Petrogenesis of the Hoy lava field, a long-lived continental mafic volcanic province in eastern Australia

Abstract Small-scale continental basaltic fields can erupt with little warning and bring deep undegassed magmas to the surface rapidly. To explore the lifetime, petrogenesis and plumbing system architecture feeding such basaltic lava fields and compare them with large-scale shield volcanoes, we have focused on the Hoy lava field, central Queensland, Australia. 40Ar/39Ar geochronology, elemental and isotopic whole-rock geochemistry and mineral chemistry on selected Hoy samples reveal long-lived volcanism of ca 50 Ma and magma storage at mantle depths, notably different from the comparatively short duration (3–5 Myr) and crustal magma storage depths of shield volcanoes. In this study, four Hoy lava-field eruptive intervals spanning ca 50 Ma were investigated: 67.5 ± 0.3 Ma, 32.3 ± 0.6–31.6 ± 0.7 Ma, 21.9 ± 0.5 Ma and 18.1 ± 0.3 Ma. In all four eruptive events, samples are porphyritic alkali basalts and trachybasalts (11.41–6.45 wt% MgO) with incompatible element concentrations and Sr–Nd–Pb isotope ratios dominantly derived from a metasomatised sub-continental lithospheric mantle (SCLM) source with an enriched mantle I (EMI) signature. Complex crystal populations show major- and trace-element variations reflecting fractional crystallisation, magma recharge, magma mixing and mantle xenocryst entrainment. Clinopyroxene–melt thermobarometry indicates magma storage in SCLM reservoirs at ∼30–47 km depths. The nearby larger but shorter-lived (3–5 Myr) Buckland central volcano has similar source compositions; however, magma storage is limited and concentrated in the crust, resulting in increased crustal contamination. The results suggest that basaltic centres of contrasting scale and longevity are linked to distinct magma production mechanisms, fluxes, ascent and differentiation. KEY POINTS The Hoy lava field erupted for over 50 Ma, with at least four eruptive periods. The Hoy magmas are all derived from the same source: SCLM with an EMI signature. Hoy lavas have a complex history of magma recharge, magma mixing, fractionation and xenocryst entrainment at mantle depths (30–47 km). Hoy lavas experienced deeper storage, limited contamination and faster ascent than the nearby Buckland central volcano.


Introduction
Small-scale basaltic magma fields are one of the most common types of volcanic systems on Earth (e.g.Canon-Tapia & Walker, 2004).They occur as spatially dispersed, smallvolume volcanoes that erupted, episodically, for over millions to tens of millions of years (Connor & Conway, 2000;Smith & N emeth, 2017;Takada, 1994).Magma petrogenesis and eruption triggers in these systems are not well understood (e.g.Cas et al., 2017 and references therein).
Crucially, many of these small-scale continental volcanic systems are located close to populated areas (e.g.Brenna et al., 2018); better constraining eruption frequency, magma generation, storage and rapidity of transport to the surface are important for assessing volcanic hazards.
Eastern Australia provides an ideal setting to investigate continental volcanism because it hosts a wide variety of eruptive systems, namely central volcanoes, lava fields and leucitites (Figure 1a; Wellman & McDougall, 1974).Central volcanoes are remnant shield volcanoes composed of both mafic and silicic rocks that young southward from ca 34 to 6 Ma (Wellman & McDougall, 1974).Leucitites are leucite-bearing basalts consisting of minor cones and flows that also young southward (Cohen et al., 2008;Johnson, 1989).The age-progressive central volcanoes and leucitites are interpreted to result from Australia's migration over a mantle plume or plumes (Davies et al., 2015;Fishwick et al., 2008;Knesel et al., 2008;Sutherland, 2003;Vasconcelos et al., 2008;Wellman & McDougall, 1974;Zhang & O'Reilly, 1997).The lava fields, our focus, are generally small-scale volcanic fields dominated by mafic lava flows, cones and plugs.In contrast with the central volcanoes, lava fields do not show any simple age progression (Johnson, 1989;Wellman & McDougall, 1974).The potential origin of the lava fields includes: rift-related decompression melting (Cohen, 2007;Johnson, 1989); far-field stresses that induced decompression melting (Cas et al., 2017); edgedriven convection (e.g.Davies & Rawlinson, 2014;Demidjuk et al., 2007;Shea et al., 2022); heat flow from the warm Pacific mantle (known as the diffuse alkaline magmatic province model, DAMP model; Finn et al., 2005); mantle swells (Sutherland et al., 2012); and bursts in slab flux (Mather et al., 2020).The close spatial, temporal and compositional relationship between central volcanoes and lava fields provides key information about the underlying mechanisms that control the formation of such diverse volcanic types (e.g.Jones et al., 2020).
The petrogenetic evolution of the lava fields located in north Queensland, New South Wales, and the Newer Volcanic province in Victoria and South Australia have been the focus of numerous investigations (e.g.Boyce, 2013;Cas et al., 2017;O'Reilly & Zhang, 1995;Van Otterloo et al., 2014;Zhang et al., 2001;Zhang & O'Reilly, 1997).These studies have shown that the dominant source components have OIB-like signatures derived from the sub-continental lithospheric mantle (SCLM), while the erupted magmas are complex and involve multiple magma batches.Curiously, the Hoy lava field province in central Queensland (Figure 1b), known to have erupted intermittently for over 50 Ma (e.g.Cohen, 2007;Lafferty & Golding, 1985;Robertson & Sutherland, 1992;Stephenson et al., 1989), lacks detailed petrogenetic studies.The Hoy lava field is spatially and, in part, temporally associated with the larger age-progressive Buckland central volcano (Figure 1a), whose ages and plumbing mechanisms have been studied in detail (Crossingham et al., 2018a;Shea & Foley, 2019;Skae, 1998;Tapu et al., 2023;Waltenberg, 2006).Investigating the Hoy lava field, particularly the centres erupted simultaneously with the Buckland central volcano, may provide key information about mantle conditions during magma generation and extraction, as well as the nature of the plumbing mechanisms that delivered magmas to the surface.
Here we combine 40 Ar/ 39 Ar geochronology with wholerock major and trace-element geochemistry and radiogenic isotope ratios, as well as major-and trace-element chemistry at the mineral scale, to investigate source-to-eruption processes in selected Hoy volcanic centres.We contrast our results with those from the nearby Buckland central volcano to compare the source of magma and the architecture of the plumbing system across the eastern Australian volcanic systems.

