Composition and Miocene deformation of the lithospheric mantle adjacent to the Marlborough Fault System in North Canterbury

ABSTRACT

Mantle xenoliths in a nephelinite in the Little Lottery River provide insight into the Miocene mantle adjacent to the Australia-Pacific plate boundary in N Canterbury. The xenoliths comprise peridotite and pyroxenite extracted from depths of ∼ 40 to 60 km. The olivine Mg# < 89, a lack of spinel and occurrence of ilmenite, elevated bulk Cr and Al, light rare Earth element (REE)-enriched clinopyroxene, and ⁸⁷Sr/⁸⁶Sr_{15 Ma} clinopyroxene populations (0.7029–0.7047) in most harzburgite and dunite samples, and in all lherzolite, wehrlite, olivine websterite and orthopyroxenites, indicate that most are the product of reaction of depleted peridotite with Fe, Ti, Al and light REE-bearing silicate melts of basaltic or similar composition. Available isotopic data indicate that the xenoliths could be derived from mantle lithosphere beneath the Hikurangi Plateau, in which case the reactive melts may have been associated with the Cretaceous plateau formation. Extensive recrystallisation of olivine indicates that portions of this lithosphere mantle were affected by the deformation close to the time of entrainment. Since the eruption occurred at ∼ 15 Ma, these textures require that the mantle lithosphere in this area was deforming before formation of today's expression of the proximal Hope Fault strand of the Australia-New Zealand plate boundary.

KEYWORDS:

• Hope Fault
• Marlborough Fault System
• mantle xenolith
• peridotite
• pyroxenite

Introduction

Peridotite and pyroxenite are coarse-grained ultramafic rocks that dominate Earth’s upper mantle. Since mantle xenoliths lock in the mantle composition at the time of their entrainment, their investigation can provide direct insight into mineralogy, temperature, buoyancy, deformation and age of the middle to lower lithosphere (e.g. Griffin et al. 1988; Carlson et al. 2005; Pearson et al. 2014, 2021). New Zealand has over 70 known mantle xenolith locations (Scott (2020) and references therein) and these give a snapshot of the composition of the lithospheric mantle underpinning the continental crust portion of Zealandia. However, the Canterbury region is relatively unstudied, with the only known mantle xenolith locations being the little-studied Le Bons Bay basanite plug in Akaroa and in the Little Lottery Intrusives in NW Canterbury (Figure 1) (Coote 1987; Sewell et al. 1993; McCoy-West et al. 2013, 2015, 2016).

Figure 1. Location map showing the mantle xenolith localities and major faults within the South Island. Figure adapted from Scott (2020).

Plate boundaries propagate into the mantle (e.g. Titus et al. 2007; Vauchez et al. 2012; Kidder et al. 2021), and the Little Lottery River nephelinite is less than 1 km from the Hope Fault – the major dextral strike-slip southern strand of the Australia-Pacific plate boundary Marlborough Fault System. Although the Hope Fault strand is generally thought to only have been active for the last 1–2 Ma (Langridge and Berryman 2005), the greater Marlborough Fault System has been active for at least the last ∼ 10 Ma (Little and Jones 1998; Ghisetti 2022). Here, we use mantle xenoliths to show the composition and evolution of the mantle beneath the Little Lottery River site at the time of eruption (∼15 Ma). The xenoliths show that the mantle was deforming in this area millions of years before the current expression of the Hope Fault formed.

Geological background

New Zealand and its surrounding islands represent the surface expression of Zealandia, a 4.9 M km² largely (94%) submarine continent (Figure 1) (Mortimer et al. 2017). Over 85 million years ago, Zealandia, Australia and Antarctica formed the paleo-Pacific margin of Gondwana (Mortimer 2004). Zealandia’s basement rocks consist of predominantly Cambrian-Cretaceous age volcano-sedimentary terranes that were accreted upon or intruded into this subduction margin (MacKinnon 1983; Mortimer 2004; Crampton et al. 2019). Today the terranes are offset by the Alpine Fault and Marlborough Fault System (Figure 1). Broadly, these terranes belong to the Western Province or Eastern Province, which are stitched by the dominantly plutonic Median Batholith and related fault zones (Mortimer et al. 1999; Tulloch and Kimbrough 2003; Scott 2013). The Western Province contains Cambrian to Early Palaeozoic age rocks of the Buller and Takaka Terranes (e.g. Coote 1987; Münker and Cooper 1999) that have been intruded by Late Devonian to Early Cretaceous age igneous rocks (e.g. Waight et al. 1999; Muir et al. 1996, 1997; Tulloch and Kimbrough 2003; Tulloch et al. 2009; Turnbull et al. 2016). In contrast, the younger Late Palaeozoic–Mesozoic age rocks of the Eastern Province form the Cretaceous forearc and are dominated by metamorphosed marine sedimentary rocks (e.g. MacKinnon 1983; Adams et al. 2002; Mortimer 2004; Wandres et al. 2005; Adams et al. 2007) intruded by scattered Cretaceous and younger intraplate igneous rocks (Grapes 1975; Baker et al. 1994; McCoy-West et al., 2010; Mortimer and Scott 2020) aside from where the province forms part of the Median Batholith (e.g. Mortimer et al. 1999; Scott 2013; and references therein).

