Pyrometamorphosed Otago Schist xenoliths cause minor contamination of Dunedin Volcanic Group basanite

ABSTRACT

We examine the pyrometamorphic changes and resulting crustal contamination that occurred when fragments of Otago Schist were entrained and pyrometamorphosed in Dunedin Volcanic Group (DVG) basanite. Otago Schist greenschist facies quartz-albite laminae were replaced by fine-grained quartz grains (inverted from β-quartz or tridymite) with orthopyroxene coronas and dacitic-rhyolitic glass. Former muscovite + chlorite + epidote ± garnet segregations were converted to plagioclase (An_30-80), olivine, spinel, cordierite, ilmenite, pigeonite and trachytic to trachy-andesitic glass. Minerals largely formed by mineral replacement and then by crystal nucleation from a quenched liquid as protolith minerals passed their liquidi, with the latter process generating swallow-tailed, skeletal, dendritic and ladder-structured crystal morphologies. Orthopyroxene-, olivine- and plagioclase-liquid geothermometry indicates xenoliths reached at least 940°C. Despite the molten state of the xenoliths, discernible chemical and isotopic (Sr, Nd) contamination of the basanite was inhibited by its enriched trace element budget. The only evidence for crustal contamination comes from basanite Pb isotopes, which have less radiogenic compositions within several cm of the xenoliths. The subtle variation in DVG Sr and Nd isotopes may therefore be due to different mantle source components rather than crustal contamination by the underlying Otago Schist.

KEYWORDS:

• Pyrometamorphism
• crustal contamination
• xenoliths
• Otago Schist
• Dunedin Volcanic Group

Introduction

Rocks struck by lightning, affected by coal combustion or transported as xenoliths in mafic magmas can experience extreme temperatures (T) at low pressure (P) that lead to the pyrometamorphic formation of new phases and textures characteristic of the sanidinite facies (e.g. Graham et al. 1988; Dempster et al. 1999; Del Moro et al. 2011; Grapes 2011; McGee et al. 2015). However, as pyrometamorphism is almost instantaneous on a geological timescale, chemical equilibration across a pyrometamorphic rock mass is inhibited. In this study, we assess the effect of pyrometamorphism on element mobility in quartzofeldspathic Otago Schist xenoliths entrained in basanite from the Dunedin Volcanic Group in Otago, New Zealand (Figure 1), and we examine whether this process could account for the subtle but significant range of Sr-Nd-Pb isotopes seen in Dunedin Volcanic Group basaltic rocks.

Figure 1. Metamorphic grades within the Otago Schist, with the studied sample site (Ram Rock) located in the highest-grade belt of upper greenschist facies. Diagram is modified from Wellnitz et al. (2019).

Geological setting

The Oligocene-Miocene Dunedin Volcanic Group (DVG) occurs in the South Island of New Zealand, which is part of the Zealandia continent. Zealandia is a Late Cretaceous rift fragment of the long-lived Australia-Antarctica-Gondwana subduction margin that is composed of an amalgamation of narrow Cambrian to Cretaceous tectonostratigraphic terranes (Landis and Coombs 1967; Mortimer 2004). The largest exposed terrane is the Rakaia Terrane, which is a large block of mainly quartzofeldspathic rocks extending ∼300 km from south of Dunedin to north of Christchurch (Mortimer 2004) (Figure 1). The contact with the adjacent Caples Terrane is obscured beneath a Jurassic-Cretaceous regionally metamorphosed belt known as the Otago Schist (Mortimer and Roser 1992). The suture between the Caples and Rakaia terranes has been overprinted by the Otago Schist, which has converted Al-poor mudstone and greywacke into pelite and psammite, respectively (Turnbull et al. 2001). Meta-volcanic (greenschist), chert, and ultramafic rock are rare components (Craw 1984; Mortimer 1993). Regional metamorphism of the Otago Schist ranges from prehnite-pumpellyite on the northern and southern flanks to greenschist facies garnet-albite zone in the centre of the belt (Figure 1) (Graham and Mortimer 1992; Mortimer 2000). Peak regional P–T conditions of the exposed rocks are estimated to be 400°C and 0.8 GPa (Mortimer 2000) and were attained during Jurassic and Early Cretaceous times (Adams and Graham 1997; Little et al. 1999; Gray and Foster 2004), although the base of the Otago Schist has been affected by a high-temperature Late Cretaceous granulite facies metamorphic event (Jacob et al. 2017).

The DVG is an ∼7800 km² Oligocene-Miocene (25–9 Ma) intraplate field centred around Dunedin and erupted through the Otago Schist (Figure 1). It mostly comprises small volume alkaline basanites, but compositions extend from basanite to trachytic and phonolitic compositions at evolved centres such as the main Dunedin Volcano edifice (Coombs et al. 1986; Price et al. 2003; Coombs et al. 2008; Scott et al. 2020). The trace element and Sr-Nd-Pb isotopic properties have been summarised in Scott et al. (2020), who used their own data and that from Price et al. (2003), Hoernle et al. (2006), Sprung et al. (2007) and Timm et al. (2010) to show that the analysed volcanic rocks generally have isotopic compositions that are very similar to metasomatised lithospheric mantle peridotite xenoliths (Scott, Hodgkinson, et al. 2014; Scott, Waight, et al. 2014; McCoy-West et al. 2016; Dalton et al. 2017). However, the DVG isotopic data do have a range of values with the ⁸⁷Sr/⁸⁶Sr of some components extending towards slightly more radiogenic values and ¹⁴³Nd/¹⁴⁴Nd extending towards less radiogenic values than the known mantle rocks (e.g. Price and Compston 1973; Scott et al. 2020). These more evolved compositions could be due to either the magmas having been derived from (or interacted with) a mantle source that is not represented in the lithospheric mantle peridotite xenolith suites, or explained by the magmas having been crustally contaminated by Otago Schist. Therefore, the relationship of pyrometamorphosed Otago Schist xenoliths to their DVG magmas is important because it provides an opportunity to examine the effects of crustal contamination on a primitive magma.

