Eruptive history of Mason Spur, a Miocene—Pleistocene polygenetic volcanic complex in southern Victoria Land, West Antarctic Rift System, Antarctica

Mason Spur is a deeply eroded Middle Miocene to Pleistocene (c. 13 to 0.37 Ma) volcanic complex in southern Victoria Land, within the West Antarctic Rift System (WARS). The oldest rocks include a large volume of trachyte ignimbrites that provided abundant volcanic detritus recovered in McMurdo Sound drill cores. The ignimbrites together with early-formed intrusions were strongly deformed during a substantial caldera collapse at c. 13 Ma. Intense erosion modified the volcanic landscape, creating a paleo-relief of several hundred metres. Deep ravines were cut and filled by deposits of multiple lahars probably linked to gravitational collapses of trachyte dome(s). Small-volume trachytic magmas were also erupted, forming lavas and at least one tuff cone. The youngest trachytic activity comprises a lava dome and related block-and-ash-flow deposits, erupted at 6 Ma. Basanite erupted throughout the history of the complex and eruptions younger than 12 Ma are almost exclusively basanite, forming scoria cones, water-cooled lavas, and tuff cones. Three peripheral outcrops are composed of basanitic ‘a‘ā lava-fed deltas, probably erupted from vents on neighbouring volcanoes at Mount Discovery and Mount Morning. Abundant ignimbrite deposits at Mason Spur differentiate this volcanic complex from others in the WARS. Eruptions were triggered by rift extension initially, yielding the voluminous trachytes sourced from a magma chamber on the margin of the WARS. Later mafic eruptions were associated with deep crustal faults related to residual intraplate deformation. These results add important details to the eruptive history of the intracontinental WARS.


Introduction
Antarctica hosts one of Earth's largest intracontinental rift provinces, the West Antarctic Rift System (WARS), which contains numerous large polygenetic volcanoes (Wilch et al. 2021; Fig. 1). They have a range of ages spanning the past 37 Myr but most are Middle Miocene and younger (≤ 14 Ma) and the volcanism is still active at prominent centres such as Mount Erebus, Mount Berlin, and Mount Melbourne (Geyer 2021). Antarctica is also covered by Earth's largest ice sheet and most of the volcanoes inland are largely obscured by ice. The Antarctic ice sheet has had a predominantly cold-based thermal regime since c. 14 Ma (Lewis et al. 2007) and, with rare exception (e.g., Crary Mountains and Mount Sidley, Marie Byrd Land; Panter et al. 1994;Wilch and McIntosh 2002), volcanoes that are exposed above the present-day ice have undergone only minimal erosion (Rocchi et al. 2006). Thus, their volcanic evolution is very poorly known. By Editorial responsibility: M.H. Ort; Deputy Executive Editor: L. Pioli contrast, volcanoes within the WARS that are situated in coastal areas have frequently undergone extensive erosion, mainly marine, during interglacial periods (Smellie 2021). This has resulted in several volcanoes being exposed deep into their interiors, for example in northern Victoria Land and eastern Marie Byrd Land (LeMasurier et al. 1994;Wilch and McIntosh 2002;Smellie et al. 2011a).
This paper describes the geology and volcanic evolution of Mason Spur, a prominent south-east-facing volcanic massif 1300 m high and c. 14 km in length that overlooks a deep embayment in the Ross Ice Shelf, southern Victoria Land (Fig. 2). It is situated below the south-east flank of the large mainly Plio-Pleistocene Mount Morning volcano (< 5 Ma; though ages as old as 25 Ma are inferred; Fig. 1). Mason Spur has been extensively and deeply eroded because of its coastal location and age, resulting in > 1000 m of relief and significant exposure of the deep stratigraphy. The locality has been visited rarely and the previous studies focussed on the age, geochemistry and petrological evolution of the basanite-trachyte magmatic lineages present (Wright-Grassham 1987;Martin et al. , 2013Martin et al. , 2021a. However, those studies described an unusually wide variety of volcanic products, including lava flows, a lava dome, scoria cones, hyaloclastite, pillow lava, volcanic breccia, tuff cones, tuff rings, lahar and fluvial deposits, and diamictite. In this paper, we review the volcanic evolution of the long-lived Mason Spur volcanic complex using stratigraphical studies and a new geological map (Fig. 3). Because of the excellent exposure, the outcrops provide a rare opportunity to document the geological history and evolution of one of the large WARS volcanic centres. Twenty-one new 40 Ar/ 39 Ar dates give an improved eruptive chronology and allow a Fig. 1 Map showing the location of Mason Spur within the Erebus Volcanic Province and the West Antarctic Rift System (WARS; inset, with selected large volcanoes shown as red dots). Possible bounding fractures of the Terror Rift are also shown (from Horgan et al. 2005), as are the locations of two drillsites, AND-1B and AND-2A. The field area described in this paper is outlined by a black polygon. TAM-transantarctic mountains (stippled ornament) comprehensive reassessment of the volcanic evolution to be made. An important result of our study is the discovery of a thick caldera-filling sequence of ignimbrites, reported initially by Martin et al. (2018). Similar deposits are very rare and much less voluminous elsewhere in the WARS and our study thus fills a significant gap in our knowledge of the eruptive and structural evolution of the large polygenetic WARS volcanoes.

