Basanite cobbles in Pleistocene sediments in Central Otago and their implications for intraplate volcanism and Clutha River paleo-drainage

ABSTRACT

The occurrence of volcanic basanite cobbles in Pleistocene terraces at Galloway and in the upper Clutha valley in Central Otago, in an area devoid of known volcanic edifices, has implications for Cenozoic intraplate volcanism and Pleistocene drainage in the region. We present petrographic, mineralogical, whole rock geochemical and Sr-Nd-Pb isotopic data, as well as an ³⁹Ar-⁴⁰Ar date of 24.4 ± 0.06 Ma, together with mineral analyses of enclosed peridotite xenoliths. We conclude from this dataset that these alkaline basanite cobbles are most likely derived from the Alpine Dike Swarm in northwest Otago, probably in the Lake Wanaka-Hawea-Luggate area, which extends the volcanic field east of its known extent. The pilotaxitic textures and presence of intersertal glass, along with the volume of basanitic material, suggests that the source rocks were likely effusive, which would make this the first such known component of the Alpine Dike Swarm and requires emergent land at that time. The occurrence of the Galloway basanite cobbles implies that Pleistocene Clutha River flowed through or over what is now the Dunstan Range. The young (late Pleistocene) age of all the basanite-bearing gravels means that the basanite cobbles have likely been reworked several times before final deposition.

KEYWORDS:

• Basanite
• geochemistry
• isotopes
• Alpine Dike Swarm
• Dunedin Volcanic Group

Introduction

Volcanic cobbles are common components of river gravels in east Otago, especially in the Taieri River and Waikouaiti and Shag valleys (Figure 1a,b), where they are derived from outcrops of the Late Oligocene-Miocene Dunedin Volcanic Group (Figure 1a,b; Coombs et al. 1986, 2008; Scott et al. 2020a). Volcanic cobbles are also common in the rivers of northwest Otago, principally the Makarora Valley, as well as the upper Clutha that drains Lakes Hawea and Wanaka catchments, and the Shotover-Kawarau catchments farther south (Figure 1a,b). These latter cobbles are derived from the Alpine Dike Swarm, which forms dikes, sills and diatremes emplaced during the Late Oligocene-Early Miocene (Figure 1a,b; Cooper 1986, 2020; Cooper et al. 1987). Other than a single lamprophyric dike at the Nevis Bluff (Hutton 1943), there are no known outcrops of either the Dunedin Volcanic Group or Alpine Dike Swarm within the mountains or valleys around Cromwell and Alexandra in Central Otago. However, lamprophyre and basanitic cobbles occur in the main channel of the Clutha River and in stranded glacial outwash sediments on the western flanks of the Dunstan Range (Figure 1b), and basanitic cobbles occur in Pleistocene gravels in the Manuherikia River valley in the Galloway district north of Alexandra (Figure 1b and 2a,b).

Figure 1. Locations of key features mentioned in this study. (a) The Otago Schist belt in the South Island of New Zealand. (b) DEM image of Otago Schist and adjacent areas showing the general distributional areas of Alpine Dike Swarm (ADS) and Dunedin Volcanic Group (DVG). The Upper Clutha and Galloway areas of this study occur in the intervening valleys.

The modern Clutha River-hosted lamprophyre cobbles can be explained via river drainage patterns as originating from the Alpine Dike Swarm (Figure 2a). However, the cobbles at Galloway are > 5 km upstream of the Clutha River-Manuherika River confluence and are separated from the Dunedin Volcanic Group by the Rough Ridge range (Figures 1b and 2a,b). Hence, defining a source for these cobbles within the present drainage system is problematic. Previous studies have inferred river drainage adjustments during the Pleistocene (e.g. Youngson et al. 1998; Bennett et al. 2006; Craw et al. 2012; Craw 2013) that could be invoked to argue for derivation of Galloway volcanic cobbles from either the Alpine Dike Swarm to the northwest (Figure 2a) or the Dunedin Volcanic Group from the east (Figure 2b). Fluvial transport of Dunedin Volcanic Group basaltic cobbles to the west of Galloway, into what is now the Upper Clutha valley (Figure 2b), is also theoretically possible prior to rise of Rough Ridge and the Dunstan Range.

Figure 2. Digital elevation model (DEM) images that summarise contrasting scenarios for possible river redistribution of volcanic cobbles in Central Otago. Low relief valley areas (green shades) are almost entirely Plio-Pleistocene sediments. (a) Northwest Otago, showing derivation of cobbles from presumed eruptive basanites in the Hawea catchment via the ancestral Clutha River through what is now the Dunstan Range. (b) East Otago, showing alternative scenario (rejected in this study) for derivation of Central Otago volcanic cobbles from DVG in the Maniototo Basin through what is now Rough Ridge to ancestral Ida Burn, followed later by diversion through Raggedy Range.

The aim of this paper is to characterise the volcanic cobbles, identify their source and distinguish between the two major Pleistocene drainage scenarios presented above. To do so, we present a petrographic, geochemical and isotopic (Sr, Nd, Pb and Ar) dataset for the Galloway cobbles and previously undescribed volcanic cobbles in the Upper Clutha valley. These data are compared with the recently established extensive geochemical databases for the Dunedin Volcanic Group (Scott et al. 2020a) and Alpine Dike Swarm (Cooper 2020). Furthermore, the occurrence of mantle xenoliths in these volcanic rocks means that these can be compared to the West Otago mantle through which Alpine Dike Swarm magmas travelled (Wallace 1975; Brodie and Cooper 1989; Scott et al. 2014b, 2016a, 2019; Shao et al. 2022) or the East Otago mantle through which the Dunedin Volcanic Group magmas travelled (Reay et al. 1987; McCoy-West et al. 2013; Scott et al. 2014a, 2014b; McCoy-West et al. 2015, 2016; Scott et al. 2019; Dalton et al. 2017; Auer et al. 2020; Shao et al. 2021).

