow-spreading ridges produce less primary igneous material and can therefore preserve larger bodies of deformed metabasites, including schistose rocks (Schroeder & De Pan 2006). On the slow-spreading Mid-Atlantic Ridge, gabbros initially developed high -temperature magmatic foliations (probably similar to S₁ of the metabasic clasts described here) and lineations, postdated by brittle deformation associated with hydrothermal alteration (Gaggero & Cortesogno 1997). Transform faults can also be sites for schistose rocks (Peive et al. 2001). Metasomatic alteration by invading fluids is mentioned for all stages of development of such ocean-floor rocks (e.g. MacLeod & Manning 1996; Schroeder & De Pan 2006), which can range from magmatic conditions to zeolite facies (Blackman et al. 2005). A conclusion that the basic rocks ejected by the St Heliers volcanoes come from a piece of ocean floor generated at a slow-spreading ridge would be in agreement with the interpretation of analogous rocks from the northern end of the South Island (e.g. Sivell & McCulloch 2000).

Implications for eruptive processes

The occurrence of abundant lithic clasts in tuffs of the Glover Park and Taylors Hill volcanoes clearly indicates the excavation of crustal lithologies during phreatomagmatic eruptions. The association of metamorphic basement rocks with non-metamorphic Cenozoic sedimentary rocks including Waitemata Group sandstones, that all appear to have undergone the same interaction with the ascending magma, suggests that they were mined from a region near the basal unconformity of the Cenozoic sediments onto the basement rocks, analogous to the processes proposed by Bryner (1991) for clasts preserved in tuff on Motukorea. The depth to the unconformity is difficult to gauge, but it must be at least 268 m because drill holes to that depth in the vicinity of the two volcanoes did not reach basement (Edbrooke et al. 1998; Kenny et al. 2011, 2012), but could be as deep as 500–600 m as indicated by the Mt Roskill drill hole (MRD in Fig. 2A) and the overall geophysical estimate by Williams et al. (2006). It may be possible to further refine the depth of origin of individual xenoliths by pressure–temperature studies of the latest post-cataclasite veins.

We suggest that the interception of the ascending magma with the base of a groundwater reservoir hosted by the permeable Cenozoic rocks triggered a phreatomagmatic eruption. The following simplified sequence of events appears to have occurred: (1) magma ascent into basement rock and entrainment of some fragments, with some rounding occurring; (2) further ascent of magma, degassing and some solidification; (3) rise through the basal Cenozoic unconformity and intrusion into fractures in the sedimentary rocks; and (4) interaction with the aquifer or wet sediments, leading to an explosive phreatomagmatic eruption excavating larger blocks of basement and overlying sedimentary rocks and rounding some fragments. A schematic cross-section illustrating the inferred subsurface structure beneath Glover Park is shown in Fig. 7.

Figure 7 Schematic cross-section of Glover Park volcano illustrating the lithic-bearing tuff ring and underlying diatreme and country rock. Small angular shapes represent lithic clasts. Not to scale.

While most of the lithic clasts (especially the largest and the smallest) are angular in shape, having mostly been cut out along joints or schistosity, others both from the weak Waitemata sediments as well as from the more resistant metabasic fragments are rounded. This could either be due to abrasion during eruption or be inherited from processes pre-dating inclusion in the magma, as is the case for rounded clasts derived from a Cenozoic basal conglomerate on Motukorea (Bryner 1991).

Because we have not made a study of it, the sedimentology of the tuff deposit at Glover Park volcano will not be discussed here. However, this will be an important aspect for further deciphering the eruption mechanisms of this volcano.

Conclusions

• 1. We describe a suite of lithic clasts from phreatomagmatic deposits of two volcanoes at St Heliers that provide a window into the crust below the AVF. The clasts are dominated by an assemblage of metabasites: mostly amphibolitic schist, but also pyroxenite and gabbro. Also present are rodingite, numerous cataclasites and rare terrigenous low-grade metamorphic sandstones, one of which contains Atomodesma shell fragments. Other clasts are of younger, non-metamorphic clastic rocks, including many derived from the Miocene Waitemata Group underlying the volcanoes.

• 2. The metamorphic clasts display foliations, folding, textures, metasomatism, minerals and associations that are consistent with derivation from a slow-spreading divergent plate boundary and have never resided at great depths.

• 3. The range of metamorphic rock types, their position relative to the JMA and the presence of Atomodesma lead us to correlate the source of these clasts with an eastern part of the Dun Mountain–Maitai terrane as it crops out in the South Island.

• 4. The association of the metamorphic rocks with non-metamorphic sedimentary rocks, all occurring both as lava-coated and non-lava-coated fragments, suggests that they were excavated from near the unconformity (>268 m deep) of Cenozoic sediments overlying a steeply dipping Dun Mountain–Maitai terrane melange unit, possibly from within a north–south- striking fault zone, as indicated by the pervasive cataclasite networks in the metamorphic clasts.

Supplementary file

Supplementary file 1: Excavation of buried Dun Mountain–Maitai terrane ophiolite by volcanoes of the Auckland Volcanic field, New Zealand: additional structural analysis.

Associate Editor: Dr Richard Wysoczanski.

Acknowledgements

Collecting and report-writing by the students in KBS's structural geology/tectonics graduate classes during the years 1999–2004 extended the initial collections of Searle and helped to lay the groundwork for this study. We would like to thank Jennifer Eccles for initial scanning of thin sections and organisation of samples. Elaine Smid, Madison Frank and Isabelle Chaillou brilliantly tackled the complex task of curating the final sample set. Neville Hudson patiently organised access to the Searle collection and the eventual storage of the whole sample set. Pat Browne kindly let KBS use his microscope during this study. Louise Cotterall assisted with improving some of the photographs. Hamish Campbell is thanked for advice on the shell fragment-bearing xenolith. James Scott, Nick Mortimer and Richard Wysoczanski critically reviewed the manuscript and helped to improve its style. The DEVORA project provided financial support.