The Early Cretaceous Pahau Terrane is a dominantly (meta-)sedimentary component to the Eastern Province (MacKinnon 1983; Wandres et al. 2004; Adams et al. 2009) and makes up the ‘basement’ rock of the field area (Figure 1). The rocks are broadly categorised into four main protolith lithofacies; mudstone, interbedded mudstone-sandstone, sandstone, and conglomerate, all suggested to have formed in a shallow-water fan delta system in the accretionary prism on the Gondwana margin (Bassett and Orlowski 2004; Wandres et al. 2004). The long-lived subduction system on the margin of Gondwana was terminated when the oceanic ∼120 Ma Hikurangi Plateau (Figure 1) collided with the Zealandia at ∼105 Ma (Jacob et al. 2017).

The Pahau Terrane in the area of study is dissected by the Marlborough Fault System (e.g. Little and Jones 1998; Wilson et al. 2004; Wannamaker et al. 2009; Ghisetti 2022). This network of faults connects the oblique dextral-strike slip Alpine Fault that runs along the western side of the South Island with the Hikurangi subduction margin running offshore of the east of the North Island (Norris et al. 1990). The Marlborough Fault System is approximately 100 km wide and consists today of four prominent dextral-slip faults; the Wairau, Awatere, Clarence and Hope faults (Little and Jones 1998). The Hope Fault runs approximately 1 km to the northeast of the Little Lottery field site. The Marlborough Fault system migrated southwards during the Negoene, with the Wairau Fault forming at ∼10 Ma (Ghisetti 2022), followed by the Awatere Fault at ∼ 6 Ma (Little and Jones 1998), the Clarence Fault at ∼ 3 Ma (Browne 1992) and then the Hope fault at 1–2 Ma (Langridge and Berryman 2005). The initiation of lateral movement on the Hope Fault, which is relevant to this study, is calculated from present-day slip rate that is assumed to have remained constant over the 20–25 km of lateral displacement. The age of the initiation could be older than 1–2 Ma if there was an earlier component of vertical movement on this structure.

Methods

A diamond-tipped rock drill was used where the mantle xenoliths could not be easily separated from the very tough host rock with a sledge hammer. Polished microscope slides or epoxy resin pucks of the samples were analysed using a Zeiss Sigma field emission scanning electron microscope fitted with an XMax 20 silicon drift EDX detector at the Otago Micro and Nanoscale Imaging facility at the University of Otago. The beam was calibrated against cobalt metal and analyses were standardised against Smithsonian microbeam mineral standards. The precision and accuracy are consistently within 2% of the standard values. The instrument was operated with an accelerating voltage of 15 kV, an aperture of 120 microns and a working distance of 8.5 mm. Analyses were recorded using the Oxford Instruments Aztec software.

In situ trace elements were obtained by laser ablation inductively coupled mass spectrometry at the Centre for Trace Element Analysis at the University of Otago. The samples were examined using an Agilent 7900 Quadrupole mass spectrometer coupled to a Resonetics RESOlution M-50-LR excimer 193 nm laser ablation system, running at 10 Hz with a beam diameter of 75 μm. Glass standards NIST 610 and NIST 612 were measured every 10 analyses to correct for instrumental drift. Data were processed using the Iolite software (Version 3; Paton et al. 2011) where the concentrations of trace elements were calculated through normalisation of the elemental count rates of Si (orthopyroxene) and Ca (clinopyroxene) in the sample and electron microbeam-measured SiO₂ and CaO concentrations. The lower limit of detection was treated as the background-subtracted value within 3 standard deviations of the average background counts. For the quantification level, the only data used are those outside of 7 standard deviations of the average background-subtracted counts.

⁸⁷Sr/⁸⁶Sr isotope ratios of clinopyroxene were analysed using the same laser ablation system coupled to a multiple-collector inductively coupled plasma mass spectrometer MC-ICP-MS at the Centre for Trace Element Analysis following the methods and standards documented in Scanlan et al. (2018) Data were collected as transect lines for 20–40 s at 10 Hz with a beam diameter of ∼90 μm. Data reduction was accomplished using an in-house Excel-based spreadsheet. Data were normalised to the marine carbonate Tridacna clam (⁸⁷Sr/⁸⁶Sr = 0.709176; Neymark et al. 2014), with the results monitored using an in-house feldspar (KAN; n = 13, ⁸⁷Sr/⁸⁶Sr = 0.70290 ± 0.00006) and a clinopyroxene (JGG; n = 7, ⁸⁷Sr/⁸⁶Sr = 0.70472 ± 0.00010) as secondary standards. These both gave results that overlap with published values (KAN ⁸⁷Sr/⁸⁶Sr = 0.70290 ± 0.00002 (Burgin et al. 2023); JGG = ⁸⁷Sr/⁸⁶Sr = 0.70494 ± 0.00031 (Zhao et al. 2020)).