Geological context

Volcanism in Otago

There are three intraplate volcanic fields in Otago: the East Otago ∼34 Ma Waiareka-Deborah field near Oamaru (Coombs et al. 1986, 2008; Mortimer and Scott 2020; Scott et al. 2020b), the 24–29 Ma Dunedin Volcanic Group (Coombs et al. 1986, 2008; Scott et al. 2020a), and the ∼24 ± 3 Ma alkaline Alpine Dike Swarm in the mountains and valleys of West Otago and South Westland (Cooper 1986; Cooper et al. 1987; Cooper 2020; Serre et al. 2020) (Figure 1b). The Waiareka-Deborah volcanic field erupted in a submarine setting and is mostly composed of sub-alkaline tuffs and submarine shallow sills (Scott et al. 2020b). In contrast, the Dunedin Volcanic Group and Alpine Dike Swarm volcanic fields are both dominantly alkaline. Although the early phases of the Dunedin Volcanic Group overlapped with eruption of the Alpine Dike Swarm, the two fields are separated by approximately 100 km without any evidence of intervening magmatism. The primary magmas of the Dunedin Volcanic Group are mostly basanitic but evolved to phonolite and trachyte (Price and Compston 1973; Coombs et al. 2008; Scott et al. 2020a; Pontesilli et al. 2021; Baxter and White 2022), whereas the Alpine Dike Swarm tends to mostly be slightly less silica-saturated and its composition ranges from carbonatite to ultramafic lamprophyre and phonolite (Cooper 1986, 2020); basanite is largely absent, except for a single basanitic diatreme located in the mountainous area between the lamprophyric Mt Alta, Minaret, Niger and Lake Wanaka diatremes (Perry 2022) (Figure 1b).

Upper Clutha cobbles

Lamprophyre cobbles occur in Pleistocene gravels as well as the active Clutha riverbed from the Wanaka and Hawea catchments to Alexandra and beyond (e.g. Cooper and Beck 2009). Abundant sources occur in the Lake Wanaka area, with rare outcrops at The Neck in Lake Hawea (Cooper 1986, 2020). Basanite and lamprophyre cobbles occur in gravels exposed in road cuttings south of Red Bridge, which spans the Clutha River near Luggate (−44.732155, 169.28088), and in Q12 (Q, Quaternary) correlated terrace gravels exposed at Bendigo (−44.94865, 169.34957), Bendigo Loop Road (−44.92828, 169.32665), Quartz Reef Point (−44.99796, 169.24599) and in Q6 correlated gravels at Middleton Road (−45.01756, 169.22874) near Cromwell (Figure 3a). Accompanying the Upper Clutha volcanic cobbles are clasts of greywacke, quartzofeldspathic schist, metabasite and metachert (Figure 4a,b).

Figure 3. Basanite localities in Pleistocene gravels in Central Otago. (a) Oblique DEM showing topographic setting of sample sites in this study (red dots). (b) Oblique DEM view of the Galloway terrace gravel area beside the Manuherikia River, upstream of the Clutha River, and at the foot of the Waikerikeri Fan.

Figure 4. Outcrop photographs of the Pleistocene gravels that host the basanite cobbles. (a) Outcrop photograph of basanite cobbles (B) together with greywacke (G), high textural grade schist (lower left) and metabasite (M) in late Pleistocene (Q12; Turnbull 2000) terrace gravels near Quartz Reef Point. (b) Close view of a large basanite cobble at Quartz Reef Point. . (c) Location of the terrace gravels unconformably on Miocene mudstone at Galloway. (d) Typical view of Galloway terrace gravels with abundant quartzofeldspathic greywacke cobbles (G) and schist-derived debris, with subordinate basanite. (e) A large well-rounded basanite cobble at Galloway.

Galloway volcanic cobbles

The studied volcanic cobbles come from near the base of river gravels 20–40 m above the modern Manuherikia River in the Galloway district, north of Alexandra. The rocks were taken from road cuts and river cliffs at −45.224309, 169.429479 and −45.224128, 169.425616 (Figures 1b, 3a,b and 4c–e). The gravels rest unconformably on Miocene Bannockburn Formation mudstone (Figure 4c; Douglas 1986) and form the top of a prominent late Pleistocene (300-400 ka) fluvial terrace (Turnbull 2000). A prominent Pleistocene alluvial fan complex, the Waikerikeri Fan, is of similar age and is still active, with streams that have eroded into the older gravels within and below the complex (Turnbull 2000) and channels from this fan cut across the Galloway terrace containing the volcanic cobbles (Figure 3b). The fan was a tributary draining to the Manuherikia River from the northwest and does not contain volcanic cobbles. We have not found volcanic cobbles in Manuherikia River gravels upstream from Galloway, or in terraces on the eastern side of the Manuherikia River valley. The cobbles are accompanied by a wide range of clasts derived from Otago Schist basement, and greywacke from northeast of the schist belt (Figures 1a, 2 and 4d). Schist clasts are dominantly quartz-rich, and include material from Textural Zones 2, 3 and 4 (Figure 4d; Turnbull et al. 2001). Metabasite (greenschist) cobbles are rare. All the gravel clasts, including the volcanic cobbles, are well rounded (Figure 4c,e) and imbrication indicates the depositional flow direction was towards the south, parallel to the modern Manuherikia River.