Geological background
The Hoy lava field, located on the western edge of the central volcanoes in central Queensland, Australia (Figure 1a), has a diameter of about 50 km and contains at least 70 alkali basalt plugs, cones and minor lava flows (Figure 1b; Stephenson et al., 1989).The plugs range from low rises to steep cones and dominantly trend north-northeast (Robertson & Sutherland, 1992).The largest plug is known as Mount Leura, which is approximately 400 m in diameter (Stephenson et al., 1989).
The Hoy province is located on moderately thick, $100 km, lithosphere (Davies et al., 2015).The basement rocks include parts of the Drummond Basin, Retreat Granite and the metasedimentary rocks of the Anakie Inlier.Most of the plugs, cones and flows in the northern part of the province are located within the Retreat Granite, specifically the mid Devonian Mount Morgan Trondhjemite, and to a less extent the Neoproterozoic to lower Cambrian Wynyard Metamorphics of the Anakie Metamorphics (Figure 1).The plugs, cones and flows in the southern part of the province are partially separated by northwest-trending faults and overlie the Carboniferous Ducabrook Formation of the Drummond Basin (Figure 1).The area contains several regional and localised structures (Figure 1) that could have acted as conduits for the rising magmas of Hoy.

Samples
The samples used in this study are five fresh samples collected from different plugs in the Hoy lava field (Figure 1b).These samples were previously dated by Cohen (2007) and are being redated here with the aim to improve precision.The samples are fresh and porphyritic, and host variable macrocryst assemblages (>0.2 mm), as outlined below.The term macrocryst is used here in place of the more traditional 'phenocryst' to avoid genetic implications.The rock groundmass consists of variable volume fractions of plagioclase, olivine, clinopyroxene and titanomagnetite.Thin-section images are provided in Supplemental data (Appendix 1).
Sample BC-83 was collected from the summit of Mount Llandillo and contains macrocrysts ($22 vol%) of olivine, three different types of clinopyroxene and rare plagioclase.The clinopyroxene types include: clear, resorbed, cores surrounded by sieved, reddish clinopyroxene (Type 1); reddish, sometimes sieved, subhedral, clinopyroxene crystals (Type 2); and resorbed greenish/brown cores with reddish rims (Type 3).BC-79, collected from the summit of Mount Leura, has the same variety of macrocrysts ($23 vol%) as sample BC-83.Sample BC-84 comes from the northeastern side of Black Mountain and contains macrocrysts ($44 vol%) of olivine and rare clinopyroxene with green cores and red rims (Type 4), differing from all other types of clinopyroxene.Sample BC-78, from the north side of Policeman's Knob, contains only olivine macrocrysts ($10 vol%), while sample BC-76, from Anakie Hill, has macrocrysts of olivine and two types of clinopyroxenes.The first is reddish, with sieved cores and subhedral to euhedral rims, similar to Type 2 clinopyroxenes.The second includes reddish glomerocryst 'flowers' (Type 5) unique to this sample. 40Ar/ 39 Ar geochronology The five mafic samples were crushed to 1-2 mm grains.The grains were sonicated in distilled water and ethanol.The freshest, most equant whole-rock grains from all samples were loaded into a 21-pit Al disk with the neutron fluence monitor Fish Canyon sanidine (age of 28.201 ± 0.04 Ma; Kuiper et al., 2008) and additional grains of GA-1550 biotite as secondary standards (98.5 ± 0.8 Ma; Spell & McDougall, 2003), following the geometry illustrated in Vasconcelos et al. (2002).
The disks were irradiated for 14 hours in the Cd-lined (CLCIT) facility at the Oregon State University TRIGA-type reactor.Ten whole-rock grains were incrementally heated using a defocussed Coherent Verdi diode laser beam and the purified gas fractions were analysed in the MAP215-50 mass spectrometer installed at The University of Queensland Argon Geochronology in Earth Sciences Laboratory (UQ-AGES) following procedures similar to those of Vasconcelos et al. (2002).

Bulk rock major and trace elements
Five whole-rock samples were powdered in an agate ring and puck mill.Major and trace elements and loss on ignition (LOI) were measured in the Geochemistry Laboratory and the Radiogenic Isotope Facility (RIF) at the School of Earth and Environmental Sciences, The University of Queensland.
Whole-rock sample powders (0.1 g) were placed into platinum crucibles with 0.4 g of lithium metaborate flux and fused at $1000 C in a Katanax Automatic Fluxer.The resulting glass bead was dissolved in 100 mL of 5% nitric acid.A Perkin Elmer Optima 8300DV inductively coupled optical emission spectrometer (ICP-OES) was used to measure the samples, standards (BHVO-2, JA2, JA3, JB2, JGb1), repeats, duplicates, monitors and blanks.Lu, Sc and Y were used for quality and drift corrections.Multiple analyses of standard glass BHVO-2, JA2, JA-3 and JB3 were used to determine accuracy (within 3%) and precision [within 5%, except for P 2 O 5 in BHVO2 (23%) and JA3 (18%), and K 2 O in JA3 (6%)].LOI was determined by placing 2 g of sample into a crucible heated at 105 C for 1 h.
Samples were prepared for trace-element analysis by multi-acid open-beaker hotplate digestion.The sample powder (100 mg) and nitric acid (2 mL) were sealed in Teflon beakers and heated at 140 C overnight.After drying, hydrofluoric acid (3 mL) and nitric acid (1 mL) were added, and the beakers were heated to 140 C overnight.The sequence was repeated using hydrochloric acid (6 mL).Following the final dry-down of the sample, 2 mL of 6 N nitric acid was added and heated for 2 hours.The solution was brought up to 10 mL with milli-Q water.Samples, duplicates, repeats and standards were analysed on a Thermo X Series inductively coupled plasma mass spectrometer (ICP-MS).The accuracy of the trace-element analysis was monitored using standard W2a and was within 10%, except Th (88%), Zr (88%) and Y (87%).Duplicates of unknown samples BC-79 and BC-83 showed a ±1% precision.