Six samples were selected for whole-rock geochemistry. Preparation began with cutting samples into small pieces with a diamond saw and removal of any visible alteration. Samples were sanded with carborundum and washed to remove any sawblade contaminants. The samples were sealed into plastic bags and smashed into pieces less than 1 cm in diameter with a hammer, and then crushed in an agate rock mill with pure quartz sand analysed between samples. The rock powders were analysed via inductively coupled plasma-atomic emission spectroscopy or by coupled plasma-atomic mass spectroscopy at ALS Minerals in Brisbane, Australia. Loss on ignition (LOI) was calculated by weighing one gram of the sample before and after being analysed at 1000^◦C for one hour.

Results

The Little Lottery River field site is located within the Amuri Range, approximately 17 km east of Hanmer Springs in North Canterbury (Figure 1). The area investigated was an ∼400 m stretch of the river and the surrounding outcrops centred on 42°31’58.34”S and 173°2’10.36”E. The country rock comprises Oligocene Dog Hill Volcanics, which is part of the poorly dated intraplate Cookson Volcanic Group. The Dog Hill Volcanics unit is composed of red-weathering lava, tuff breccia and massive red lapilli tuff (Coote 1987). The Dog Hill Volcanics is intruded by alkaline mafic dikes and plugs of the Little Lottery Intrusives (Coote 1987).

The Little Lottery Intrusives are a suite of alkaline intraplate intrusions that, at the field site, occur as large grey columnar-jointed nephelinite boulders within the river and as scree deposits on the hillside (Figure 2A). The samples in the river are typically very fresh and comprise olivine microphenocrysts set in a sideromelane groundmass with clinopyroxene and Ti-oxides. Coote (1987) has an unpublished K-Ar analysis of 15.1 ±  0.7 Ma on a glass-bearing sample from this location. The nephelinite boulders and scree contain abundant ultramafic xenoliths (Table 1; Figure 2B). When plotted on the ternary olivine-orthopyroxene-clinopyroxene ultramafic rock classification (Le Bas and Streckeisen 1991), 23 of the 40 analysed samples are classified as peridotite and 17 are pyroxenite, although there is gradation between groups (Figure 3). Of these, the peridotites comprise dunite, harzburgite, lherzolite and wehrlite. Most pyroxenites have an olivine-websterite composition, although four samples are orthopyroxenite.

Figure 2. A. Boulder of the Little Lottery Intrusives nephelinite displaying columnar jointing within the Little Lottery River. B. Four xenoliths entrained within the nephelinite. The scale is approximately 12 cm long.

Figure 3. The modal abundance of the Little Lottery River xenoliths plotted on an ultramafic rock classification, following Le Bas and Streckeisen (1991). Each sample is based on 500 points.

Table 1. Modal and textural summary of the Little Lottery River xenoliths.

				Temperature calculations
Sample	Modal classification^a	Texture	Olivine Mg#	Taylor
LLR-15	Dunite	Porphyroclastic
LLR-19	Dunite	Porphyroclastic	83.5
LLR-27	Dunite	Equigranular	86.4	985	±67.9
LLR-34	Dunite	Porphyroclastic	83.1
LLR-38	Dunite	Porphyroclastic	83.2
LLR-50	Dunite	Protogranular	92.2
LLR-9	Harzburgite	Protogranular	88.3
LLR-12	Harzburgite	Protogranular	89.6	±38.6
LLR-18	Lherzolite	Protogranular	86.6	1045	±1.4
LLR-29	Harzburgite	Protogranular	90.6	1010	±114.6
LLR-33	Harzburgite	Porphyroclastic	90.9	1025	±19.7
LLR-31	Lherzolite	Protogranular	89.5	1016	±53.7
LLR-5	Lherzolite	Porphyroclastic	83.2
LLR-10	Lherzolite	Porphyroclastic		1016	±2.8
LLR-16	Lherzolite	Porphyroclastic
LLR-17	Lherzolite	Protogranular
LLR-22	Lherzolite	Porphyroclastic	83	967	±72.9
LLR-25	Lherzolite	Porphyroclastic	83.6
LLR-28	Lherzolite	Protogranular
LLR-37	Lherzolite	Protogranular
LLR-42	Lherzolite	Porphyroclastic	83.6
LLR-47	Lherzolite	Protogranular
LLR-41	Wehrlite	Equigranular	84.5
LLR-1	Olivine-websterite	Porphyroclastic	85.5	1041	±90.1
LLR-3	Olivine-websterite	Protogranular		1046	±9.5
LLR-6	Olivine-websterite	Porphyroclastic	75.6	1060	±33.0
LLR-8	Olivine-websterite	Porphyroclastic	86.1
LLR-11	Olivine-websterite	Porphyroclastic	86.5	1079	± 22.1
LLR-14	Olivine-websterite	Protogranular
LLR-20	Olivine-websterite	Porphyroclastic	83.9
LLR-21	Olivine-websterite	Protogranular
LLR-24	Olivine-websterite	Porphyroclastic
LLR-26	Olivine-websterite	Equigranular
LLR-40	Olivine-websterite	Porphyroclastic	83.8	1086	±50.9
LLR-43	Olivine-websterite	Equigranular	84.1	1061	±43.8
LLR-45	Olivine-websterite	Porphyroclastic
LLR-7	Orthopyroxenite	Porphyroclastic		1042	±5.8
LLR-13	Orthopyroxenite	Equigranular
LLR-23	Orthopyroxenite	Equigranular
LLR-30	Orthopyroxenite	Protogranular

^a Based on 500 points per sample.