Methods

Since pyrometamorphic rocks typically cool rapidly and grain sizes are extremely fine-grained (e.g. Grapes 2011), a scanning electron microscope equipped with an electron dispersal spectrometer (EDS) and an electron backscatter diffraction detector (EBSD) was utilised to characterise the mineralogy, textures, and mineral chemistries. The major element geochemistries of minerals were obtained using a Zeiss Sigma VP FEG Scanning Electron Microscope (SEM) fitted with a HKL INCA Premium Synergy Integrated EDS/EBSD at the Otago Centre for Electron Microscopy at the University of Otago. Operating conditions for EDS were an accelerating voltage of 15 kV, an aperture of 120 µm, and a working distance of 8.5 mm. The electron beam intensity was calibrated using a pure cobalt standard and minerals were standardised using several Smithsonian microbeam standards (Jarosewich et al. 1980). EDS data collection, analysis and standardisation were carried out on AZtec software by Oxford Instruments. EBSD operating conditions were an accelerating voltage of 30 kV, an aperture of 300 µm, a working distance of 30 mm with the stage tilted at 70° and the camera at a distance of 207 mm. Maps were collected using a 2—5 μm step size. EBSD data collection was carried out on AZtec software with data analysis via the HKL Channel 5 tango map program, both by Oxford Instruments.

Whole rock geochemistry was measured by ALS Brisbane. Major elements were measured on a dissolved lithium boron fused disk by inductively coupled plasma (ICP) atomic emission spectrometry, and trace elements were measured by ICP-mass spectrometry.

Strontium and neodymium isotopes were measured using a Nu Instruments NuPlasma HR MC-ICP-MS at the Department of Geological Sciences, University of Cape Town, South Africa. Whole rock sample powders were digested over a 48-h period in a 1:1 concentrated HF-HNO₃ solution at 140°C before being dried down and converted to nitrate. Element separation was carried out using column chromatography following the methods described in Pin et al. (2014). Two columns were used for elution of digested rock powders, with a Sr Spec resin for extraction of Sr and Pb, and a TRU spec resin for REE separation. A sequential elution of the REE using a Ln Spec resin separated Nd from Sm (Pin et al. 2014). Strontium was analysed as a 200 ppb 0.2% HNO₃ solution. The ⁸⁷Sr/⁸⁶Sr data was corrected for mass interference of ⁸⁷Rb using the ⁸⁵Rb signal and the naturally occurring ⁸⁷Rb/⁸⁵Rb. Mass fractionation on the ICP-MS was corrected using an ⁸⁶Sr/⁸⁸Sr value of 0.1194 and the exponential law. Data was normalised to the NIST SRM987 reference material with an ⁸⁷Sr/⁸⁶Sr value of 0.710255 (McArthur 1994). Neodymium was analysed as a 50 ppb 2% HNO₃ solution on a Nu Instruments DSN-100 desolvating nebuliser. Data was corrected for Sm and Ce interferences using the measured signals of ¹⁴⁷Sm and ¹⁴⁰Ce and known isotope ratios of Sm and Ce. Correction for instrumental mass fractionation was carried out using a ¹⁴⁶Nd/¹⁴⁴Nd value of 0.7219 and the exponential law. Normalisation of the corrected ¹⁴³Nd/¹⁴⁴Nd data was carried out using the JNdi-1 reference standard, which has a value of 0.512115 (Tanaka et al. 2000). Pb isotopes were analysed as 50 ppb 2% HNO₃ solutions using Nu Instruments DSN-100 desolvating nebuliser. A NIST SRM997 Tl standard was added to all standards and samples at a ∼10:1 Pb:Tl ratio. NIST SRM981 was analysed as reference standard, with ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb, ²⁰⁶Pb/²⁰⁴Pb results normalised to values of 15.4963, 16.9405 and 36.7219 (Galer and Abouchami 1998). All Pb isotope data were corrected for Hg interference and instrumental mass fractionation using the exponential law and a ²⁰⁵Tl/²⁰³Tl value of 2.3889 (after Woodhead 2002; Baker et al. 2004).

Ram Rock basanite

The intermittently quarried Ram Rock basanite at 45°56′68″S, 170°50′07″E (Figure 1) occurs over an area of ∼0.08 km² and has a K-Ar age of 16.4 ± 0.7 (Coombs et al. 2008). The fan-like columnar jointing (Figure 2A) and lower elevation compared to nearby basanitic flows are interpreted to mean it is a plug. The basanite contains an array of mantle and crustal xenoliths that range up to 15 cm in diameter (Figure 2B). The crustal xenoliths form the basis of the paper; for detailed information on mantle xenoliths from this location and elsewhere in Otago, the reader is directed to Dalton et al. (2017), Li et al. (2018), Liu et al. (2015), McCoy-West et al. (2013, 2015, 2016), Scott, Hodgkinson, et al. (2014), Scott, Waight, et al. (2014), Scott, Brenna, et al. (2016), Scott, Liu, et al. (2016), Scott et al. (2019), and Wang et al. (2016). The fine-grained dark grey basanite contains olivine and augite phenocrysts set in a groundmass of plagioclase, augite and ilmenite. Olivine phenocrysts are commonly 0.2–0.4 mm in size, although larger (∼1 mm) phenocrysts are present. Large olivine phenocrysts have higher Mg contents in the cores (Mg# 83 – Mg# 71, where Mg# = 100*Mg/(Mg + Fe²⁺)) than the rims (Table 1). The smaller phenocrysts have Mg# = 66–72.

Figure 2. A, Ram Rock is a columnar-jointed basanite plug that hosts crust and mantle xenoliths. B, This schist xenolith in the basanite displays dark mafic domains and yellow felsic domains. Purple layers are cordierite. C, Crustal xenolith in outcrop of the basanite. Coin is 26.5 mm. D, Sample of the regional Otago Schist collected from near to the basanite outcrop. E, Outcrop of basanite with a small peridotite xenolith and small crustal xenolith. Coin is 26.5 mm.