Mineral chemistry
Major-element mineral compositions were determined by electron microprobe using a JEOL JXA-8200 microprobe equipped with five wavelength-dispersive spectrometers at the Centre for Microscopy and Microanalysis, The University of Queensland.Analyses were performed using an accelerating voltage of 15 kV, a beam current of 15 nA and a fully focused beam ($2 Â 2 mm interaction surface).Counting times for all elements were 30 s on the peak and 5 s on each of two background positions.A ZAF method was used for matrix correction.Calibration of major and minor elements used the following standards: orthoclase (K), albite (Na and Al), wollastonite (Si and Ca), chromite (Fe and Cr), spessartine (Mn), F-apatite (P), rutile (Ti), P-140 olivine (Mg) and Ni-olivine (Ni).VG2 was used as a quality monitor; for major elements with concentrations of >1 wt%, the percentage RSD was within 3%, and the accuracy relative to accepted values was within 4% (Supplemental data, Appendix 1).
Trace-element mapping of eight pyroxene crystals, using the rastering technique developed by Ubide et al. (2015), was carried out by laser ablation-inductively coupled plasmamass spectrometry (LA-ICP-MS) at the RIF, School of Earth and Environmental Sciences, The University of Queensland.An ASI RESOlution 193 nm excimer UV ArF laser ablation system with a dual-volume Laurin Technic ablation cell was coupled to a Thermo iCap RQ quadruple mass spectrometer.Details on the instrument set up and gas flows are provided in Ubide, Mollo, et al. (2019).The system was tuned on NIST612 glass reference material.NIST610 glass was used as a calibration standard, and Ca concentrations determined by electron microprobe (21.7 wt% CaO in clinopyroxenes from samples 20.8 wt% CaO in sample BC-83) were used as internal standards.Accuracy and precision were monitored measuring glass standards BHVO-2G and BCR-2G during the runs.Typical precision was within 5% and accuracy within 10-15%.We used Iolite software v2.5 (Paton et al., 2011) to build element maps, and the Monocle add-on (Petrus et al., 2017) to extract average compositions for individual crystal zones.

Bulk rock radiogenic isotopes
The five samples were further analysed for Pb, Nd and Sr isotope ratios at the RIF, School of Earth Sciences, The University of Queensland.Samples were prepared following the method outlined in Crossingham et al. (2018a).A VG Sector 54 thermal ionisation mass spectrometer was used to measure the Sr isotopes, which were then calibrated using SRM-987 (N ¼ 15), which yielded an average of 0.710225 ± 0.000014, normalised to 0.710249.A Nu Instruments multi-collector-inductively coupled plasmamass-spectrometer (MC-ICP-MS) measured the Pb and Nd isotopes.Pb isotopes were fractionation-corrected by doped thallium (Tl) using a value of 2.3882 for the 205 TI/ 203 TI ratio and compared with the standard reference material SRM0981.Nd isotopes were calibrated using Nd metal, which yielded an experimental average (N ¼ 5) of 143 Nd/ 144 Nd ¼ 0.511958 ± 0.000007 and calibrated to JNDi (0.511966).

Results
40 Ar/ 39 Ar geochronology A summary of 40 Ar/ 39 Ar results for 10 whole-rock fragments from five samples from the Hoy lava field is provided in Figure 2 and Table 1.The complete set of results is presented in the Supplemental data (Appendix 3).
Most whole-rock grains produced plateau ages (Table 1).All plateau ages are within error of the ages produced by the combined isochron and age-probability diagrams.We used the combined inverse isochrons for interpreting geochronological results for all samples except Policeman's Knob (BC-78; Figure 2), as inverse isochrons combine data for all individual steps analysed for a given sample into a single age and allow identification and mitigation of excess argon, if present (e.g.BC-84: 40 Ar/ 36 Ar intercept of 410 ± 80).The isochron for sample Policeman's Knob (BC-78) did not converge because all steps contain highly radiogenic argon and cluster at the right-hand side of the curve; therefore, we used the probability density plot age (Figure 2) for the combined grains from this sample. 40Ar/ 39 Ar geochronology results for the five samples collected from the Hoy lava field identify four eruptions spanning ca 50 Ma: 67.5 ± 0.3 Ma, 32.3 ± 0.6-31.6 ± 0.7 Ma, 21.9 ± 0.5 Ma and 18.1 ± 0.3 Ma.Our results are within error to those of Cohen (2007).We were, however, able to improve the precision on Policeman's Knob (BC-78; 67.5 ± 0.3 Ma) relative to the previous age of 66 ± 1 Ma (Cohen, 2007).

Bulk rock elemental and isotope geochemistry
The complete set of bulk rock and isotope geochemistry results are presented in the Supplemental data (Appendix 4).The Hoy lavas range from 46.7 to 50.2 wt% SiO 2 and from 6.7 to 11.7 wt% MgO, classifying them as alkali basalts to trachybasalts (Figure 3).The chondrite-normalised rare earth element (REE) and primitive mantle-normalised multi-element diagrams show there is no significant compositional variation among the various Hoy lavas (Figure 4).All lavas are characterised by alkaline signatures with light rare earth element (LREE) enrichment over the heavy rare earth elements (HREEs) (Figure 4a).On the multi-element diagram, there are notable negative anomalies for K and Pb (Figure 4b), and the samples show signatures similar to enriched mantle I (EMI) and high time-integrated 238 U/ 204 Pb mantle reservoirs (HIMU; Zindler & Hart, 1986).
Isotope ratios (Sr, Nd and Pb) for the five samples are presented in the Supplemental data (Appendix 4) and Figure 5. Isotopic compositions are similar, regardless of age.However, the lavas of Mount Leura (21.9 ± 0.5 Ma) have slightly less radiogenic Pb values than the other samples (Figure 5).Our radiogenic isotope results are similar to previously analysed samples from the Hoy region, including samples from Mount Llandillo and Black Peak (Figure 5; Ewart et al., 1988).The samples also show a partial overlap with some of the Buckland lavas, which trend towards increased crustal contamination (Crossingham et al., 2018a).

Mineral chemistry
A summary of mineral chemistry results is presented in Figures 6 and 7. Complete mineral chemistry data are presented in the Supplemental data (Appendix 2).The extended version of Figure 7 and the complete set of trace-element laser ablation ICP-MS maps are presented in the Supplemental data (Appendix 5).