The peridotite and pyroxenite samples, although different in mineral modes, cannot be distinguished texturally. Half of the xenoliths are either equigranular (6) or protogranular (14), with the remaining having porphyroclastic textures (Figure 4A, B, C). Eight of the xenoliths are strongly deformed; this includes the five dunite (LLR-15, LLR-19, LLR-27, LLR-34 and LLR-38) and three olivine-websterite samples (LLR-6, LLR-26 and LLR-42). In each of these, olivine is found as fine-grained aggregates, with clinopyroxene and orthopyroxene variably recrystallised (Figure 4B, C).

Figure 4. A. Undeformed dunite xenolith with main phases annotated: olivine (ol), orthopyroxene (opx). LLR-27. B, C. Deformed peridotites, with extensively recrystallised olivine. In these samples, clinopyroxene (cpx) occurs as dusty grains. LLR-42, LLR-15.

Peridotite xenoliths

Olivine within the peridotites is magnesium-rich but the sample suite covers a large range in Mg# ( = 100*Mg/Mg + Fe) (all mineral chemistry is in the Data Supplementary), with the majority below the typical primitive upper mantle peridotite composition (89.2; Ionov and Hofmann 2007) (Figure 5A). The lherzolite samples have the tightest olivine Mg# cluster (82.2–83.9), whereas the harzburgite xenoliths range from 86.6 to 91.2 and dunite xenoliths range from 82.6 to 92.3. The wehrlite sample had an average olivine Mg# of 84.5. Orthopyroxene is typically magnesium-rich with average end-member compositions of Wo₂En₈₇Fs₁₁ and therefore is enstatite. Clinopyroxene has an average end-member diopside composition of Wo₄₅En₅₀Fs₅. Six harzburgite samples contain Cr-spinel butthis phase is absent in all other samples (Table 1). However, even when present, the spinel is of small abundance (<1%). Spinel Cr# ( = 100*Cr/(Cr + Al)) ranges from 25.1 to 55.6 and the samples trend towards refractory compositions. Furthermore, the absence of spinel correlates with lower olivine Mg# (Figure 5B). Ilmenite is a minor component of the spinel-free peridotite samples (<1%), but only occurs in those with olivine Mg# < 89.6.

Figure 5. A. Comparison of the olivine Mg# from peridotite (lherzolite, harzburgite, dunite, wehrlite) and pyroxenite (olivine-websterite). Numbers below samples represent sample numbers. PUM = primitive upper mantle value (89.2; Ionov and Hofmann 2007). B. Spinel Cr# versus olivine Mg# plot for the studied xenoliths compared to spinel in both hydrous and anhydrous melting conditions. Blue and green triangles indicate data from McCoy-West et al. (2013). Figure adapted from Scott et al. (2019a).

Pyroxenite xenoliths

Olivine-websterite xenoliths display olivine Mg# ranging from 83.0 to 87.1 (Figure 5A). One exception is LLR-6, with an extremely low olivine Mg# of 74.4. All the olivine-websterite Mg# sit below primitive upper mantle values (Figure 5A). Orthopyroxene within the pyroxenite samples is typically magnesium-rich enstatite with an average end-member composition of Wo₂En₈₄Fs₁₄, similar to the peridotites. Orthopyroxene comprises 39% to 93% of the total modal abundance of the pyroxenite samples. Clinopyroxenes have an average end-member composition of Wo₄₃En₄₉Fs₉ and are diopside (Data Supplementary). These are very similar to the peridotite clinopyroxene compositions.

Xenolith extraction temperatures and pressures

Analyses of multiple proximal clinopyroxene and orthopyroxene grains were used within calculations set out by Taylor (1998) to assess temperatures of equilibration. A pressure of 15 kbar was assumed and although this is not accurate,changing the pressure by 5 kbar to represent the range of fertile spinel facies lithospheric mantle yields a difference of less than 20 °C. The calculated temperatures of the mantle xenoliths are 900 to 1080°C (Table 1). The South Island geotherm beneath Zealandia was approximately 70 mW m⁻² in the Oligocene-Miocene (Scott et al. 2014b). Using this approximate geotherm, the temperatures project onto extraction depths equivalent to pressures of 13–15 kbar, which equates to the middle to lower mantle lithosphere (Figure 6).

Figure 6. A summary of the temperatures, pressures and depths of formation of the Little Lottery River xenoliths. The Oligocene-Miocene geotherm for the South Island is estimated to be ∼ 70 mW m^-2, following Scott et al. (2014b). The thickness of the crust is from Wilson et al. (2004). On the righthand side, “lab” is the approximate lithosphere-asthenosphere boundary.