Methods

Mineral chemistry was obtained on a scanning electron microscope with an electron dispersal x-ray attachment at the University of Otago. The electron beam intensity for the peridotite xenoliths was calibrated on pure cobalt with a 15 kV accelerating voltage. Analyses of the volcanic clasts was by semi-quantitative electron dispersal x-ray and were standardised against the internal factory standards, which, at the time, produced good quality results as judged by mineral totals and stoichiometry, coupled with the results of secondary Smithsonian standards.

Alteration rinds were removed from the cobbles using a diamond-tipped rock saw and surfaces cleaned with carborundum before crushing to a fine powder in a tungsten-carbide mill. One aliquot was sent to ALS Brisbane for whole rock geochemical analysis. The elemental data were obtained by inductively coupled plasma-atomic emission spectroscopy on fused glass disks, with in-house standards run with each batch. Loss on ignition was calculated by the difference in weight of an initial one gram of sample powder after being heated at 1000°C for one hour.

A second aliquot was processed for ¹⁴³Nd/¹⁴⁴Nd, ⁸⁷Sr/⁸⁶Sr ²⁰⁶Pb/²⁰⁴Pb, ²⁰⁷Pb/²⁰⁴Pb and ²⁰⁸Pb/²⁰⁴Pb at the University of Cape Town. Samples were dissolved in concentrated HF:HNO₃ for 48 h at 140°C, converted to nitrate and then processed for chromatographic elemental separation using standard column chemistry and analysed using routine MC-ICP-MS protocols (see Harris et al. 2015). The cobbles from Galloway were analysed first, with BHVO-2 used as a monitoring standard. It yielded ¹⁴³Nd/¹⁴⁴Nd = 0.512993 ± 0.000014 (all errors are 2 s.e.), ⁸⁷Sr/⁸⁶Sr = 0.703470 ± 0.000011, and ²⁰⁶Pb/²⁰⁸Pb = 18.6141 ± 0.0009, ²⁰⁷Pb/²⁰⁴Pb = 15.5374 ± 0.0009 and ²⁰⁸Pb/²⁰⁴Pb = 38.2247 ± 0.0031, which overlap the reference values of Weis et al. (2006). BHVO-2 in the second batch, comprising three Upper Clutha rocks, yielded ¹⁴³Nd/¹⁴⁴Nd = 0.513005 ± 0.000011 (all errors are 2 s.e.), ⁸⁷Sr/⁸⁶Sr = 0.703455 ± 0.000012, and ²⁰⁶Pb/²⁰⁸Pb = 18.6493 ± 0.0011, ²⁰⁷Pb/²⁰⁴Pb = 15.5359 ± 0.0094 and ²⁰⁸Pb/²⁰⁴Pb = 38.2408 ± 0.0030, which also overlap values of Weis et al. (2006).

The groundmass separated for ³⁹Ar-⁴⁰Ar dating was ultrasonically cleaned with 5% HNO₃ (10 min), 2% HF (1 min), demineralised water (10 min) and then rinsed in acetone. The cleaned grains were weighed and wrapped in aluminium foil envelopes, and then placed into quartz glass vials together with interspersed aliquots of the flux monitor Fish Canyon Tuff sanidine (Age = 21.176 ± 0.005 Ma; Phillips et al. 2022). The package (UM#98) was then encapsulated in an outer sealed glass vial and irradiated in the CLICIT facility of the Oregon State University TRIGA Reactor for 20 MWh. Following irradiation and cooling, ⁴⁰Ar-³⁹Ar analyses were undertaken in the Noble Gas laboratory at the University of Melbourne. Step-heating analyses were conducted on single weighed aliquots of groundmass grains placed into the sample chamber of a gas-handling system equipped with a Photon Machines Fusions 10.6 CO₂ laser and connected to a Thermo Fisher Scientific ARGUSVI mass spectrometer at the University of Melbourne. Apparatus details are given in Phillips and Matchan (2013) with updated Faraday detectors now equipped with 1 × 10¹³ Ω resistors as described in Heath et al. (2018). Analytical methods follow those described by Matchan and Phillips (2014) and Heath et al. (2018). For incremental heating a 6 mm laser beam-size was utilised with laser power varied (2–30%), dependent on number of heating steps (1-13) for each aliquot. All results are corrected for system blanks, mass discrimination, radioactive decay and reactor-induced interference reactions. Correction factors vary for each irradiation are supplied in the Supplementary File. Mass discrimination was monitored by analysis of standard air volumes, assuming the air argon isotopic composition of Lee et al. (2006). Inclusion of uncertainties in the J-value and age of Fish Canyon Tuff sanidine have a negligible impact on uncertainties. Decay constants are those of Steiger and Jäger (1977). The ⁴⁰Ar-³⁹Ar dating technique is described in detail by McDougall and Harrison (1999). Age spectra were generated using ISOPLOT (Ludwig 2003). Age uncertainties reported in the results section are 2σ.