Olivine
There are two main types of olivine macrocrysts.Mantle xenocrysts have forsterite content Fo > 90, calcium concentrations CaO <0.15 wt% (e.g.Sakyi et al., 2012 and references therein) and plot above equilibrium on mineral-melt equilibria models (Rhodes et al., 1979).Magmatic olivine have lower Fo and higher CaO values.Mantle xenocrysts occur in most samples, except Mount Leura (Figure 6a, b).Magmatic olivine are present in all samples, excluding Policeman's Knob (Figure 3a, b).Resorbed, subhedral to anhedral magmatic cores range from Fo 87 to Fo 86 for Mount Llandillo, while euhedral magmatic cores range from Fo 86 to Fo 76 for Black Mountain (Figure 6a, b).Rims are similar in both samples (Fo 78-73 ).The subhedral to anhedral olivine cores from Mount Leura and Anakie Hill range from Fo 74 to Fo 69 and Fo 85 to Fo 79 , respectively (Figure 6a, b).The rims of Mount Leura range from Fo 68 to Fo 59 and Anakie Hill from Fo 81 to Fo 79 .Groundmass olivine microcysts range from Fo 76 to Fo 62 across all samples (Figure 6a, b).where concentrations are expressed on a molar basis, and Fe 2þ is 90% total Fe), TiO 2 0.38-0.1 wt%, high Cr and Ni, and low Sc, V, Zr, La and Nd relative to other compositions (Figures 6 and 7), suggesting that they are mantle xenocrysts (e.g.Griffin et al., 1987;Roach, 2004).The Al 2 O 3 concentration is distinct between the two samples (Figure 6c).The surrounding sieved, reddish rims have low Mg# (73-79) and are depleted in Cr, Ni, V, Zr, La and Nd, and enriched in Sc (Figure 6c, d).A small area of enrichment occurs within the low Mg# zoned of Mount Llandillo (Mg# ¼ 81; enriched in Cr, Sc, Ni, Zr and Nd) (Figures 4 and  6c, d).
Type 3 are reversely zoned clinopyroxene crystals present in Mount Llandillo and Mount Leura samples.The brownish-green cores in Black Mountain have 68-69 Mg#  A plateau age comprises at least 50% of the total 39 Ar released.A plateau also includes three or more consecutive steps, where the age values overlap within a 95% confidence interval (Fleck et al., 1977).The mean weight by inverse variance and the errors, which include errors in irradiation corrections factors and errors in J, are used to calculate the plateau age.Ages are reported at the 95% confidence level.
Error-overlap, with a 2-sigma error, is used to define a plateau. b The assumption that a Gaussian distribution occurs for the errors in an age is used to construct a probability density plot.When each age is plotted, the totals for every Gaussian curve are taken (Deino & Potts, 1990). c Isochron age errors include irradiation correction factors and the errors in J.The uncertainty in the potassium decay constant is excluded.Isochron ages are measured to the 95% confidence level (2r).6c, d and 7), suggesting that they are mantle xenocrysts (e.g.Griffin et al., 1987;Roach, 2004).They are overgrown by reddish rims with Mg# 81-89 (TiO 2 1.6-2.1 wt%) and depleted in Cr and Ni (Figures 6c, d and 7).
Type 5 clinopyroxenes are sector-zoned, reddish 'flower'shaped aggregates present only in samples from Anakie Hill lavas.Crystal sectors were identified based on the morphological model of titanoaugite (Leung, 1974  sectors have Mg# ranging from 72 to 76 (TiO 2 2.76-3.81wt%; Figure 3c, d).The hourglass sectors have Mg# ranging from 78 to 79 (TiO 2 2.76-3.81wt%; Figure 3c, d).These sector-zoned crystals show no evidence of multiple magma injections but because trace-element analysis was aimed at characterising multiple magma injections, we chose to allocate the limited laser time to other samples and did not measure the trace-element compositions of Type 5 clinopyroxenes.

Discussion
The Hoy lavas have high MgO (6.7-11.7 wt%) and Cr (131-749 ppm) contents, which partly reflect accumulation of mafic macrocrysts (Ubide et al., 2022) but may also retain evidence of their mantle sources.Whole-rock geochemistry shows little variation between samples over the four eruptive periods investigated here: 67.5 ± 0.3 Ma, Policeman's Knob; 32.3 ± 0.6-31.6 ± 0.7 M, Black Mountain and Mount Llandillo; 21.9 ± 0.5 Ma, Mount Leura; and 18.1 ± 0.3 Ma, Anakie Hill.However, mineral chemistry reveals significant differences among the olivine and clinopyroxene macrocrysts in the porphyritic samples.The limited number of volcanic edifices sampled and dated in this study is insufficient to determine whether the Hoy lava field is episodic or whether it hosts a continuous history of eruptions.However, the large time span represented by the samples, the contemporaneity of one of the Hoy eruption periods with the eruption of the nearby Buckland volcano, and a carefully age-constrained volcano stratigraphy (Crossingham et al., 2018a), permits comparison and contrasting of magma genesis and emplacement mechanisms between lava fields (Hoy) and central volcanoes (Buckland).In the following sections, we assess the role of magma sources, lithospheric contamination and differences in the plumbing systems of the various Hoy lavas.We then compare the small-scale Hoy lava field with the spatially and temporally associated but much larger Buckland central volcano.