Major and trace element geochemistry

The six samples analysed for whole-rock geochemistry include two harzburgites, two lherzolites, an olivine-websterite and an orthopyroxenite. The whole-rock concentrations of major and trace elements are summarised in the Data Supplementary. For the purposes of comparison, the major elements are normalised to 100% anhydrous.

Rock types from the Little Lottery River can be easily separated based on their major elemental composition. The harzburgite samples, LLR-31 and LLR-33, have the highest Mg# (89.4 and 90.1, respectively) and the highest Cr# (16.3 and 24.8, respectively) of all the samples. Figure 7 displays the major elemental compositions of the samples normalised to LLR-33, which appears to have the most mantle-like standard peridotite composition based on > 43 wt% MgO, highest Mg# (90.1) and ∼ 0.5 wt% Cr₂O₃. The lherzolite, olivine-websterite and orthopyroxenite xenoliths are enriched in Al₂O₃, CaO, FeO, Na₂O, K₂O and TiO₂ but depleted in MgO and Cr₂O₃ compared to LLR-33. The other harzburgite sample (LLR-31) is chemically similar to LLR-33 but is depleted in CaO and enriched in K₂O, TiO₂ and Cr₂O₃. The two harzburgite xenoliths have low Al₂O₃ (0.96 and 1.09 wt %) and TiO₂ (0.11 and 0.18 wt %) whereas the two lherzolite samples have higher Al₂O₃ (2.37 and 1.49 wt %) and TiO₂ (0.39 and 0.35 wt %). The olivine-websterite sample has the highest concentration of TiO₂ at 0.51 wt% and the orthopyroxenite has the highest Al₂O₃ (4.08 wt%).

Figure 7. Whole-rock anhydrous major element compositions of the Little Lottery River samples normalised to the inferred most mantle-like peridotitic sample; LLR-33.

Clinopyroxene chondrite-normalised rare Earth element (REE) plots show relative depletion in heavy REE and elevated light REE (Data Supplementary; Figure 8). Samples LLR-33 and LLR-12 show a slight curvature in their REE patterns, with elevated middle REE over light REE. Internal variation in REE abundance is especially evident in clinopyroxene analyses within LLR-41, LLR-31 and LLR-29. Scattering and variation are most common in the LREE. Apart from LLR-33 and LLR-31, all other peridotite samples display positive Eu anomalies. Clinopyroxene within the pyroxenite samples display the same enrichment in the light REE compared to the heavy REE that occurs within the peridotite samples. Clinopyroxene within both olivine-websterites (LLR-40 and LLR-3) exhibits a positive Eu anomaly and high variability inlight REE (Figure 8).

Figure 8. CI Chondrite-normalised clinopyroxene plots displaying the patterns of REE. Normalising values from Sun and McDonough (1989).

Orthopyroxene analyses in most of the peridotites are depleted in light REE compared to heavy REE (Supplementary Data; Figure 9). This depletion is most prominent in the harzburgite samples, with the lherzolite samples tending to show a flatter pattern. Variation between the concentration of different orthopyroxenes within each xenolith is also evident, with most variation tending to occur within the light REE. This could in part be due to the low concentrations or inclusions of a LREE-rich phase in the ablation stream. Orthopyroxenes within the pyroxenite samples exhibit relatively flat chondrite-normalised REE patterns (Figure 9). Large amounts of internal variation are evident through the scatter of REE plots for these xenolith samples. When comparing the REE patterns of co-existing clinopyroxene and orthopyroxene, the samples display relatively uniform linear heavy REE trends (Figure 10). This linearity typically indicates that the two mineral phases were likely in equilibrium (Agranier and Lee 2007).

Figure 9. CI Chondrite-normalised orthopyroxene plots displaying the patterns of REE within orthopyroxene. Normalising values from Sun and McDonough (1989).

Figure 10. Comparison of the trends of orthopyroxene diivded by clinopyroxene REE i. Figure adapted from Agranier and Lee (2007).

In situ clinopyroxene ⁸⁷Sr/⁸⁶Sr isotopes

Of the eight samples inspected for in situ Sr isotopes (Supplementary Data; Figure 11), three were harzburgites (LLR-9, 31, 33). Four clinopyroxene analyses of harzburgite LLR-9 were collected and the mean ⁸⁷Sr/⁸⁶Sr ratio was 0.70483 ± 0.00034, excluding the distinctly less radiogenic LLR-9_4 analysis. The average estimated ⁸⁷Rb/⁸⁶Sr was 0.34 and, when age-corrected to 15 Ma – the inferred age of the intrusion (Coote 1987) – the age-corrected ⁸⁷Sr/⁸⁶Sr_{15 Ma} is 0.70474. Three analyses of clinopyroxene within harzburgite LLR-31 produced an average ⁸⁷Sr/⁸⁶Sr ratio of 0.70321 ± 0.00030. The average estimated ⁸⁷Rb/⁸⁶Sr was 0.15 and age correction yielded a value of ⁸⁷Sr/⁸⁶Sr_{15 Ma} as 0.70318. Four analyses of harzburgite LLR-33 yielded a ⁸⁷Sr/⁸⁶Sr ratio of 0.70337 ± 0.00054. The average estimated ⁸⁷Rb/⁸⁶Sr was 0.10 and age-correction gave a ⁸⁷Sr/⁸⁶Sr ratio of 0.70335.