Results

Petrography and geochemistry

Since the volcanic cobbles collected from terrace gravels in the Galloway and the Upper Clutha valley are petrographically and geochemically identical, the following description compiles observations from both areas. The volcanic rocks are porphyritic with euhedral micro-phenocrysts (typically 0.23–0.60 mm) of olivine, with subordinate magnetite, and, in a few samples, Ti-augite (Figure 5a,b). In some specimens, the olivine (typically Fo_70-73, but with rims zoned to Fo_64; Supplementary Table 1) is fresh, but typically shows variable degrees of marginal or vein serpentinisation. Aggregates of fine- to very fine-grained fawn-brown to pinkish Ti-augite and titano-magnetite, typically 0.05–0.07 mm, dominate the intergranular and in places intersertal spaces between the framework of plagioclase laths (typically 0.06–0.17 mm long) (Figure 5b). Grains of Ti-augite yield both Ti-Al-rich (5.06 wt. % TiO₂, 9.18 wt. % Al₂O₃) and Ti-Al-poor (2.56 wt. % TiO₂, 4.84 wt. % Al₂O₃) compositions (Supplementary Table 1). Magnetite analyses are titaniferous (TiO₂ ranging up to 22.8 wt. % TiO₂).

Figure 5. Thin section images. A. The basanites contain fine-grained plagioclase with phenocrysts of olivine (ol) and titan-augite (Ti-aug). B. Some samples display a pilotaxic alignment of plagioclase (pl) grains set in glass, titano-magnetite (mag), titan-augite and olivine. C. A brown glass occurs between plagioclase laths. A–C are Upper Clutha sample 8C. D. Peridotite xenolith in very fine-grained basanite (Galloway 10J). The peridotite contains spinel (sp)-orthopyroxene (opx) vermicular intergrowths.

Pilotaxitic flow-aligned textures of plagioclase occur in some specimens (Figure 5b). Plagioclase compositions are typically labradorite (as calcic as An₅₆Ab₄₂Or₃) but are zoned to rims of andesine (An₄₃Ab₅₃Or₄) (Supplementary Table 1). Interstitial material is invariably present and has an intergranular to intersertal texture. In most specimens, the interstices consist of a pale brown, apparently isotropic glass phase containing feathery, in places dendritic (quench?), microlites, some of which are acicular needles of an opaque phase (ilmenite?) (Figure 5c). These dendrites are cut by hollow needles of apatite. In some specimens, the interstitial material is more yellow-brown and birefringent than the intersertal glasses and may have resulted from devitrification/recrystallisation. Compositions of this pale brown interstitial phase are highly variable, with low analytical totals probably reflecting variable degrees of hydration and devitrification. Some specimens contain coarser leucocratic patches, with crystals of isotropic analcite on the margins and finer-grained feldspathic or feldspathoidal crystal aggregates in the interior. The analcite may result from devitrification and recrystallisation of glass. The least hydrated glass compositions are typically rich in Na₂O (e.g. 9.71%), Al₂O₃ (e.g. 23.37%), with K₂O typically < 1%, SiO₂ ∼54% and FeO < 1% (Supplementary Table 1) and contain Cl, although this element was not analysed quantitatively. Such compositions are similar to glasses analysed in Alpine Dike Swarm lamprophyres, which Cooper (1979) likened to compositions of phonolites that accompany lamprophyres in the evolved part of the Alpine Dike Swarm. Wilkinson (1966) determined that trachytic and phonolitic intersertal glasses in basalts are dependent on the degree of undersaturation of the host magma. A very similar conclusion was also reached by Tsypukova et al. (2014) who scanned vitreous groundmass areas, 800 mm square, by a focused beam. For basanites of similar composition to the Clutha samples, this method determined phonotephrite residual magma compositions. One specimen each from the Upper Clutha and Galloway suites contain nepheline (Ne_75.7-79.5Ks_12.26-12.25Qtz_12.05-8.26), and the same Galloway sample also showed the presence of interstitial pools of sodalite (Supplementary Table 1). The presence of sodalite, nepheline and Cl-bearing glasses in the Upper Clutha-Galloway basanites indicates a very similar fractionation sequence to that observed in the Alpine Dike Swarm.

Peridotite xenoliths

One specimen from Galloway (10J) and three specimens from Upper Clutha (C6A, 8A and 9A) contain polycrystalline peridotite xenoliths up to 4 cm diameter and composed of a granular harzburgitic assemblage of olivine and orthopyroxene with minor spinel (Figure 5d). Other basanites contain single xenocrysts of olivine exhibiting kink banding and orthopyroxene characterised by fine lamellae of birefringent clinopyroxene. In the xenoliths, clinopyroxene is very rare, and only one grain was found in each xenolith from the Upper Clutha samples. Compositions are chrome-diopside, with Mg# (100*Mg²⁺/(Mg²⁺ + Fe²⁺) ranging from 83.7 (sample C9A) to 92.9 (C6A) (Supplementary Table 2). Orthopyroxene is in the form of slightly lamellate, simple-twinned grains (3–4.5 mm in diameter) that are consistently highly magnesian (Mg# = 89.4–93.2). Olivine is slightly finer grained (1.7–2.3 mm), in places is kink banded, with compositions from Mg# 87.4 in C8A to Mg# 92.8 in 10J. Spinel has a variable texture. It forms equant textures in C8A, 0.35–0.5 mm diameter, that have Cr# (100*Cr³⁺/(Cr³⁺ + Al³⁺)) chromite compositions ranging from 82.0–87.1. In C9A and 10J, chromite is part of a symplectite texture forming vermicular grains of a translucent deep coffee brown colour, intergrown with orthopyroxene (Figure 5d). Spinel Cr# is 78.0 (9A) to 78.6 (10J). In C6A, the granular Cr-spinel has Cr# of 58.2–62.6.