Table 1. Representative analyses of mineral geochemistry from Otago Schist and crustal xenolith samples.

Sample #	OU85425_03	OU85425_05	OU85425_13	OU85425_09	OU85425_04.1	OU85425_04.3	OU85424_47.1	OU85424_47.2	OU85423_13	OU85421_09	OU85423_33.1	OU85422_44.3	OU85421_03	OU85423_14.2	OU85422_26.1
Mineral	Plagioclase	Muscovite	Epidote	Chlorite	Garnet	Garnet	Olivine	Olivine	Orthopyroxene	Cordierite	Olivine	Olivine	Spinel	Plagioclase	Plagioclase
					Rim	Core	Rim	Core			Symplectite	Cruciform		Prismatic	Anhedral
	Otago Schist	Otago Schist	Otago Schist	Otago Schist	Otago Schist	Otago Schist	Basanite	Basanite	Xenolith	Xenolith	Xenolith	Xenolith	Xenolith	Xenolith	Xenolith
SiO2	67.85	47.01	38.70	25.27	36.84	36.76	39.81	37.48	48.80	45.19	35.61	34.20	0.00	55.89	48.67
TiO2	0.00	0.31	0.00	0.00	0	0	0.00	0.06	0.70	0.00	0.00	0.00	1.11	0.05	0.00
Al2O3	20.27	30.50	27.99	21.98	21.21	21.07	0.24	0.16	3.59	34.95	0.39	3.03	63.30	28.09	33.17
FeO	0.00	3.42	–	26.52	13.84	12.33	15.67	26.20	29.83	9.21	35.72	41.09	24.28	0.40	0.52
Fe2O3	–	–	9.41	–	0.66	0.62			–	–	–	–	–	–	–
MnO	0.00	0.00	0.64	0.85	13.92	17.28	0.21	0.50	0.87	0.00	0.6	0.99	0.00	0.00	0.00
MgO	0.00	2.29	0.00	12.68	0	0	43.97	35.07	13.92	8.39	26.91	20.18	11.19	0.14	0.00
CaO	0.00	0.00	23.46	0.00	13.59	11.91	0.24	0.41	0.63	0.00	0.17	0.63	0.00	9.60	15.89
Na2O	11.36	0.33	0.00	0.00	0	0	0.12	0.05	0.55	0.00	0.34	0.43	0.00	5.54	2.25
K2O	0.00	14.47	0.00	0.00	0	0	0.09	0.05	0.62	1.78	0.00	0.00	0.00	0.63	0.27
Cr2O3	0.00	0.00	0.00	0.00	0	0	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Total	99.48	93.33	100.20	87.30	100.07	99.97	100.35	99.98	99.51	99.52	99.74	100.55	99.88	100.34	100.77
Oxygen #	8.00	12.00	12.50	10.00	12	12	4.00	4.00	6.00	8.00	4.00	4.00	4.00	8.00	8.00
Si	2.97	3.43	3.02	1.93	2.95	2.96	1.00	1.00	1.91	2.08	1.00	0.98	0.00	2.51	2.21
Ti	0.00	0.02	0.00	0.00	0	0	0.00	0.00	0.02	0.00	0.00	0.00	0.02	0.00	0.00
Al	1.05	2.63	2.57	1.97	2.01	2	0.01	0.01	0.17	1.90	0.01	0.10	1.98	1.49	1.78
Fe^3+	–	–	0.61	–	0.04	0.04			–	–	–	–	–	–	–
Fe	0.00	0.21	–	0.00	0.93	0.83	0.33	0.58	0.98	0.36	0.84	0.98	0.54	0.02	0.02
Mn	0.00	0.00	0.04	1.69	0.95	1.18	0.01	0.01	0.03	0.00	0.01	0.02	0.00	0.00	0.00
Mg	0.00	0.25	0.00	0.05	0	0	1.65	1.39	0.81	0.58	1.12	0.86	0.44	0.01	0.00
Ca	0.00	0.00	1.96	1.44	1.17	1.03	0.01	0.01	0.03	0.00	0.01	0.02	0.00	0.46	0.77
Na	0.97	0.05	0.00	0.00	0	0	0.01	0.00	0.04	0.00	0.02	0.02	0.00	0.48	0.20
K	0.00	1.35	0.00	0.00	0	0	0.00	0.00	0.03	0.10	0.00	0.00	0.00	0.04	0.02
Cr	0.00	0.00	0.00	0.00	0	0	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Total	4.99	7.93	8.20	7.09	8.03	8.03	3.02	3.00	4.02	5.02	3.01	2.98	2.99	5.00	5.00

A whole rock analysis was reported by Scott et al. (2020) and two new basanite analyses are reported here (Table 2). The two new samples were obtained from within 1.1 to 2 cm (3C) and 0.1 to 1 cm (sample 2B) from a schist xenolith (1A). The three basanite samples show virtually no chemical variation (Figures 3 and 4) except for Rb – but the sample closest to the schist (highest Rb) has the least similar Rb abundance. With the exception of SiO₂, Th and U, the basanites tend to have greater element abundances than the schist (Figure 4; Table 2). The three basanite samples have ⁸⁷Sr/⁸⁶Sr_{(16 Ma)} values of 0.70285 to 0.70308 and the same ¹⁴³Nd/¹⁴⁴Nd_{(16 Ma)} (0.51292) (Table 2). ²⁰⁶Pb/²⁰⁴Pb values vary from 19.559 to 19.831, ²⁰⁷Pb/²⁰⁴Pb values are 15.644 to 15.656 and ²⁰⁸Pb/²⁰⁴Pb values of 39.376 to 39.243 (Table 2).