Figure 11. Age corrected ⁸⁷Sr/⁸⁶Sr_{15 Ma} ratios for clinopyroxene in different samples from the Little Lottery River. The orange line displays the average ⁸⁷Sr/⁸⁶Sr_{15 Ma} for the host Pahau Terrane (Adams et al. 2005). The green line shows the average MORB ⁸⁷Sr/⁸⁶Sr (Salters and Stracke 2004). The blue line shows the Hikurangi Plateau from Mortimer and Parkinson (1996). New Zealand mantle from Scott et al. (2014b), McCoy-West et al. (2016), Scott et al. (2016a) and Dalton et al. (2017).

Five clinopyroxene analyses of the lherzolite sample LLR-42 gave an average ⁸⁷Sr/⁸⁶Sr ratio of 0.70301 ± 0.00022 (Figure 11). The average estimated ⁸⁷Rb/⁸⁶Sr was 0.19 and age-correction to 15 Ma yielded ⁸⁷Sr/⁸⁶Sr of 0.70297.

Five clinopyroxene analyses of the wehrlite LLR-41 yielded an average ⁸⁷Sr/⁸⁶Sr of 0.70332 ± 0.00015 with a MSWD of 1.6 (Figure 11). The estimated ⁸⁷Rb/⁸⁶Sr of the sample was 0.28 and age-correction to 15 Ma gave 0.70326.

Clinopyroxene in two olivine websterites were analysed (Figure 11). Five analyses in olivine-websterite LLR-40 yielded an average ⁸⁷Sr/⁸⁶Sr of 0.70288 ± 0.00028. The estimated ⁸⁷Rb/⁸⁶Sr was 0.02 and age correction yields a ⁸⁷Sr/⁸⁶Sr_{15 Ma} ratio of 0.70288. Five clinopyroxene analyses from LLR-3 yielded a weighted mean ⁸⁷Sr/⁸⁶Sr of 0.70352 ± 0.00017 . The average ⁸⁷Rb/⁸⁶Sr was 0.54 and the age-corrected ⁸⁷Sr/⁸⁶Sr_{15 Ma} was 0.70341.

Clinopyroxene is rare in the orthopyroxenite LLR-7 and only two analyses could be made (Figure 11). These yielded an average ⁸⁷Sr/⁸⁶Sr ratio of 0.70295 ± 0.00043. The average ⁸⁷Rb/⁸⁶Sr was 0.29 and age-correction gave ⁸⁷Sr/⁸⁶Sr_{15 Ma} = 0.70289.

Discussion

Composition and evolution of the mantle beneath the Little Lottery River

The xenoliths reveal that the mantle sampled by the 15 Ma nephelinite was extremely heterogeneous and made of peridotite and pyroxenite. The separation of peridotite versus pyroxenite is an arbitrary subdivision made on the basis of an ultramafic rock containing either more or less than 40% olivine and does not mean that rock types cannot be related. Pyroxenite can, for example, form from melt-peridotite reaction where a silicate melt metasomatises peridotite by enriching the element budget and causing formation of new minerals and the destruction of others (e.g. Garrido and Bodinier 1999; Xu et al. 2003; Sobolev et al. 2005). Our samples appear to form an array from peridotite to pyroxenite (Figure 12) and may therefore be related. The pyroxenites have high concentrations of Cr for this type lithology, which suggests that these were likely to be highly modified peridotites rather than basalitic-type melts. Furthermore, both lithologies come from similar temperatures (Figure 6) and have similar ⁸⁷Sr/⁸⁶Sr ratios (Figure 11). Although all samples are metasomatised, as judged by the clinopyroxene LREE trace element patters being greater than 10 * chondrite (the typical value for unmodified mantle peridotite), the spinel-bearing harzburgites with high olivine Mg# (LLR 31, 33) would represent the least metasomatised mantle, whereas the rocks with olivine Mg# < 89.6 – some harzburgites and dunites, the lherzolites, wehrlite, olivine websterites and orthopyroxenite– would represent the most Fe-enriched and therefore the most metasomatised. It is also evident that the most mantle-like peridotitic samples (LLR-33 and 12) have no clinopyroxene Eu anomaly, whereas the more inferred metasomatised lithologies have positive Eu anomalies (Figure 8) as well as higher normalised LREE concentrations. There is also a positive relationship between lower olivine Mg# and higher LREE.

Figure 12. Ultramafic rock classification showing the spread of the Little Lottery River data and inferred evolution. The pink field represents the inferred depleted mantle range prior to metasomatism. Blue arrows represent typical peridotite melting trajectories from Niu et al. (1998).