Geological background
Mason Spur is situated in southern Victoria Land within the WARS, which measures 500-700 km in width, over 3000 km in length and covers an area of c. 1.2 × 10 6 km 2 (Fig. 1, inset;LeMasurier 2008;Jordan et al. 2020 Smellie and Martin 2021). Mason Spur represents one of the early stages of volcanism of the Mount Morning Volcanic Field. Previous geological studies of Mason Spur were a PhD project by Wright-Grassham (1987; a brief summary is given by Wright and Kyle 1990) and papers by Martin et al. ( , 2013Martin et al. ( , 2018. The study by Wright-Grassham (1987) was reconnaissance in nature and investigated Mount Discovery, Mount Morning and Minna Bluff in addition to Mason Spur. Martin et al. ( , 2013 examined the petrology of Mount Morning and undertook isotopic dating but they included a brief study of Mason Spur. Wright-Grassham (1987) informally defined the "Mason volcanic complex," a name that we also adopt in this paper, and identified two petrological suites: an older suite ranging from mugearite to peralkaline trachyte and a younger suite of mainly basanites/tephrites and benmoreites (also Martin et al. , 2013Martin et al. , 2021a. Helms Bluff, situated 5 km north of Mason Spur (Fig. 3), is also included in our study for completeness since it was previously regarded as part of the Mason Spur sequence (Wright-Grassham 1987). However, on the basis of transport directions for the volcanic units present (lava-fed deltas; see later) and an isotopic age that is younger than volcanic units at Mason Spur apart from late-stage scoria cones, it is regarded as an outlier of the neighbouring Mount Discovery volcano (see Smellie and Martin 2021, and below).
The lithological descriptions of Mason Spur by Wright-Grassham (1987) include references to numerous sedimentary and volcaniclastic units. However, the study by Wright-Grassham (1987) is now 35 years old and, given the more focused nature of our study and numerous significant advances in volcanology that have taken place since the initial study, we provide a new stratigraphy and geological interpretation.

Methods and terminology
For this study, fieldwork by JS, AM, and DT at Mason Spur occurred over 21 days during January 2017. It built on 10 days of fieldwork by AM in 2005. Outlying areas near Dunbar Head, and at Helms Bluff and Snow Petrel Peak were accessed during short visits by helicopter. A new and significantly revised lithostratigraphical geological map is presented in Fig. 3. Rock samples were taken for petrography and isotopic dating. More than seventy thin sections were examined and 21 geochronology  (Dunbar Head;see Fig. 3 (inset) for location). The base of the crag is mostly above c. 400 m above sea level and the summit is at c. 1300 m. It has been exposed by a combination of glacial and marine erosion samples analyzed. The isotopic dating was carried out at the Istituto di Geoscienze e Georisorse, Consiglio Nazionale delle Ricerche (Pisa, Italy) and at the New Mexico Geochronology Research Laboratory (USA). A full description of the dating results and methods used is presented in Supplementary Information S1. A summary of the petrology of Mason Spur is presented by Martin et al. (2021a, b) and representative major oxide compositions are provided in Supplementary Information S2. Elevations and unit thicknesses were determined by hand-held GPS (Garmin eTrex), adjusted for daily pressure drift and calibrated to fixed locations in the field, with an estimated error of ± 10 m.
"Ignimbrite" is used to denote any pumiceous ash-rich deposit laid down by the wide spectrum of pyroclastic density currents, ranging from fully dilute to concentrated (granular fluid-based), following the recommendations of Branney and Kokelaar (2002;also Giordano and Cas 2021). The terminology for other volcaniclastic rocks follows that of White and Houghton (2006), and rock compositional names are after Le Maitre et al. (2002). Unofficial place names used are indicated by inverted commas at first use in the text and on maps.