Whole rock analyses

The whole rock analyses of 8 of the visually freshest Galloway samples and 8 Upper Clutha samples (Supplementary Table 3) have small to moderate loss on ignition values (up to 2.24 wt%), probably due to the alteration of glass, and therefore the data are normalised to 100% anhydrous for the following description. The SiO₂ content forms a tight cluster of 42.0–44.7 wt%. MgO ranges from 9.8–6.1 wt%, with the highest values also having the highest Cr (223 ppm). When SiO₂ is compared to Na₂O + K₂O on a total alkali versus silica diagram, the rocks display a tight cluster that falls entirely within the alkaline (> 4.8% normative nepheline) basanite field (Figure 6), although in detail sample 8H has low MgO (6.1 wt%) and a normative plagioclase composition that indicates classification as nepheline hawaiite to be more appropriate. The primitive mantle-normalised data show the volcanic rocks to be enriched in large ion lithophile elements and light rare-earth elements over heavy rare-earth elements, but to have a positive P anomaly in all samples and prominent negative anomalies in Rb, K and Ti in the most enriched samples (Figure 7).

Figure 6. Total alkali versus silicate diagram showing the basanite cobbles compared to the Dunedin Volcanic Group (dataset of Scott et al. 2020a) and Alpine Dike Swarm (dataset of Cooper 2020). Maniototo basalts are from Scott et al. (2020a) and Wilson (2023).

Figure 7. Primitive mantle-normalised diagram showing basanite cobbles compared to the Dunedin Volcanic Group (from Scott et al. 2020a) and the Alpine Dike Swarm (from Cooper 2020).

Groundmass ³⁹Ar-⁴⁰Ar date

Aliquot 8H-1, from sample 8H from Galloway, has a descending ‘staircase’ pattern with apparent ages decreasing from ∼25 Ma, before flattening at ∼24 Ma and then decreasing in the final heating steps to ∼22 Ma (Figure 8a; Supplementary Table 4). The weighted-mean age of the nearest-to-flat steps is 24.00 ± 0.11 Ma. Although this and the weighted mean ages presented below may not be considered statistically robust (i.e. MSWD > 2; p = 0), they may still provide useful estimates for the time of isotopic closure (i.e. crystallisation) of these groundmass grains. The total gas age, an aggregate age for all heating steps weighted by the proportion of ³⁹Ar released in each step, for this sample is 24.2 ± 0.04 Ma.

Figure 8. ⁴⁰Ar/³⁹Ar age spectra for aliquots from basalt groundmass sample and inverse isochron derived from sample 8H from Galloway.

Aliquot 8H-2 yielded a contrasting and more disturbed ³⁹Ar release spectrum relative to 8H-1, with apparent ages decreasing from ∼25.5 Ma, before levelling out at ∼24.7 Ma and then increasing once more to ∼25.5 Ma in the latter steps (Figure 8b). The weighted mean age of ‘middle’, flatter steps is subtly older at 24.69 ± 0.008 Ma, and outside uncertainty of the weighted mean age for 8H-1. Similarly, the total gas age, at 25.1 ± 0.04 Ma, is outside uncertainty of that for 8H-1.

The step-heating spectrum for 8H-3 is notably more concordant than for the aliquots mentioned above, with a subtle decrease in apparent ages from steps 2 to 8 from 24.6 to 24.0 Ma (Figure 8c). The final heating steps are anomalous, as per the above examples, and culminate in ages younger than 23 Ma. Depending on the choice of selected steps in the flattest part of the spectrum, the weighted mean age varies from 24.16 ± 0.02 Ma to 24.26 ± 0.14 Ma. These weighted mean ages are within uncertainty of that calculated for 8H-1. It is also notable the total gas age for 8H-3 is within uncertainty of the total gas ages for both 8H-1 and 8H-2.

To assess the presence of extraneous argon and ensure that the assumption of an air argon ratio is valid, we also constructed inverse isochrons for each aliquot. The inverse isochron for 8H-3 is shown in Figure 8d. All inverse isochrons yield ages consistent with those described above (8H-1, 23.92 ± 0.57 Ma; 8H-2, 25.14 ± 0.34 Ma; 8H-3, 24.26 ± 0.39 Ma). Most importantly, the ⁴⁰Ar/³⁶Ar_(i) ratios (8H-1, 285 ± 21; 8H-2, 277 ± 24 Ma; 8H-3, 285 ± 18 Ma) are all within uncertainty of the accepted ‘air’ value of 298.56 ± 0.31 (Lee et al. 2006); these atmospheric values suggest that extraneous argon is likely not a complicating factor in these analyses.