Figure 3. Total alkali versus silica (TAS) classification diagram of Le Maitre et al. (2005). The host lava plots as a basanite (Reay et al. 1991; Scott et al. 2020; this paper). Glass compositions from the different segregations of the xenoliths plots in three clusters that display little overlap. Glasses within the felsic domains have sub-alkaline compositions while glasses from the mafic domains and xenolith-basalt interface display alkaline compositions. The Dunedin Volcanic Group field is based on compiled data from Price et al. (2003), Hoernle et al. (2006), Sprung et al. (2007), Timm et al. (2010) and Scott et al. (2020).

Figure 4. Multi-element plot of measured Ram Rock basanite and an Otago Schist xenolith with two samples of basanite at varying distances from an Otago Schist xenolith.

Table 2. Whole rock geochemistry and Sr-Nd-Pb isotopic analyses for Ram Rock basanite samples and an Otago Schist xenolith.

	Otago Schist	± 2s	Basanite	± 2s	Basanite	± 2s	Basanite
	1A		2B		3C		RAM-1^a
SiO2	68.6		43.7		43.4		42.9
TiO2	0.5		2.52		2.42		2.35
Al2O3	14.2		13.95		14		13.6
Fe2O3t	3.54		13.4		12.9		12.35
MnO	0.08		0.2		0.2		0.19
MgO	2.14		9.93		9.85		9.31
CaO	2.09		10.6		10.25		9.92
Na2O	3.45		3.72		3.9		4.11
K2O	3.1		1.36		1.58		1.83
P2O5	0.09		0.64		0.62		0.61
L.O.I.	2.31		0.65		0.91		0.17
Total	100.1		100.67		100.03		97.34
Ba	348		481		518		502
Rb	77		45		45		51
Sr	461		827		799		845
Th	11.5		7.83		7.85		8.24
U	2.4		2.08		2.03		2.05
Nb	8.9		100		103		95.3
Ta	0.7		5.8		6.5		6.7
La	31.2		56.9		57.2		62.3
Ce	62.7		109		108		115
Pr	7.19		12.15		12.05		12.2
Nd	25.7		45.2		44.9		45.8
Zr	234		317		310		285
Hf	6.1		6.9		6.4		6.2
Sm	4.64		8.51		8.19		8.66
Eu	1.05		2.75		2.64		2.85
Gd	3.63		7.4		6.95		6.56
Tb	0.56		1.08		0.99		1.02
Ga	16.1		20.4		19.8		19.4
V	75		222		220		214
Dy	3.23		5.85		5.59		5.35
Ho	0.6		1.04		1.06		0.97
Er	1.85		2.75		2.64		2.59
Tm	0.29		0.39		0.38		0.37
Y	18.4		29.3		28.6		27.4
Yb	1.77		2.27		2.22		2.13
Lu	0.26		0.37		0.33		0.32
Cr	30		350		320		360
87Rb/86Sr^b	0.48		0.12		0.14		0.18
86Sr/87Sr	0.70848	0.000011	0.7031	0.00001	0.70294	0.000012	0.70289
86Sr/87Sr(16 Ma)	0.70837		0.70308		0.7029		0.70285
147Sm/144Nd^b	0.115		0.112		0.112		0.115
143Nd/144Nd	0.51237	0.000011	0.51292	0.000013	0.51292	0.000011	0.51292
143Nd/144Nd(16 Ma)	0.51236		0.51291		0.51291		0.51291
206Pb/204Pb	18.98	0.0011	19.559	0.0012	19.741	0.001	19.831
207Pb/204Pb	15.664	0.001	15.656	0.0013	15.647	0.0011	15.644
208Pb/204Pb	38.992	0.0034	39.243	0.0038	39.3212	0.0034	39.376

^aData from Scott et al. (2020).

^bEstimated using trace element data and isotopic weights and fractions.

Otago Schist

As a reference frame for the pyrometamorphosed xenoliths described below, samples of un-pyrometamorphosed quartzofeldspathic Otago Schist from near to the basanite have been inspected. The rocks are pale green-grey, have a pervasive foliation and prominent millimetre-scale segregations common to the Otago Schist (Figure 2C). The segregations comprise alternating quartz-albite and phyllosilicate-rich layers (Figure 5A). The quartz-albite layers are typically 0.5–2 mm wide and separate 1–3 mm wide microlithons that are dominated by white mica, chlorite and epidote that are sometimes accompanied by small garnet grains (∼0.1 mm). The albite is end-member albite (Ab₁₀₀An₀Or₀) (Table 1) . The white mica is muscovite and contains 14.5 wt% K₂O, ∼3.4 wt% MgO and ∼2.5 wt% FeO. Epidote has ∼9 wt% Fe₂O₃ and 22.7 wt% CaO. Chlorite is rich in Fe (FeO ∼27 wt%) compared to Mg (MgO ∼12 wt%). Garnet grains have ragged margins and are zoned, with a core to rim variation of Alm_24-33Grs_34-38Sps_39-27Andr_2-2.

Figure 5. A, Photomicrograph in plane polarised light (PPL) of a garnet-bearing sample of Otago Schist collected near Ram Rock. B, Photomicrograph in PPL of a felsic domain in one of the collected Ram Rock crustal xenoliths. Quartz has lobate boundaries and a dark corona. Glass is heterogeneous in colour, ranging from colourless to a dark brown. C, Backscattered electron (BSE) image of skeletal orthopyroxene grains within glass in a felsic layer. D, Photomicrograph in PPL of a mafic domain in one of the collected Ram Rock crustal xenoliths. The mafic segregations are heterogeneous with a variety of minerals that are too small for optical identification. Glass varies from a medium brown to a dark brown/black colour. E, BSE image of plagioclase grains with a rectangular prismatic crystal shape. This morphology of plagioclase is always found within glass in the mafic layers. F, Inverted BSE image of a zoned anhedral plagioclase grain. Zoned plagioclase grains in the studied xenoliths consistently have a Ca-rich core and Na-rich rim. mus = muscovite, chl = chlorite, grt = garnet, qtz = quartz, opx = orthopyroxene, plag = plagioclase, dv glass = devitrified glass.