Crustal contamination and magma source characteristics
Upper and lower crust show positive K and Pb anomalies, and negative Nb and Ti anomalies on primitive mantle-normalised multi-element diagrams (Figure 4b).The Hoy lavas, in contrast, show negative K and Pb anomalies and enriched values for Nb and Ti with respect to the primitive mantle (Figure 4b).Similarly, the Hoy lavas have Nb/U values of 37.94-46.72 and Nb/Th values of 12.64-15.75,which fall within the global average compositions of uncontaminated mid-ocean ridge basalts (MORB) and ocean island basalts (OIB) (Nb/U ¼ 47 ± 10 and Nb/Th 10-20; Hofmann et al., 1986;Hofmann, 2004).Crustal contamination is assessed by increasing 87 Sr/ 86 Sr values with decreasing MgO and increasing 1/Sr values (e.g.Barry et al., 2003).The data do not define linear trends following these tests.The least radiogenic 87 Sr/ 86 Sr (BC-76) sample has the lowest MgO values, which may suggest minor crustal contamination, but this sample does not have the highest 1/Sr value (Supplemental data, Appendix 6).This suggests that the Hoy lavas have not been significantly contaminated by the crust.
The depth and degree of melting of the Hoy lavas were assessed using REE and shape coefficients of chondrite-normalised REE patterns following O'Neill (2016).The incompatibility of light REE in garnet and spinel compared with the compatibility of the HREE in garnet (D > 1), but not spinel (D < 0.001), can be used to provide information on the depth of melting (McKenzie & O'Nions, 1991).All Hoy lavas show LREE enrichment with respect to HREE [(La/Yb) N ¼ 11.49-17.05;(Sm/Yb) N ¼ 3.35-5.25],suggesting a garnetbearing source.Chondrite-normalised REE shape coefficients (after O'Neill, 2016) show that the Hoy lavas plot within the ocean island basalt field, further indicating a garnet-bearing source (Figure 8a, b).
The Hoy lavas have high k 1 values (Figure 8a, b), which further indicate LREE/HREE enrichment and suggest low degrees of partial melting.Non-modal batch melting models, used to assess the degree of melting at the Buckland central volcano (Crossingham et al., 2018a), suggest less than 5% partial melting for the Hoy samples (Figure 8c, d).Thus, we conclude that the Hoy lavas formed from low degrees of partial melting (<5%) in the garnet stability field.
Radiogenic isotopes reveal that the Hoy lavas have both an enriched and a depleted source component.The isotopic values suggest that the depleted source is Indian MORB, as Pacific MORB is ruled out by the higher 87 Sr/ 86 Sr and lower 143 Nd/ 144 Nd values (Figure 5).Previous studies have suggested that the shallow asthenosphere is a likely source component for the lava fields of eastern Australia (O'Reilly & Zhang, 1995;Sun et al., 1989;Zhang et al., 2001).Some of these studies have proposed Indian MORB as a potential source, especially for the Queensland lava fields (e.g.Jones et al., 2020;Van Otterloo et al., 2014;Zhang et al., 2001).
In turn, Sr-Nd isotopes suggest either HIMU or EMI as potential enriched sources for the Hoy lavas (Figure 5a).Multi-element patterns also agree with an EMI or HIMU source (Figure 4b).Lead isotope ratios, however, are relatively low and do not suggest a dominant HIMU component, agreeing with an EMI signature (Figure 5b-d).
Enriched mantle source components are commonly interpreted as indicative of a deep source (e.g.Hart, 1988;Hofmann, 1997) or a metasomatised SCLM source.In fact, SCLM has been proposed as a potential source component for the Cenozoic volcanic rocks of eastern Australia (Crossingham et al., 2018a;Ewart et al., 1988;Jones et al., 2020;O'Reilly & Zhang, 1995;Shea et al., 2022;Zhang & O'Reilly, 1997;Zhang et al., 2001).Negative K anomalies indicate the presence of residual K-rich phases in the source (Class & Goldstein, 1997;Panter et al., 2006).Moreover, negative Ba and Rb anomalies, relative to Nb, suggest that the K-rich phases may be either amphibole and/or phlogopite (Furman & Graham, 1999;La Tourrette et al., 1995).Importantly, the presence of these K-rich phases in the source suggests the involvement of the SCLM, as neither amphibole nor phlogopite is stable in the convecting mantle (Class & Goldstein, 1997).Nonmodal batch melting models also support the presence of amphibole and phlogopite in the source (Figure 8c, d).Furthermore, the negative Pb anomalies observed in the Hoy basalts are also evident in other mafic samples in eastern Australia, including the late Cenozoic Cooktown nephelinites and Old lavas at Buckland (Crossingham et al., 2018a;Shea et al., 2022;Zhang et al., 2001).The negative Pb anomalies have been attributed to Pb loss because of shallow fluid-loss during Paleogene subduction, preserved in the SCLM (Shea et al., 2022;Zhang et al., 2001).The subduction zone may have introduced the sediments and recycled crust that resulted in carbonatite to silicate melts that contributed to metasomatism of the SCLM, giving it an EMI-like signature (Shea et al., 2022).Therefore, we interpret from this that EMI signatures in the Hoy lavas are indeed derived from the metasomatised SCLM.

Magma plumbing systems
The presence of mantle xenocrysts and the lack of complexly zoned crystals in the lavas from small-scale volcanic fields have been interpreted as evidence of a direct connection from the source to the surface (e.g.Griffin et al. 1987;Jankovics et al., 2015;Re et al., 2017;Van Otterloo et al., 2014).Once xenoliths have been entrained in the magma, magma ascent is generally interpreted to be continuous and rapid (e.g. 10 À2 to 10 ms À1 ; Lensky et al., 2006;Spera, 1980;Szab o & Bodnar, 1998).However, recent studies on the chemical stratigraphy of minerals have shown that the ascent and storage of magmas in these systems are more complex than previously thought and may include periods of storage in the upper mantle and crust (Brenna et al., 2018;Jankovics et al., 2015).
then used for thermo-barometric analysis using the model of Putirka et al. (2003), appropriate for alkaline magmas with calibration errors of 34 C and 1.7 kbar, to estimate the depth of magma storage for all volcanoes, except Policeman's Knob (67.5 ± 0.3 Ma), where clinopyroxene macrocrysts are not present.

Policeman's Knob
The macrocryst population of Policeman's Knob (67.4 ± 0.3 Ma) is represented solely by olivine xenocrysts of mantle origin (Fo 89-91 ).The presence of olivine xenocrysts and absence of magmatic olivine derived from the melt itself suggest that the Policeman's Knob magmas ascended quickly through the lithosphere, with little to no storage.

Mount Llandillo
The Mount Llandillo (32.3 ± 0.6 Ma) macrocryst population contains both lithospheric mantle-derived xenocrysts and melt-derived olivine and clinopyroxene crystals.Magmatic olivine plots above the equilibrium line in mineral-melt equilibrium diagrams, which suggests that they are primitive antecrysts.Thus, they are crystals recycled from early mafic magmas (Figure 9a; Davidson et al., 2007).The brown/green cores from Type-3 clinopyroxenes have majorand trace-element chemistry distinct from the rest of Mount Llandillo clinopyroxenes (Figures 6c, d and 7) and plot below the equilibrium line (Figure 9b).Thus, they are considered evolved antecrysts.The 'enrichment' zone of Type 1 and Type 2 clinopyroxenes and the high Mg# rims of Type 3 clinopyroxene crystals plot along or slightly below the equilibrium line (Figure 9b) and are considered phenocrysts, that is crystals that have grown in the host magma (Davidson et al., 2007).Barometric estimates from samples that passed the equilibrium tests of Fe-Mg exchange between clinopyroxene and liquid (K d ; Figure 9b), and the comparison between the observed DiHd component in clinopyroxene and that predicted from the composition of the liquid (Supplemental data; Appendix 7, Putirka 2008), indicate a crystallisation depth of 17.6-17.8kbar and 1270-1305 C for the low Mg# red rims of Type 1 clinopyroxene (Figure 10).Barometric estimates of Type 2 clinopyroxene cores indicate a slightly shallower crystallisation depth to Type 1 rims (14.3-15.7 kbar and 1286-1316 C; Figure 10).In contrast, Type 2 clinopyroxene rims Figure 9. Mineral-melt equilibria models for (a) olivine and (b) clinopyroxene from the Hoy lava field.Models using bulk rocks as liquids have the higher Mg#, whereas modelled liquids are more evolved (lower Mg#).The curved lines (Rhodes et al., 1979) show the range mineral-melt equilibrium compositions.The iron-magnesium distribution coefficient K d was 0.30 ± 0.03 for olivine (Roeder & Emslie, 1970) and K d 0.28 ± 0.08 for clinopyroxene (Akinin et al., 2005;Putirka, 2008).Melts were modelled using mass balance calculations, subtracting the average compositions of xenocrystic olivine and clinopyroxene from the bulk rock composition, using mineral volume fractions estimated via point counting (see Supplemental data (Appendix 7) for full calculations and point counting results).
Considering a crustal density of 2.8 g/cm 3 and a mantle density of 3.3 g/cm 3 , we infer the Mount Llandillo magma stalled in a deep mush reservoir ($49 km) that contained rare primitive olivine antecrysts, and crystallised primitive normally zoned (Type 2) clinopyroxene crystals.Olivine, clinopyroxene, and rare anorthite xenocrysts would have been entrained with incoming primitive melts, and the clinopyroxene xenocrysts were overgrown with low Mg# rims.This now crystal-rich magma sampled a more fractionated mush reservoir ($37-40 km) that contained green/ brown clinopyroxene (Type 3) antecrysts.The high Mg# rims surrounding green-brown cores of Type 3 clinopyroxenes and the rims of the Type 2 crystallised upon intrusion into the reservoir.The injection of magma from a deep into a shallower and more fractionated reservoir likely triggered the eruption.Further fractionation occurred during unobstructed, rapid ascent.Mount Llandillo magma erupted at 32.3 ± 0.6 Ma.