Although true plateau ages could not be calculated from the groundmass aliquots, two aliquots yield approximately consistent weighted mean ages. It is possible that the discordance observed may therefore be caused by recoil loss and or redistribution of ³⁹Ar^K, with different phases outgassing at different temperatures. In addition, ‘ascending’ and ‘descending’ spectra of these aliquots is often observed for fine-grained mafic volcanic groundmass samples (e.g. Jourdan and Renne 2014; Heath et al. 2018). This pattern is widely considered to reflect sub-micrometre scale nuclear recoil of ³⁹Ar and ³⁷Ar between fine-grained K-rich/Ca-poor and K-poor/Ca-rich phases during neutron irradiation, leading to decoupling of ⁴⁰Ar*/³⁹Ar ratios. Older apparent ages in early (lower temperature) heating steps, such as observed in this study, have commonly been attributed to ³⁹Ar recoil in basaltic samples (e.g. Koppers et al. 2000). For this reason, the crystallisation age is predicted to be intermediate between the oldest and youngest apparent ages for the descending portion of the age spectrum and the total gas age is thus considered a good approximation for the crystallisation age (Jourdan and Renne 2014). It has been shown for both basalts and other igneous lithologies that total gas ages can be within uncertainty of ‘true’ emplacement ages for recoil impacted samples (Heath et al. 2018; Dalton et al. 2020). Noting that aliquot 8H-3 produced the most concordant ³⁹Ar release spectrum, the total gas age for this aliquot, at 24.4 ± 0.06 Ma is therefore the best estimate for this basanite.

Sr-Nd-Pb isotopes

⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd and Pb isotopes gathered on eight rocks from Galloway and three from the Upper Clutha have an extremely tight distribution (Figure 9a–d; Supplementary Table 5). The measured ⁸⁷Sr/⁸⁶Sr ranges from 0.70307 to 0.70401, with age correction to 25 Ma yielding virtually no change (0.70307–0.70398) due to the very low ⁸⁷Rb/⁸⁶Sr. There is a slight positive correlation between Sr concentration and ⁸⁷Sr/⁸⁶Sr, which may mean that some samples have been modified by post-igneous addition of Sr (Figure 9a), as has been found comparing in-situ feldspar and whole rock Sr isotope data for basalts in the Maniototo (Scott, unpublished data). As a result, the least radiogenic Sr samples are probably most representative of the magmatic ⁸⁷Sr/⁸⁶Sr (Figure 9a). Measured ¹⁴³Nd/¹⁴⁴Nd is between 0.51286 to 0.51291 and the age-corrected ¹⁴³Nd/¹⁴⁴Nd_{25 Ma} values are 0.51284–0.51289 with a corresponding ϵNd_{25 Ma} =  + 4.0. to +4.9. There is insufficient Pb trace element data to undertake age correction but the data presented are distinctively radiogenic: ²⁰⁶Pb/²⁰⁴Pb varies from 20.340 to 20.621, ²⁰⁷Pb/²⁰⁴Pb is 15.633 to 15.680 and ²⁰⁶Pb/²⁰⁴Pb is 40.136 to 40.401.

Figure 9. Sr-Nd-Pb isotopic data for the Upper Clutha-Galloway basanites compared to the Dunedin Volcanic Group (data from Price and Compston (1973), Hoernle et al. (2006), Sprung et al. (2007), Timm et al. (2010), Scanlan et al. (2020) and Scott et al. (2020a) and Alpine Dike Swarm (data from Barrerio and Cooper (1987), Hoernle et al. (2006), Timm et al. (2010) and Serre et al. (2020)). The data for the Maniototo basalts, which are part of the DVG, are from Scott et al. (2020a) and Wilson (2023).

Discussion

Comparisons with the Dunedin Volcanic Group and Alpine Dike Swarm

The Upper Clutha-Galloway cobbles are alkaline volcanic rocks, which are common in the Dunedin Volcanic Group and the Alpine Dike Swarm (Figure 6). However, a point of difference between those suites is that most of the primitive magmas in the Dunedin Volcanic Group are basanites (Reay et al. 1991; Coombs et al. 2008; Scott et al. 2020a; Pontesilli et al. 2021) whereas those in the Alpine Dike Swarm are lamprophyric (Cooper 1986, 2020). The normalised trace element abundances of the Galloway-Clutha cobbles are, however, slightly enriched in Zr, Hf and Ti relative to most of the Dunedin Volcanic Group but comparable to the Alpine Dike Swarm (Figure 7).

The new ³⁹Ar-⁴⁰Ar age, 24.4 ± 0.6 Ma (Figure 8), overlaps with the Alpine Dike Swarm, which has a relatively tight age clustering of ∼ 24 ± 3 Ma (Cooper et al. 1987; Cooper 2020). While this new date is indicative, it is not diagnostic because it also overlaps the very earliest Dunedin Volcanic Group eruptions such as at Foulden Maar (Fox et al. 2015) and The Crater (Hoernle et al. 2006) (see Scott et al. 2020a for a summary of age data). The ⁸⁷Sr/⁸⁶Sr_i, ¹⁴³Nd/¹⁴⁴Nd_i and ²⁰⁷Pb/²⁰⁴Pb data of the Galloway and Upper Clutha volcanic cobbles also overlap both volcanic fields (Figure 9). However, the Galloway-Clutha ²⁰⁶Pb/²⁰⁴Pb and ²⁰⁸Pb/²⁰⁴Pb data are more radiogenic than most Dunedin Volcanic Group samples and more like the Alpine Dike Swarm. Furthermore, recent work on the large exposures of Dunedin Volcanic Group basaltic rocks in the Maniototo region (Scott et al. 2020a; Wilson 2023), which could potentially be a source as the closest Dunedin Volcanic Group rocks (see Figure 2b), has revealed that they were erupted at ∼10–11 Ma, the compositions of many lavas are transitional to sub-alkaline (Figure 6), and although the Sr and Nd isotopic data are comparable to the cobbles, the Pb isotopes are quite distinct (Figure 9a–c). Collectively, the petrographic and geochemical data therefore point to a West Otago Alpine Dike Swarm source for the cobbles.