Pyrometamorphic Otago Schist xenoliths

Like the in-situ outcrop schist, the Otago Schist xenoliths in the Ram Rock plug display prominent segregation layering, except that the former white quartz-albite layers are now yellow-white and the former mica-rich brown bands are now dark grey (Figure 2B). Bands and lenses composed of light purple cordierite occur in several samples (Figure 2B). Calculation of the effective bulk composition of the major domains using mineral modes and mineral chemistries indicates that the yellow-white segregations are very felsic (SiO₂ > 80 wt%), whereas the dark domains are very mafic (SiO₂ < 40 wt%).

The felsic domains comprise lobate polycrystalline quartz grains within a colourless to pale brown glass (Figure 5B). Quartz grains are 1–20 mm and are surrounded by a thin rim composed of very fine-grained (<50 µm) needles of orthopyroxene. Dauphine twins were identified in the quartz grains using EBSD analysis; these twins are related to each other through 180° rotations of the c-axis (Tullis 1970). Orthopyroxene grains can also be found within glass and commonly display skeletal rims or swallow tail crystal terminations (Figure 5C). The orthopyroxene grains are enstatite, varying from 20 to 30 wt% FeO and 13–20 wt% MgO, with pigeonite only identified in a large symplectite (Table 1). The glass is silica and aluminium-rich with 4–5 wt% FeO, 5–7 wt% K₂O and ∼3 wt% Na₂O (Table 3) and has a dacite to rhyolitic composition when plotted on a TAS diagram (Figure 3).

Table 3. Representative geochemical analyses of glasses within the crustal xenoliths. Glass was measured in both the felsic and mafic domains and from the basanite-xenolith interface.

Sample #	OU85422_03.2	OU85422_09.2	OU85423_16.2	OU85420_14	OU85421_04	OU85421_09	OU85421_08.1	OU85423_10.3
Segregation	Felsic	Felsic	Felsic	Felsic	Mafic	Mafic	Mafic	Interface
SiO2	66.31	73.04	68.12	71.12	60.72	65.90	58.79	57.10
TiO2	0.89	0.42	0.62	0.44	0.00	0.99	1.08	0.92
Al2O3	16.69	14.10	15.44	15.77	20.62	15.89	18.67	20.92
FeO	4.77	3.72	5.66	2.63	1.92	3.62	3.01	3.48
MnO	0.00	0.00	0.03	0.00	0.00	0.00	0.00	0.00
MgO	0.92	0.00	0.56	0.19	0.63	0.00	0.53	0.24
CaO	2.13	0.67	0.51	0.86	2.75	0.71	2.30	0.50
Na2O	4.86	1.89	2.79	2.49	5.78	2.41	5.71	6.45
K2O	2.10	3.77	5.08	4.81	6.44	7.71	5.84	9.04
Cr2O3	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Total	98.67	97.61	98.81	98.31	98.86	97.23	95.93	98.65

The mafic segregations have more variable compositions compared to the felsic segregations. Their mineral assemblages comprise combinations of plagioclase, cordierite, olivine, spinel, ilmenite and glass (Figure 5D). Plagioclase is present as euhedral laths, prismatic grains or anhedral grains that vary from 10–50 µm (Figure 5E,F). Feldspar compositions vary from anorthite to oligoclase and anorthoclase, with most grains either being calcic (Ab_15-25An_70-80Or₅) or sodic (Ab_50-60An_40-50Or_2-15) plagioclase (Figure 6; Table 1). Normal zoning and replacement features of Ca-rich plagioclase are present (Figure 5E,F). The purple colouration noted in some hand samples (Figure 2B) is made up of fine 10–20 µm prismatic grains of cordierite within glass (Figure 7A). Cordierite grains contain 7–9 wt% FeO and 7.5–8.5 wt% MgO and have a 38–40 mol% sekaninaite (Fe₂Al₃(AlSi₅O₁₈)) component (Table 1). Olivine occurs mainly in the form of very fine (1–5 µm) symplectic intergrowths with spinel (Figure 7B). Stunning cruciform olivine grains, ∼4 µm in size, are common (Figure 7C). The symplectic olivine grains have a composition of Fo_45-53Fa_47-65Tp₁, whereas the cruciform grains have higher iron contents (Fo₃₁Fa₆₈Tp₁; Table 1). The intergrown spinel is hercynite and has 22–40 wt% FeO and 7–14 wt% MgO, with V₂O₃ reaching up to 1.6 wt%. One very large symplectite (∼150 µm) contains anorthite, spinel, olivine, and ilmenite surrounded by pigeonite (Figure 7D)_. Ilmenite occurs in a lobate morphology and is surrounded by glass. Ilmenite often contains minor MgO (∼2 wt%) and up to 2.6 wt% V₂O₃. An acicular mineral is present within all samples but remains unidentified due to the fine grainsize; this mineral often has a dendritic morphology and may be mullite (Figure 7E). Glass in the mafic section is trachyandesite to trachydacite (Figure 3) and the totals slightly lower than 100% indicate that it is hydrous (Table 3).

Figure 6. Ternary feldspar diagram showing the wide variety of plagioclase compositions in the crustal xenoliths in comparison to the pure albite of the Otago Schist protolith. Plagioclase with a prismatic morphology tends to have an andesine composition while anhedral plagioclase is comparatively enriched in Ca. Considerable overlap in the plagioclase chemistry is present between the different plagioclase morphologies. Feldspar isotherms are from Benisek et al. (2010).