Whole rock geochemistry indicates that the metasomatic fluids were enriched in LREE, Al₂O₃, CaO, FeO, Na₂O, K₂O and TiO₂ (Figure 7). High Ti and Fe within peridotites correlates to the high abundance of oxide phases, such as ilmenite, which typically forms in peridotite via reaction of a Fe and Ti-rich fluid, such as a silicate melt (Drury and Van Roermund 1988; Klemme et al. 1995; Perinelli et al. 2008; Rehfeldt et al. 2008; Lorand and Gregoire 2010; Gervasoni et al. 2017). This type of fluid would also account for the high incompatible REE abundances in these rocks, which are demonstrated in the clinopyroxene and orthopyroxene data (Figures 8 and 9). The clinopyroxene within the strongly metasomatised xenoliths display positive anomalies of Eu relative to Sm and Gd (Figure 8). Such anomalies are sometimes linked to the presence of plagioclase (Weill and Drake 1973), or to so-called ‘ghost plagioclase’ signatures suggested to form by a plagioclase-rich melt or a melt derived from plagioclase-rich rock (Sobolev et al. 2000; Marchesi et al. 2013). On the other hand, Eu anomalies may be linked to a change in oxidation state caused by metasomatism. In either case, the Eu anomaly in the Little Lottery River xenoliths is evidently related to the metasomatic event.

The Sr isotope data show the weakly to highly metasomatised samples to have a tight age-corrected ⁸⁷Sr/⁸⁶Sr_{15 Ma} population range. When plotted against average MORB (∼0.7027; Salters and Stracke 2004) and the ⁸⁷Sr/⁸⁶Sr of New Zealand peridotite xenoliths (from Scott et al. 2014b, 2016a; McCoy-West et al. 2016; Dalton et al. 2017), they sit just above average MORB but within the known range of the New Zealand mantle lithosphere (Figure 11). One exception may be harzburgite LLR-9, which has a slightly elevated ⁸⁷Sr/⁸⁶Sr of 0.70474 ± 0.00034. One sample is insufficient to draw major conclusions on its origin, but a simple explanation of this deviation, assuming it occurred in the mantle, is that it could be due to mixing of crustally-derived and more radiogenic Sr-bearing fluids with mantle material (e.g. Zhao et al. 2013).

The metasomatism recorded in the Little Lottery River xenoliths is quite different to the carbonate metasomatism that has been inferred to have widely affected the mantle below other parts of Zealandia (Scott et al. 2014a, 2014b; McCoy-West et al. 2015, 2016; Scott et al. 2016a, 2016b; Dalton et al. 2017; Delpech et al. 2022). A prominent indicator of carbonate metasomatism in peridotite is low Ti/Eu in clinopyroxene (Rudnick et al. 1993; Klemme et al. 1995; Coltorti et al. 1999; Scott et al. 2014a, 2014b); however, the Little Lottery River xenolith clinopyroxenes have high Ti/Eu (average 5096). Wittig et al. (2010) suggested that a clinopyroxene Th/U ratio greater than 4 indicates carbonatite metasomatism, and most of the data plot well below this value (not shown). The data therefore point to the depleted peridotitic mantle beneath the Little Lottery River having been metasomatised by silicate melts. The composition of such melts would likely be of basaltic or slightly evolved composition to cause the Fe-enrichment recognised in the xenoliths.

The northern side of the Marlborough Fault System is inferred to represent the extent of the oceanic Hikurangi Plateau, which collided with Zealandia at about ∼ 105 Ma and is today inferred to be at the base of the crust in Canterbury and abut or even extend under the crustal Marlborough Fault System (Reyners 2018; Eberhart-Phillips et al. 2022) (Figure 1). If this is the case, the Little Lottery River xenoliths should come from the mantle lithosphere beneath the crustal portion of the oceanic plateau, since the plateau is modelled as at the base of the crust in this region (Reyners 2018; Eberhart-Phillips et al. 2022) and the xenoliths are from temperatures and pressures equivalent to the mid mantle lithosphere (Figures 6 and 13). The ⁸⁷Sr/⁸⁶Sr isotopic composition of some of the Hikurangi Plateau crustal rocks (0.70361–0.70374; Mortimer and Parkinson 1996) overlaps with the isotopic composition of the Little Lottery River xenoliths, and it is therefore possible that the extensive silicate melt metasomatism could be related to the Cretaceous percolation through the mantle lithosphere of plateau-forming melts . Portions of the mantle lithosphere beneath the Hikurangi Plateau are thought to have elevated ²⁰⁷Pb/²⁰⁶Pb compared to the Zealandia lithospheric mantle (>15.7; McCoy-West et al. 2016; Van der Meer et al. 2017), and published Pb data on two Little Lottery River samples are close to that value (²⁰⁷Pb/²⁰⁶Pb = 15.67 and 15.71; McCoy-West et al. 2016). Additional Pb isotopic work on the xenoliths would greatly help refine or refute this interpretation.