Black Mountain
Similar to the Mount Llandillo magmas, magmas from the Black Mountain volcano (31.6 ± 0.7 Ma) contain both olivine and clinopyroxene xenocrysts and magmatic olivine and clinopyroxene.Magmatic olivine are in equilibrium with the modelled melt and are interpreted as melt-derived phenocrysts (Figure 9a).Similarly, Type 2 clinopyroxenes and the low Mg# rims of Type 4 clinopyroxenes plot within the equilibrium field or are fractionated with respect to equilibrium (Figure 9b).This suggests that they are phenocrysts, some of which may have experienced a certain degree of fractionation, possibly during ascent.Barometric estimates show that the low Mg# rims of Type 4 clinopyroxenes that pass equilibrium tests with the carrier liquid crystallised at 8-10.6 kbar (1253-1288 C; Figure 10); in contrast, barometric estimates of Type 2 clinopyroxene cores range from 6.2 to 13.0 kbar (1247-1299 C; Figure 10).
As the Black Mountain magma began to ascend, olivine and Cr-rich clinopyroxene mantle xenocrysts were entrained and the low Mg# rims (Type 4) formed around the clinopyroxene xenocrysts at depths of $40 km.Further Type 2 clinopyroxene and olivine phenocrysts crystallised upon ascent.Black Mountain magmas erupted at 31.6 ± 0.7 Ma.

Mount Leura
Mount Leura (21.9 ± 0.5 Ma) lavas contain only clinopyroxene xenocrysts and magmatic olivine and clinopyroxenes.Magmatic olivine plot well below the equilibrium line (Figure 9a), suggesting that they are fractionated antecrysts.Similarly, the green cores of Type 3 clinopyroxenes plot just below equilibrium (Figure 9b), and have traceelement compositions distinct from the Mg# rich rims (Figures 6 and 7).Therefore, they are probably antecrysts.Low Mg# rims in Type 1 and Type 2 clinopyroxenes, and high Mg# rims in Type 3 clinopyroxenes, plot within equilibrium or just below it, suggesting that they are phenocrysts, with some experiencing fractionation during ascent.(2003).We only considered pairs that passed the equilibrium tests of Fe-Mg exchange between clinopyroxene and liquid (K d ), as well as observed DiHd component of the clinopyroxene composition matching that predicted from the composition of the liquid (Putirka, 2008).Depths were calculated using a crustal density of 2.8 g/cm 3 and a mantle density of 3.3 g/cm 3. The depth of the base of the crust (Moho) and garnet-spinel transition are from Griffin et al. (1987).
Based on barometric estimates and mineral-melt equilibrium diagrams, a reservoir at a depth of $40 km contained fractionated olivine, green clinopyroxene and plagioclase crystals.The mush was recharged by a hotter more primitive magma that carried clinopyroxene xenocrysts.Previous studies on garnet websterite from Mount Leura show that they were derived from mantle pressures of 13.5 kbar (Griffin et al., 1987).The recharging magma promoted the growth of enriched rims and disequilibrium textures in the clinopyroxene and plagioclase crystals, and it drove the formation of the low mg# rims around the entrained xenocrysts.Recharge likely triggered the Mount Leura eruption at 21.9 ± 0.5 Ma.

Anakie Hill
Olivine xenocrysts and magmatic olivine and clinopyroxene macrocrysts make-up the crystal population of Anakie Hill (18.1 ± 0.2 Ma) lavas.The magmatic olivine plots above and within equilibrium (Figure 9a).The olivine crystals that plot above equilibrium are considered primitive antecrysts, and those that plot within equilibrium are considered phenocrysts.Type 5 sector-zoned clinopyroxene crystals plot within equilibrium or are fractioned with respect to equilibrium and are, thus, considered phenocrysts (Figure 9b).Barometric estimates show that they crystallised at shallower depths (5.0-11.1 kbar) and lower temperatures of 1205-1243 C (Figure 10) near the crust-mantle boundary.
We interpret that the magma was injected into a reservoir at the Moho that contained primitive olivine crystals.The sector-zoned clinopyroxene crystals crystallised because of low degrees of undercooling (e.g.Ubide, Caulfield, et al., 2019;Ubide, Mollo, et al., 2019;Wass, 1973).During ascent of the Anakie Hill magmas, olivine xenocrysts were entrained.This magma erupted at 18.1 ± 0.6 Ma.

Summary
The diverse crystal populations found in the Hoy lavas suggest a complex history of magma recharge, magma mixing, fractionation and xenocryst entrainment at mantle depths.These processes were more prevalent in the intermediate age (Mount Llandillo and Black Mountain) and young (Mount Leura and Anakie Hills) lavas, as they appear to have sampled deep crystal-bearing reservoirs (dominantly between $29 and 60 km) in the SCLM.The older Policeman's Knob lavas may have ascended to the surface more rapidly.However, a recent olivine diffusion study on the Auckland Volcanic Field concluded that even volcanoes with simple crystal cargoes, which imply a direct source-tosurface connection, may have experienced some degree of stagnation (Brenna et al., 2018).