Figure 7. A, BSE image of cordierite in a mafic segregation with angular grain boundaries. Cordierite is typically found with a rectangular to square crystal morphology and hosted in glass. B, BSE image of an olivine-spinel symplectite. C, BSE image of cruciform and laddered olivine grains within devitrified glass. D, BSE image of a large symplectite composed of intergrown anorthite, ilmenite, olivine, and spinel with a rim of larger subhedral pigeonite grains. E, BSE image of an unknown mineral within the xenoliths. This mineral has an acicular morphology and can also be found as dendrites. F, BSE image of acicular SiO₂ grains that have been identified at the xenolith-basanite interface. crd = cordierite, ol = olivine, spl = spinel, dv glass = devitrified glass, pig = pigeonite, ilm = ilmenite, an = anorthite, pl = plagioclase, un-id = unidentified mineral, qtz = quartz, cpx = clinopyroxene.

Discussion

Geothermometry

Since most of the pyrometamorphic minerals appear to have formed from precipitation from a liquid, the chemistries of minerals co-existing with glass can be used to estimate mineral-liquid equilibration conditions. Selected mineral-glass pairs occur within 50 µm of one another and a P = 1 bar is applied to reflect the shallow crystallisation. Using 1 bar or 1 kbar makes very little difference to the results of calculations; for example, the difference is only 31°C for the olivine-liquid geothermometer of Beattie (1993), which has an endemic standard error associated with the equation of 53°C. The error used for the average calculated temperatures is 2SD to reflect the small population sizes used for each geothermometer. Orthopyroxene-liquid thermometry (Beattie 1993, equation 11; Putirka 2008, equation 28a) is based on Mg-Fe-Mn exchange. Euhedral orthopyroxene crystals within the felsic segregations are set within a glass and over half of the samples are in equilibrium with this glass when plotted on a Rhodes diagram (Roeder and Emslie 1970). The pairs that display equilibrium give averages of 1086 ± 73°C and 940 ± 112°C for the two calibrations, respectively (Table 4). The Beattie equation gives higher temperatures, but does not consider the presence of H₂O within the liquid, which may cause overestimation. Plagioclase-liquid geothermometry (Putirka 2005, equation A; Putirka 2008, equation 24a) was used on coexisting phases within the mafic domains and yielded averages of 978 ± 96°C and 942 ± 111°C (Table 4). Since the mafic domains also contain olivine and glass, four olivine-liquid geothermometers were used; equation 10 from Beattie (1993), a H₂O corrected version of the same equation, and equations 2 and 4 from Putirka et al. (2007). Olivine-glass pairs were plotted on a Rhodes diagram to evaluate equilibrium, and those satisfying equilibrium criteria yielded 1044 ± 86°C (for both Beattie equations) and 856 ± 167°C and 987 ± 112°C (Putirka equations), respectively (Table 4).

Table 4. Average values and the standard deviation for geothermometry of selected liquid-mineral pairs.

Geothermometer		n	Average T	2SD	Minimum T	Maximum T
Opx-Liquid	Beattie (1993)	9	1086°C	73°C	1037°C	1134°C
Opx-Liquid	Putirka (2008)	9	940°C	112°C	853°C	996°C
Plag-Liquid	Putirka (2005)	8	978°C	96°C	923°C	1049°C
Plag-Liquid	Putirka (2008)	8	942°C	111°C	886°C	1038°C
Ol-Liquid	Beattie (1993)	4	1044°C	86°C	986°C	1087°C
Ol-Liquid	Beattie (1993)	4	1044°C	86°C	986°C	1087°C
Ol-Liquid	Putirka et al. (2007)	4	856°C	167°C	752°C	927°C
Ol-Liquid	Putirka et al. (2007)	4	987°C	112°C	933°C	1036°C

n = number of mineral-glass pairs used. The number of mineral-glass pairs used was limited by those that displayed equilibrium. Opx = orthopyroxene, Plag = plagioclase, Ol = olivine.

Reactions of minerals

Two stages of mineral reaction have likely occurred in the felsic and mafic domains in the xenoliths following conversion from older greenschist facies assemblages. The first is prograde heating, where the Otago Schist greenschist facies minerals reacted to form new phases such as the olivine-spinel symplectites in the mafic domains or plagioclase co-existing with melt in the felsic domains (Figures 5E, 7B). The occurrence of olivine, orthopyroxene, cordierite, pigeonite with glass indicates formation at sanidinite facies conditions. The second stage of mineral reactions occurred as the xenoliths began to cool and new minerals nucleated from the liquid that is estimated via orthopyroxene-, olivine- and plagioclase-melt equilibria to have been >940°C (Table 4). Examples of this are the cruciform olivine in the mafic layers and the skeletal quench-textured orthopyroxene grains in the felsic layers (Figures 5C, 7C). The bimodal composition of plagioclase may represent these two reaction stages. On heating, plagioclase melts through dissolution of pre-existing grains and growth of new plagioclase is progressively enriched in the anorthite component (Grapes 2011). Ca-rich plagioclase grains in the studied xenoliths are anhedral and likely formed in the first stage of mineral reactions. The evidence for the this may be the calcic cores and sodic rims (Figure 5F). This indicates Ca-rich grains were overgrown by Na-rich plagioclase, and the Na-rich grains have tabular textures and are set within glass and probably crystallised directly from the melt (Figure 5E).

One of the main phases that does not appear to have precipitated from melt is quartz (Figure 5B). Electron backscatter diffraction analysis indicates that these grains are α-quartz, which is the low-T polymorph; however, the presence of a high frequency of dauphine twins (relative frequency of misorientation angles of 0.26; Figure 8) compared to the representative Otago Schist samples (relative frequency of 0.18; Figure 8) indicates that the grains in the pyrometamorphic rock have inverted from β-quartz (or higher) during cooling (Barber and Wenk 1991). The presence of needles of quartz surrounding sub-spherical quartz grains (Figure 7F) is similar to textures that have been interpreted as a result of partial melting of β-quartz and precipitation of tridymite (Clark and Peacor 1992; Holness et al. 2005; Grapes 2011), which occurs at > 870°C (Klein and Philpotts 2013).

Figure 8. Distribution of misorientation angles of analysed quartz in an Otago Schist sample (A) and a crustal xenolith (B). The peak of ∼60° represents the dauphine twins. A higher relative frequency of dauphine twins is present in the crustal xenolith (0.26) in comparison to the Otago Schist (0.18).