Deformation in the mantle beneath the Little Lottery River

Deformation textures in xenoliths can be used to show the level of fault zone propagation into the upper mantle (Titus et al. 2007; Vauchez et al. 2012; Kidder et al. 2021). The Hope Fault is the southern strand of the Marlborough Fault System, which is an expression of the Australian-Pacific plate boundary and is less than 1 km from the Little Lottery River (Figure 1). The fine-grained recrystallised dunites and olivine websterite xenoliths (Figure 4B, C) require their extraction from an active (or very recently active) mantle shear zone at the time of entrainment because grain growth calculations indicate that olivine should essentially geologically instantaneously coarsen if the grain boundaries are not pinned (Karato 1989; Kidder et al. 2021). Since only a portion of the samples are deformed, yet the temperatures of equilibration and corresponding pressure and depths estimates indicate derivation from ∼ 40 to 60 km depth (Figure 6), this 15 Ma shear zone must have comprised thin strands at the time of xenolith extraction A similar inference has been reached by Kidder et al. (2021) and Shao et al. (2022) from mantle peridotites associated with New Zealand’s Alpine Fault.

Figure 13. Model of the plate boundary lithosphere during the time of the Little Lottery River intrusion. The four prominent dextral-slip faults of the Marlborough Fault System: the Wairau (Wa), Awatere (Aw), Clarence (Cl) and Hope (Hp) Fault, are annotated in blue. The range of formation depths (40-60 km) for the Little Lottery River xenoliths is indicated in Fig. 6. We show the fault as a strike-slip fault, although these may have been reverse faults early in their history (Collett et al. 2019). The thickness of the crust is from Wilson et al. (2004), with the lithospheric mantle thickness extrapolated from Scott et al. (2014a, 2014b).

The deformation textures in the Little Lottery River xenoliths are, however, difficult to directly attribute to the Hope Fault. This is because of the age disparity between xenolith entrainment (~15 Ma) and the inferred age of the structure (∼1 to 2 Ma) assummed from offset markers. It is possible that the Hope Fault is a reactivated feature that accommodated early vertical movement, in which case the mantle deformation could be related to this early phase. On the other hand, the mantle deformation could potentially be related to a precusor mantle structure in the area. Neither case, however, affects the key conclusion that deformation was occurring in the lithospheric mantle beneath NorthCanterbury at ∼ 15 Ma in an area today that is very close to the position of the Hope Fault. Baker and Seward (1996) and Collett et al. (2019) have shown from fission data that the Marlborough region experienced uplift in the late Oligocene and Early Miocene, and so the mantle deformation could be related to this.

The northern Marlborough Fault System strand, the Wairau Fault, marks the transition in the region from Australian Plate to the Pacific Plate and probably offsets and sheared the lithospheric mantle in this region (Figure 13). It is unclear if the mantle composition beneath the Little Lottery River had any bearing on the location of the Hope Fault. However, the closest mantle xenolith location on the northwestern side of the Marlborough Fault System occurs at Lake Moana (Figure 1) and he mantle beneath there, at least at the time of Cretaceous extraction, is extensively depleted and only marginly metasomatised (Tulloch and Nathan 1990; Scott et al. 2016b; Cooper 2021). This is potentially importantbecause even small changes in mineral properties or modes affect density and rheology of the mantle lithosphere (e.g. Pearson et al. 2021). Thus,there is probably significant compositional variation in the Zealandia mantle lithosphere across the Marlborough Fault System.

Conclusions

Mantle xenoliths entrained in a nephelinite exposed in Little Lottery River provide insight into the composition and evolution of the mantle at the junction of the Australia and Pacific plates. The xenoliths form a spectrum from peridotite to pyroxenite, with the pyroxenites here interpreted on the basis of geochemistry, and similar equilibration temperatures and indistinguishable ⁸⁷Sr/⁸⁶Sr isotopes to the peridotiites, to be the result of reaction of depleted peridotite with silicate melts. These melts were probably basaltic or similar composition. Deformation textures, especially the occurrence of fine-grained dunites, indicate that the mantle lithosphere was undergoing deformation at the time of xenolith entrainment. Since the deformation features occur in only a subset of the samples, this deformation was occurring in discrete zones. Furthermore, the age of the nephelinite carrying the xenoliths, 15 Ma, means that this deformation predated the proximal translithopheric Hope Fault by ∼13 to 14 Ma and the overall Marlborough Fault System by ∼ 5 Ma assuming the ∼ 10 Ma age of formation for this fault system is correct. There may therefore be a precusor plate boundary structure beneath NorthCanterbury that is preserved in the xenolith histories.

Acknowledgments

We thank Amuri Helicopters for travel into the field. Fieldwork was supported by the Bob Carter Memorial Fund, and analyses were sponsored by the Geology Department. Thanks to B. Pooley for high quality thin sections, M. Brenna for comments on a draft, M. Palin for managing the in-house Sr isotope spreadsheet, H. Dalton and T. Waight for reviews, and J. Hopkins for handling the manuscript.

Disclosure statement

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

Data availability statement

The data that support the findings of this study are openly available in Figshare at https://doi.org/10.6084/m9.figshare.22939460.