Lava fields vs central volcanoes in eastern Australia
The lava fields of eastern Australia are diverse in their areal distribution and span a large age range, but as a group, they do not generally have an age-progressive pattern (e.g.Johnson, 1989;O'Reilly & Zhang, 1995;Zhang et al., 2001;Zhang & O'Reilly, 1997).Some are, however, spatially and temporally associated with the age-progressive shield volcanoes, known as central volcanoes (Figure 1).Hoy is spatially associated with the Buckland central volcano, and it hosts eruptions that pre-dated, overlapped with, and postdated the Buckland central volcano (30-27 Ma;Crossingham et al., 2018a).This presents us with a unique opportunity to compare magmas from these two distinct types of volcanic systems and compare their sources, magma plumbing systems and eruption styles.Jones et al. (2020) showed that central volcanoes and lava fields have similar bulk rock compositions, suggesting little geochemical differences in their sources, regardless of age or spatial association.Both Hoy and Buckland have similar bulk rock compositions (Figures 3-5).In particular, the bulk rock composition of the Hoy lava field is strikingly similar to the older Buckland lavas (30.3 ± 0.1 Ma).They both suggest low degrees of partial melting of an asthenospheric source (Figure 8; Crossingham et al., 2018a) that has interacted with the metasomatised SCLM, acquiring an EMI-like signature (Figures 4 and 5;Crossingham et al., 2018a;Shea & Foley, 2019).
The main distinction between the Hoy lava field and the Buckland central volcano is the depth and complexity of their plumbing systems (Figure 11).Mount Llandillo and Black Mountain, which temporally overlap with older Buckland, experienced magma mixing and magma rejuvenation at greater depths (49-30 km) than Buckland; their lavas also entrained both olivine and clinopyroxene mantle xenoliths before fast ascent to the surface (Figure 11).The older Buckland lavas (30.3 ± 0.1 Ma), in contrast, suggest no magma mixing and rejuvenation events.They lack mantle xenocrysts and have bulk compositions that suggest fractionation and crustal assimilation (AFC) at shallower 34-11 km depths, as recorded in clinopyroxene crystals (Figure 11: Crossingham et al., 2018a).
Unlike the Buckland volcano, where the plumbing system is relatively simple, central volcanoes from the southern segment of the eastern Australian track (e.g.Belmore, Warrumbungles, Comboyne, Nandewar, Ebor, Canobolas) have highly complex plumbing systems that sample multiple mushes, have experienced recharge and magma mixing, and may have been contaminated by the crust (Crossingham et al., 2018a(Crossingham et al., , 2018b;;Ewart et al., 1988;Skae, 1998;Sutherland et al., 2005Sutherland et al., , 2020;;Tapu et al., 2022).Recent work by Tapu et al. (2023) suggests that volcanoes in the north, experienced higher magma flux, and less magma stagnation than southern volcanoes.
What is similar across all the central volcanoes, yet distinct from the Hoy lava field, is the relatively shallow (crustal) depths at which central volcano magmas differentiated (e.g.Crossingham et al., 2018aCrossingham et al., , 2018b;;et al., 2022, 2023).The lavas from Hoy have not experienced crustal contamination, host abundant, mantle-derived olivine and/or clinopyroxene xenocrysts, and stagnated at depths below or at the crust-mantle boundary.Further work on the crystal cargoes of other lava field provinces is needed to determine if our observations from the Hoy lava fields can be extrapolated elsewhere.This is particularly important, as the porphyritic lavas of the Southern Highlands lava field (54-30 Ma; O'Reilly, 1989) are interpreted to have erupted from shallow mush reservoirs (Wass, 1973), and other lava fields supposedly experienced crustal contamination (Mason, 1989;O'Reilly & Zhang, 1995;Sutherland & Fanning, 2001), interpretations that differ from Hoy.
Interestingly, the depth at which magma stagnates in the crust has been linked to magma production rates (e.g.Gleeson et al., 2021;Smith & N emeth, 2017).In ocean island basalt intraplate settings, where magma production has been shown to be high (e.g.Hawaii), magmas mostly experienced stagnation and modification at shallow depth (Smith & N emeth, 2017, and references therein).In contrast, volcanic provinces with lower magma production rates, such as the Azores, experienced stagnation at depths at or below the Moho (e.g.Smith & N emeth, 2017 and references therein), with volatile saturation at subcrustal depths potentially triggering eruption (Ubide et al., 2022).The depth of magma stagnation appears to be the main difference between Hoy and the larger, shorter-lived central volcanoes.Estimates of eruptive volumes are not available for Hoy, as estimating magma production rates for the scattered small volcanoes is challenging owing to variable degrees of erosion.On average, however, estimated volumes of individual lava fields are substantially lower ($186 km 3 ) than those of central volcanoes (678 km 3 ; Duncan & McDougall, 1989, and references therein).
Compared with Buckland, the Hoy lava field is volumetrically smaller and has a more intermittent but longer lifespan.This suggests that the mechanism/s for magma production at Hoy potentially led to episodic magma generation separated by periods of quiescence.The larger and shorter-lived (3-5 Myr; Cohen, 2007) central volcanoes are believed to be derived from a mantle plume (e.g.Fishwick et al., 2008;Knesel et al., 2008;Tapu et al., 2023;Vasconcelos et al., 2008;Wellman & McDougall, 1974;Zhang & O'Reilly, 1997) or plumes (Sun et al., 1989;Sutherland, 2003), where the near-continuous melt production by plume activity would result in the growth of large shield volcanoes over a relatively short period of time.The  11. Schematic showing differences between the plumbing system of the Hoy lava field and Buckland central volcano at ca 30 Ma (not to scale).At Hoy, rejuvenating magma stagnated and mixed in deep, upper mantle reservoirs between the garnet-spinel transition zone ($55 km) and the Moho ($30 km) and, subsequently, ascended to the surface rapidly.At Buckland, there is no evidence to suggest that magma mixing and magma rejuvenation played a major role before eruption (Crossingham et al., 2018a).In contrast, crustal assimilation and fractional crystallisation (AFC) are common at Buckland, and clinopyroxene barometry indicated crystallisation between 34 and 11 km (Crossingham et al., 2018a), much shallower than Hoy.The Hoy magmas contain a complex assortment of crystals, including an abundance of mantle xenocrysts, antecrysts and phenocryts.In contrast, crystal populations at Buckland are relatively simple, containing only rare phenocrysts and even rarer olivine mantle xenocryts (Crossingham et al., 2018a) and suggesting relatively direct transfer of magma-feeding volcanism in the central volcano, similar to other age-progressive northern volcanoes in eastern Australia (Tapu et al., 2023).
mechanisms responsible for the generation of lava fields still elude us.
The distinction documented here between the Hoy lava field and the Buckland central volcano suggests that estimating the depth of magma storage may help to determine mechanisms controlling lava generation and eruption.Further work on temporally associated lava fields and shield volcanoes may bring new insight into the relevance of magma production rates and plumbing system architecture in determining the style of volcanism.