The Otago Schist in the area is phyllosilicate-rich, yet neither chlorite nor mica were identified in the xenoliths (Figure 9). Published occurrences of thermal decomposition in pyrometamorphic rocks has predicted that breakdown of intermediate Mg-Fe chlorite – similar to the chlorite present in the Otago schist – occurs at T > 650°C and leads to the formation of olivine, orthopyroxene, cordierite, spinel, magnetite (McOnie et al. 1975; Worden et al. 1987). Magnetite and corundum are absent in the studied xenoliths, but common olivine-spinel symplectites found within the mafic domains could be the result of chlorite breakdown (Figures 7B, 9A). Prismatic cordierite grains within glass are often associated with these symplectites. Muscovite breakdown occurs via dehydroxylation, which converts the hydroxyl groups into H₂O, lowering the solidus within the xenoliths and resulting in melting at T > 900°C (Guggenheim et al. 1987; Gridi-Bennadji et al. 2008). Studies on high temperature muscovite breakdown show the formation of mullite or corundum, often with sanidine (Guggenheim et al. 1987; Rodriguez-Navarro et al. 2003; Gridi-Bennadji et al. 2008; Grapes 2011). The sanidine component is often incorporated within the melt, as K⁺ ions are highly mobile (Devineau et al. 2006; Grapes 2011). Corundum and mullite are both absent in the xenoliths, although acicular grains that are too fine-grained for SEM analysis do resemble mullite (Figure 7E). However, the bulk rock composition may not allow for formation of these minerals as spinel is a common Al-phase instead.

Figure 9. A, Schematic diagram of the mineralogical and textural changes that the regionally metamorphosed Otago Schist underwent during pyrometamorphism. Not to scale. B, P-T-t diagram of regional metamorphism of the Otago Schist around 160–140 Ma, followed by the pyrometamorphic event at 16 Ma (Coombs et al. 2008). Pressure and temperature range of Otago Schist metamorphism from Mortimer (2000). Entrainment of the crustal xenoliths occurred at a shallow depth, with peak metamorphism reached quickly before the xenoliths were erupted and cooled.

While epidote is present in greenschist facies Otago Schist, all epidote has been destroyed in the xenoliths and no obvious replacement textures are evident (Figure 9A). Experimental data on the breakdown of epidote at T > 625°C indicates anorthite is a product, as indicated by increasing Ca content of plagioclase (Apted and Liou 1983). Anorthite, cordierite, and H₂O are the likely products of epidote breakdown when the Otago Schist epidote is plotted on an ACF (alkali-calcium-iron) diagram (from Eskola 1939 in Grapes 2011). Cordierite and anorthite-rich feldspar are common in the xenoliths.

Garnet is a minor component within the Otago Schist. A rare symplectite anorthite, olivine, spinel, and ilmenite in one pyrometamorphosed xenolith is very similar to the breakdown products of garnet documented in pyroxene hornfels facies mafic rocks (Figures 7D, 9A) (Scott et al. 2013). The analysed symplectite has a bulk major element composition similar to Otago Schist garnet with the exception of Mn (Table 5). Misch and Onyeagocha (1976) documented manganese mobility in non-isochemical pyrometamorphic garnet breakdown and suggest that Mn could be mobilised and then hosted in orthopyroxene, olivine, spinel, cordierite and ilmenite. Titanium and Mg were added during symplectite formation, most likely from the surrounding melt. A pigeonite corona surrounds the symplectite, and this may reflect a reaction front and signify interaction with the surrounding melt or minerals.

Table 5. Comparison of major element mineral chemistry of garnets from the analysed Otago Schist of a large anorthite-ilmenite-olivine-spinel-pigeonite symplectite in one of the crustal xenoliths.

	Garnet	Symplectite
	Otago Schist	Xenolith
SiO2	37	40
Al2O3	21	25
TiO2	0.0–0.3	1.3
MgO	0	5
FeO	13–19	16
MnO	8–15	0.12
CaO	12–14	12

Model for the evolution of the xenoliths

The cellular mineral growth textures, particularly skeletal and swallow tail orthopyroxene grains, dendritic morphologies, and cruciform and ladder olivine (Figures 5C, 7C, 7E), have been documented in minerals that crystallise rapidly by quenching of melts (Bryan 1972). Common quench textures include spherulites (Swanson 1977), dendritic mineral shapes (Evans et al. 1994), skeletal grains, swallow-tails and mineral chains (Bryan 1972; Nabelek et al. 1978). These textures have been documented in minerals including magnetite (Long and Wood 1986), olivine (Bryan 1972; Tegner et al. 1993), and plagioclase (Bryan 1972). The presence of these quench textures in the Ram Rock Otago Schist xenoliths leads to the question of how they formed within a basanite that itself does not display evidence of quenching.

The morphology of crystals is a function of the growth rate and the mobility of crystal-forming components in the melt, measured as the diffusion coefficient (Kirkpatrick 1975). Crystal growth theory states that skeletal grains or dendritic morphologies form where compositional gradients and diffusion near the crystal-melt interface impact the crystal growth, causing the crystal to break up into ‘cells’ (Kirkpatrick 1975). Large undercooling typically results in spherulitic crystal growth, then dendritic, followed by skeletal grains to finally form tabular grains as the degree of undercooling decreases (Lofgren 1974; Swanson 1977). Experimental work has shown that diffusion-controlled growth results in the formation of swallow-tail morphology in plagioclase crystals (Nabelek et al. 1978). Diffusion controlled growth becomes the dominant form over interface-controlled growth as supersaturation increases.