The samples from Galloway and Upper Clutha are also so similar mineralogically and geochemically that it is unlikely that they are derived from different bodies, and we infer the source to have most likely have been a flow. This is because a single Alpine Dike Swarm dike of typical dimensions of given in Cooper (1986) would be unlikely to generate the volume of basanitic material seen in the Quaternary terraces (as illustrated in Figure 4a). Furthermore, the pilotaxitic textures and almost ubiquitous glassy groundmass to the rocks suggest a rapidly cooled magma with a preferred crystal alignment as would be expected for a flow. The rocks are well rounded and any vesicular lava tops are unlikely to survive the transport distances required given the basanite distribution. Admittedly, the sample suite is biased by our attempts to collect fresh-appearing samples and although we prefer an extrusive origin to the source rocks, this remains a tentative conclusion.

The composition of the rare peridotite xenoliths in the cobbles also supports an Alpine Dike Swarm origin. There are more than 70 peridotite xenolith locations in mainland New Zealand (Scott 2020) and refractory olivine and spinel compositions equivalent to the Galloway-Clutha xenoliths (Figure 10) have only been found in xenoliths from the north Westland (Tulloch and Nathan 1990; Scott et al. 2016b, 2019) and the Alpine Dike Swarm (Wallace 1975; Brodie and Cooper 1989; Scott et al. 2014b; Liu et al. 2015; Scott et al. 2016a, 2019; Shao et al. 2022). North Westland, on the western side of the Alpine Fault, is an improbable source for Central Otago cobbles, and the xenoliths are notably chemically distinct from 99% of the > 200 peridotites analysed from the Dunedin Volcanic Group (McCoy-West et al. 2013; Scott et al. 2014a, 2014b; Dalton et al. 2017; Scott et al. 2019; Shao et al. 2021).

Figure 10. Averaged olivine and spinel compositions for the peridotite xenolith in the Galloway and Upper Clutha basanites. Note that the Upper Clutha xenolith with the low olivine Mg# (∼85) relative to primitive upper mantle (89) is not shown; this xenolith requires reaction with an Fe-rich melt. Reference data are from McCoy-West et al. (2013), Scott et al. (2014a, 2014b; 2016a, 2016b) and Dalton et al. (2017). ADS, Alpine Dike Swarm; DVG, Dunedin Volcanic Group.

Implications for Pleistocene drainage

Our geochemical data favour derivation from the northwest, as previously inferred for river drainage by Youngson et al. (1998) and Craw et al. (2012, 2013). However, a probable precise source location remains unclear. To date, Alpine Dike Swarm basanite has only been found in the West Wanaka diatreme (Perry 2022), which is situated in an area surrounded by dikes and diatremes of lamprophyre. Since the Galloway location lacks lamprophyre, at least at the scale we sampled, it is unlikely this diatreme was the source. Furthermore, most of the basanites in the diatreme are heavily contaminated by Otago Schist material (Perry 2022).

We infer that basanite cobbles in the upper Clutha valley are derived from the Lake Hawea-Lake Wanaka-Luggate area (Figure 11a,b). The basanite cobbles are invariably accompanied by greywacke cobbles that do not come from exposed rocks in the current Clutha River catchment and are abundant in that same general area (McDonnell and Craw 2003). Extensive transport of greywacke cobbles from the north and northeast into what is now the upper Clutha area occurred during the Pliocene and early Pleistocene (McDonnell and Craw 2003), and these gravels may have buried the basanite flow (Figure 11a). The Ahuriri River may have originally flowed southwards into the what is now the Lindis River and upper Clutha valley and brought greywacke into Otago (Figures 1b, 2a, and 11b; Craw et al. 2012). This also permitted galaxiid fish connections between Canterbury and Otago, which ceased ∼1 million years ago as intervening mountain ranges rose, according to the DNA record (Figure 11b; Craw et al. 2012).

Figure 11. Schematic summary of basanite flow source and redistribution over time. (a) Sketch paleogeographic map of Otago Schist basement in northwest and central Otago in Late Oligocene-Early Miocene times. White outlines of modern Lakes Wanaka and Hawea are overlain to show location and scale, as in Figures 1b and 2a. The inferred location of the basanite flow is indicated in relation to the coeval diatremes and associated subsurface lamprophyre dikes of the ADS. A paleodrainage divide separated NW Otago from Central Otago rivers (Craw et al. 2013). Lowlands of the southeastern portion of the area, probably including parts of the basanite flow, were covered with greywacke-bearing gravels derived from the north in Pliocene and Early Pleistocene times. (b) Oblique modern view (GoogleEarth) looking northwest from Manuherikia valley to the ADS and inferred basanite source, with post-Pliocene drainage changes indicated. Middle to late Pleistocene glacial advances almost to Cromwell (Turnbull 2000) may have diverted the Clutha River to its present course.

There was a drainage divide between central Otago and northwest Otago in the Miocene (Figure 11a) and there are no basanite or lamprophyre cobbles in the quartz-dominated Miocene fluvial deposits of central Otago (Douglas 1986; Craw et al. 2012, 2013). Likewise, no basanite cobbles occur in more distal portions of the quartz-dominated central Otago paleo-rivers that reached the Late Oligocene-Early Miocene paleo-shoreline in eastern Southland, or in younger deposits recycled from these sediments in Southland (Stein et al. 2011; Craw et al. 2015). Any late Oligocene-Miocene erosional debris from the inferred basanite flow(s) must have been transported towards western Southland with lithic schist debris (Figures 1a and 11a; Craw et al. 2012; Upton and Craw 2016), although no evidence for such basanite debris has yet been found.