Conclusions
Hoy lavas in this study reflect a protracted 50 Ma history of punctuated volcanism.We investigated four periods of eruption: 67.5 ± 0.3 Ma, 32.3 ± 0.6-31.6 ± 0.7 Ma, 21.9 ± 0.5 Ma and 18.1 ± 0.3 Ma.A combined geochronological, petrological, geochemical and mineral chemistry approach to the Hoy lava field province permitted a better constraint on magma sources and tracing eruption processes at individual edifices.It also made it possible to compare the Hoy lava field with the temporally and spatially associated Buckland central volcano.
The Hoy lavas experienced low degrees of partial melting of an enriched EMI-like source with a minor contribution from a depleted Indian MORB source.The EMI signature was derived from interaction with the metasomatised SCLM.Mantle xenocrysts are ubiquitous in Hoy lavas, suggesting fast magma ascent to the surface.Excluding Policeman's Knob, all investigated samples record pre-eruptive residence within deep reservoirs ($30-47 km) in the SCLM, where they experienced magmatic differentiation processes including magma recharge, mixing and fractionation.
The magmas of the Hoy lava field and the neighbouring, larger, but shorter-lived, Buckland central volcano reflect similar degrees of partial melting and interaction with the SCLM.The most obvious difference between the lava fields and central volcanoes is the depth of magma stagnation and crustal contamination.This key difference may result from distinct mechanisms of magma ascent and different intensities of magma fluxes.

Figure 1 .
Figure 1.(a) Map of eastern Australia showing the location of the three types of volcanic provinces: lava fields (green), central volcanoes (black) and leucitites (orange and circled).The location of the Hoy lava field is highlighted by a red box, and the location of the Buckland central volcano is highlighted by a yellow box.After Johnson (1989).(b) Simplified geological map of the Hoy lava field, showing sample locations.After Geoscience Australia (2016).

Figure 2 .
Figure 2. Left: 40 Ar/ 39 Ar step-heating spectra for duplicate grains (seen in black and red) from each sample.Right: combined isochron and/or ideogram (Policeman's Knob) plots.

Figure 3 .
Figure 3.Total alkali vs silica (TAS) diagram (anhydrous) showing the major element chemical distribution of the Hoy lavas (whole rock: WR) and their modelled crystal-free liquid (M) (white filled symbols; Supplemental data, Appendix 7).Fields are after Le Maitre et al. (1989), and the alkaline/sub-alkaline line (grey) is after Irvine and Baragar (1971).The grey field is the major element chemical distribution of the Buckland central volcano (data from Crossingham et al., 2018a and references therein).

Figure 4 .
Figure 4. (a) Chondrite-normalised rare earth element (REE) patterns; and (b) primitive mantle-normalised multi-element patterns of bulk rock samples from Hoy.The chondrite and primitive mantle compositions and the average upper and lower crustal values are from Sun and McDonough (1989) and Rudnick and Fountain (1995), respectively.EMI values are the average for OIB Tristan and Inaccessible islands (GEOROC: http://georoc.mpch-mainz.gwdg.de/georoc/),and HIMU values are from Hofmann (1997) and references therein.Data from Buckland central volcano (grey field) from Crossingham et al. (2018a) and references therein.

Figure 5 .
Figure 5. Radiogenic isotope ratios of the Hoy lavas and mantle end-member fields.Isotopic errors are within the size of the symbol.Mantle end-member fields, including EMI, EMII, HIMU and MORB, are compiled from the GEOROC database: http://georoc.mpch-mainz.gwdg.de/georoc/.I-MORB and P-MORB after O'Reilly and Zhang (1995) and references therein.The unfilled circles with coloured rims represent other Hoy samples from Ewart et al. (1988).The grey fields representing Buckland central volcano are from Crossingham et al. (2018a), and show partial overlap with Hoy samples but also a trend towards increased crustal contamination.

Figure 7 .
Figure 7. Photomicrographs (in plane-polarised light) and backscatter electron images overlayed with semi-transparent laser ablation ICP-MS element maps for Cr of the different clinopyroxene crystal populations from the Hoy lava field: Mount Llandillo, Black Mountain and Mount Leura.Beneath these images are core-to-rim transects for Mg#, Cr and Zr.Legend is as in Figure 6.

Figure 10 .
Figure10.Clinopyroxene-melt thermobarometry from the Hoy lava field followingPutirka et al. (2003).We only considered pairs that passed the equilibrium tests of Fe-Mg exchange between clinopyroxene and liquid (K d ), as well as observed DiHd component of the clinopyroxene composition matching that predicted from the composition of the liquid(Putirka, 2008).Depths were calculated using a crustal density of 2.8 g/cm 3 and a mantle density of 3.3 g/cm 3. The depth of the base of the crust (Moho) and garnet-spinel transition are fromGriffin et al. (1987).

Figure
Figure11.Schematic showing differences between the plumbing system of the Hoy lava field and Buckland central volcano at ca 30 Ma (not to scale).At Hoy, rejuvenating magma stagnated and mixed in deep, upper mantle reservoirs between the garnet-spinel transition zone ($55 km) and the Moho ($30 km) and, subsequently, ascended to the surface rapidly.At Buckland, there is no evidence to suggest that magma mixing and magma rejuvenation played a major role before eruption(Crossingham et al., 2018a).In contrast, crustal assimilation and fractional crystallisation (AFC) are common at Buckland, and clinopyroxene barometry indicated crystallisation between 34 and 11 km(Crossingham et al., 2018a), much shallower than Hoy.The Hoy magmas contain a complex assortment of crystals, including an abundance of mantle xenocrysts, antecrysts and phenocryts.In contrast, crystal populations at Buckland are relatively simple, containing only rare phenocrysts and even rarer olivine mantle xenocryts(Crossingham et al., 2018a) and suggesting relatively direct transfer of magma-feeding volcanism in the central volcano, similar to other age-progressive northern volcanoes in eastern Australia(Tapu et al., 2023).

Table 1 .
Results of 40