Due to the different chemistry and mineralogy of the schist xenoliths compared to the basanite, the thermal properties and viscosity of the schist xenoliths will vary from that of the basanite. The mode of crystal formation in the xenoliths must have been different than in the magma, as shown by the variety of mineral textures present (Figures 5C–F, 7A–F). Limited chemical exchange between the mafic and felsic domains suggests that element mobility was extremely restricted. The xenolith quench textures may therefore have formed due to a dominance of diffusion-controlled growth and the limited mobility of elements. Nucleation dynamics were also different in the xenolith in comparison to the basanite, with a finer grain size and a higher number of crystals. The higher number of crystals will have limited the size of crystal growth during the short time of entrainment.

Crustal contamination

Although the textures show that Otago Schist xenoliths were almost entirely composed of liquids whilst in the host magma, the whole rock trace element data show little evidence for contamination of the surrounding basanite (Figure 4). Simple assimilation (A) – percentage of melt remaining (F) – crystallisation (C) models, assuming trace element fractionation of only olivine and clinopyroxene, using crystal-liquid partition coefficients for Sr, Nd and Pb from the GERM database and a r (mass of assimilation compared to mass of crystallisation) of 0.2, also indicate that the basanite Sr and Nd isotopes display negligible isotopic change adjacent to the xenolith. This is because the Otago Schist has roughly half the Sr (845 ppm) and Nd (25.7 ppm) of the basanite (Figure 10A; Table 2) and so, for example, changing the Sr isotope ratio from ∼0.7029 to ∼0.7035 in the basanite via crustal contamination would require the isotopically evolved component to have crystallised from a liquid with F of ∼0.1 (e.g. 90% already crystallised) (Figure 10A). This is unrealistic because for values of F > 0.5, the remaining melt would no longer be basaltic. In contrast to Sr and Nd, however, the average Pb concentration for DVG basanite (3.7 ± 5 ppm; Hoernle et al. 2006; Timm et al. 2010) is lower than the Otago Schist (18 ± 4 ppm; Scanlan et al. 2018; Scott et al. 2020). This means that basanite Pb isotopes should be more susceptible than Sr and Nd to modification by small amounts of Otago Schist contamination. Figure 10B,C shows that the basanite ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb and ²⁰⁸Pb/²⁰⁴Pb isotope ratios plot close to or along the AFC arrays, and that this contamination can be achieved before significant volumes of melt crystallised. The Pb analyses therefore show that there are haloes of crustally contaminated basanite extending several cm from the xenoliths, but the enriched trace element character of the basanite in most other elements obscures this contamination. Furthermore, since mantle xenolith-bearing alkali basalts are estimated to rise through the crust at rates of 0.01–10 m/s (Spera 1984; Peslier and Luhr 2006), the process of the Ram Rock basanite ascending rapidly, entraining and then rapidly quenching them should have also inhibited xenolith-magma interaction.

Figure 10. A, Plot of ¹⁴³Nd/¹⁴⁴Nd versus ⁸⁷Sr/⁸⁶Sr shows little isotopic change in the basanite samples within several cm from a crustal xenolith (2B, 3C). B, ²⁰⁷Pb/²⁰⁴Pb versus ²⁰⁶Pb/²⁰⁴Pb reveals variation in the basanite adjacent to the crustal xenolith, with the highest isotopic change occurring in the sample taken from closest to the xenolith. C, A similar trend to 10B is visible on a plot of ²⁰⁸Pb/²⁰⁴Pb against ²⁰⁶Pb/²⁰⁴Pb. The arrays represent the modelled fraction of melt remaining as the isotopic compositions evolve. The DVG basalts are from Scott et al. (2020).

The implication of the AFC modelling is that the isotopic spread that occurs in the DVG basalts (Figure 10), especially in the Sr and Nd isotopes, is unable to be explained alone by crustal contamination by Otago Schist greyschist. Exchanging the contaminant of the basalts with greenschist, which is the next major component of the Otago Schist, also fails to explain the variable compositions of some DVG basalts since the greenschists have unradiogenic ⁸⁷Sr/⁸⁶Sr (∼0.70498), radiogenic ¹⁴³Nd/¹⁴⁴Nd (0.51270) and unradiogenic Pb (²⁰⁶Pb/²⁰⁴Pb = 18.769; ²⁰⁷Pb/²⁰⁴Pb = 15.627; ²⁰⁸Pb/²⁰⁴Pb = 38.676) (Scanlan 2018). This means that the isotopic variation in the DVG basalts is either due to post-crystallisation processes (e.g. by filling of voids by crustally Sr-enriched fluids) or that mantle source to the magmas was subtly heterogeneous. We prefer the latter explanation, as many of the isotopically investigated basalts do not have large amounts of alteration.

Conclusions

Quartzofeldspathic schist xenoliths entrained within a Miocene Dunedin Volcanic Group basanite magma display pyrometamorphic phase assemblages that are constrained by the pre-existing metamorphic segregations in their Otago Schist protolith. Mineral-melt equilibria and feldspar compositions indicate that temperatures achieved during pyrometamorphism were >940°C. The effective reacting domains were either extremely felsic or extremely mafic, with rhyolite-dacite glass in the felsic layers and the trachyte-trachyandesite glasses in the mafic layers having formed as different minerals passed their thermal maximums and then dissolved other components. A combination of the elevated trace element composition of the basanite compared to the crustal xenoliths, coupled with rapid quenching of the partially melted xenoliths, inhibited chemical exchange. The host basanite largely retains a primitive Sr and Nd isotopic composition but Pb isotopes reveal evidence for crustal contamination within several cm of the Otago Schist xenoliths. Nonetheless, crustal contamination cannot account for the range in Sr, Nd and Pb isotopes in the Dunedin Volcanic Group basalts, which means that this volcanic province is probably derived from several subtly different isotopic reservoirs.

Acknowledgments

We thank Brent Pooley for making the polished sections and Kat Lilly and Marianne Negrini for SEM assistance with analyses. Dave Prior helped identification and interpretation of the dauphine twins. We thank Alex McCoy-West and Christian Timm for their insightful reviews.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors confirm that the data supporting the findings of this study are available within the article.