The Manuherikia River catchment that the Galloway gravels sit in is confined between the Dunstan Range to the west and Rough Ridge to the east and is therefore currently isolated from the probable source area (Figures 1b and 11a,b). However, these ranges, as well as the Pisa Range to the west, have grown progressively since the middle Pleistocene (Jackson et al. 1996; Bennett et al. 2006; Craw et al. 2012, 2013). For example, uplift of the Raggedy Range was initiated at ∼1.2 Ma (Bennett et al. 2006) and similar timing has been proposed for the Dunstan Range (Craw et al. 2013). Hence, today's Manuherikia River catchment is more confined by topographic features than it was in the early Pleistocene and preceding times (Craw et al. 2012, 2013). The ancestral Clutha River may therefore have flowed across the present location of the Dunstan Range, possibly through the structurally-controlled low point at Thomsons Saddle, to transport basanite cobbles from the NW into the Galloway area (Figures 2a and 11b). This ancestral Clutha River course was abandoned in the middle-late Pleistocene, probably because glacial advances in the upper Clutha valley (Turnbull 2000) diverted the river to its present course through Cromwell Gorge (Figure 11b; Craw et al. 2012, 2013). A consequence was that this halted the supply of basanite cobbles into the Manuherikia catchment.

All the known occurrences of the basanite cobbles are in gravel deposits that are up to a million years younger than the ancestral Clutha River that we suggest carried the basanites to Galloway. The present hosting gravels at Galloway, for example, were formed at ∼350 ka (Turnbull 2000) and their terrace morphology and internal imbrication are consistent with deposition by an ancestral Manuherikia River (Figure 4c-e). The basanite cobbles have therefore likely been recycled from older gravel deposits as the Manuherikia valley evolved following diversion of the ancestral Clutha. Likewise, repeated recycling of quartz and greywacke clasts and detrital gold has been documented for Manuherikia Valley gravels through the middle-late Pleistocene as the Dunstan Range was rising (Craw et al. 2013), and in sedimentary evolution of the area going back at least to the Miocene (Youngson and Craw 1996).

The oldest basanite-bearing gravels in the upper Clutha valley are < 0.5 Ma (Turnbull 2000), which means that some or all of these clasts have been recycled over the past million years from ancestral Clutha River gravels, facilitated by uplift of valley-bounding topography and several glacial events. Greywacke clasts, which occur in all gravels in the Upper Clutha valley, have undergone similar repeated recycling events since the source(s) outside the present catchment was curtailed (Turnbull 2000; McDonnell and Craw 2003; Craw 2013). The greywacke and basanite clasts at the Middleton Road locality (Q6; ∼150 ka) were, for example, likely derived from erosion of older gravels, including the nearby and up-slope ∼450 ka (Q12) Quartz Reef Point locality (Figure 3a). At least two older greywacke-bearing gravel deposits occur up-slope of the Quartz Reef Point locality (Turnbull 2000) and these may have contributed clasts to younger deposits. Irrespective of the complexities of middle-late Pleistocene fluvial and glacial recycling processes in the upper Clutha Valley, direct connection between basanite occurrences and their primary source(s) may have been severed by up to a million years.

Conclusions

Basanitic cobbles occurring in Pleistocene terraces in the Galloway area near Alexandra are geochemically indistinguishable from basanites cobbles found in the Upper Clutha valley. Neither location is proximal to a known source, but the occurrence of these alkaline basanites with distinctive Sr-Nd-Pb isotope ratios, as well as a 24 Ma age, strongly points to an association with the Alpine Dike Swarm located in the Southern Alps. Furthermore, the Otago area contains several chemically distinct mantle reservoirs and the peridotite mantle xenoliths in the basanite cobbles are most similar to the ultra-refractory mantle that underlies the Southern Alps in northwest Otago. The petrological characteristics of the cobbles indicate that the source was a basanitic lava flow (or flows), a feature which has not previously been documented in the Alpine Dike Swarm. The inferred source area is somewhere in the Lake Wanaka-Hawea-Luggate area, although it is probably now either fully eroded or buried beneath Pleistocene and younger gravels. This inferred source expands the extent of that magmatic field southeast of the known Alpine Dike Swarm and also requires that there was emergent land at the time of eruption. A northwest Otago source means that the ancestral (early-middle Pleistocene) Clutha River carried cobbles of basanite across the site of the present Dunstan Range and then down the (present-day) Manuherika Valley. This pathway was then abandoned in the middle-late Pleistocene, when the Clutha River moved to the south and formed the modern Cromwell Gorge. The cobbles have likely been reworked several times, to account for them now to be stranded in < 0.5 Ma terraces.

Acknowledgements

We thank Stan Szczepanski for his expertise in the AuScope-supported ⁴⁰Ar/³⁹Ar dating laboratory at The University of Melbourne. Discussions with John Youngson over many years was stimulating and helped to develop some of the ideas in this paper. John Smellie and Tod Waight are thanked for constructive reviews, and Sebastian Naeher is thanked for handling the manuscript.

Disclosure statement

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

Data availability statement

All data for this paper are available in tables freely hosted at: https://doi.org/10.6084/m9.figshare.22679827.

Additional information

Funding

This study was funded by University of Otago.