Stratigraphy of the Mason Spur volcanic complex
A schematic cross section of Mason Spur illustrating the lithostratigraphy proposed in this study and the vertical and lateral relationships is shown in Fig. 4. Details of the stratigraphical units are given in Tables 1 and 2 and representative images of the lithofacies in Figs. 5, 7, and 8. Full descriptions of the lithofacies are presented in Supplementary Information S3, and additional supporting images are presented in Supplementary Information S4 and S5 to give a comprehensive view of the geology.
Unlike most other volcanoes within the WARS (Smellie 2021;Wilch et al. 2021), the stratigraphy at Mason Spur is dominated by clastic units, varying from ignimbrites, scoria, and agglutinate to sandstones and breccio-conglomerates, but lavas and intrusions are also locally prominent. The bulk of the outcrop is dominated by ignimbrite and breccia deposits and hypabyssal intrusions of MSP1. It has a minimum thickness of 800 m. Subsequent units are volumetrically Fig. 4 Schematic west south-west-east north-east cross section of Mason Spur (not to scale) showing the relationships between the different lithostratigraphical units identified in this study. Some of the units crop out on the modern slope or above and behind the section but they are included in the sketch where indicated. Note how unit MSP1 has been severely eroded prior to the deposition of younger units, with > 800 m of relief present. The eroded upper surface of MSP1 has a topographical relief of several hundred meters minor in comparison and they unconformably drape MSP1 and each other. The most prominent erosional surface is at the base of unit MSP3. It defines a ravine cut at least 150 m into MSP1 (Fig. 7). Unit MSP3 is also distinctive in being largely formed of bedded sandstones and breccio-conglomerates. Sedimentary rocks are largely absent elsewhere apart from within unit MSP6 and at the base of unit MSP5, although they are just a few meters thick in the latter. Compositions at Mason Spur are essentially bimodal (basanites/ tephrites to trachytes; Martin et al. , 2013Martin et al. , 2021aSupplementary Information S2). However, trachyte magmas dominate basally whilst mafic magmas were erupted mainly after unit MSP4 and form the higher part of the succession.

Chronology
Results of previous studies Wright-Grassham (1987) determined K-Ar ages for two samples from Mason Spur equivalent to our unit MSP1 (Table 3). Another three samples from the same unit were dated by 40 Ar/ 39 Ar to try and counteract possible effects of the pervasive alteration affecting MSP1 and yielded ages consistent with the K-Ar ages.  also published one K-Ar and two 40 Ar/ 39 Ar ages, for samples obtained within our MSP1. All of the previously determined ages for MSP1 are for the less altered intrusive units. They have a wide range of 12.80-11.4 Ma, with high 2σ errors (typically 0.2-0.4 Ma). A benmoreite sample, also dated by  and from a location within our MSP1, yielded an age of 12.1 ± 0.2 Ma. Only two of the younger units were dated, yielding K-Ar ages of 11.7 ± 0.4 Ma for MSP7 and 6.13 ± 0.20 Ma for MSP8. Additionally, a sample of lava from Helms Bluff was dated by Wright-Grassham (1987) as 4.51 ± 0.31 Ma (by K-Ar).

Results of new 40 Ar/ 39 Ar dating
Twenty-one samples were dated by 40 Ar/ 39 Ar (Table 3; Supplementary information S1). Eight of the eleven eruptive units identified at Mason Spur are now dated, and the undated units (MSP2, MSP3, and MSP4) are stratigraphically above or below dated units so their ages are approximately constrained.

Units dated directly
Twelve samples from unit MSP1 were dated. The 2σ errors on the new ages are smaller than those already published, and range from 0.01 to 0.07 Ma (Supplementary information S1). All the new ages for MSP1 were determined on intrusions, which provide minimum limiting ages for MSP1.  Vesicles in the flattened bombs are usually spherical signifying volatile expansion after deposition; the distinctive bomb-rich variant is probably that described by Wright-Grassham (1987), p. 39) as 'massive bomb lapillistones and agglutinate' in her unit MS2 Massive fine lapilli tuff with flattened obsidian bombs Distinctive variation of the fine lapilli tuff with flattened bombs lithofacies; at least 80 m thick; comprises massive monomict trachytic lapilli tuff formed of finely vesicular angular pumice lapilli 2 mm to 1 cm in diameter, varying up to 5 cm, associated with 10-20% blocky obsidian lapilli; variably clast-supported to matrix-supported with < 10-20% fine ash matrix; distinguished by the presence of 10-15% conspicuous elongate, finely fractured obsidian bombs, many with wavy fluidal shapes; other obsidians angular, partly broken or jigsaw fractured; the largest obsidian bomb seen is c. 2 m long but most are c. 20-80 cm in length Ignimbrite deposit with obsidian spatter incorporated when pyroclastic density current overran co-erupting spatter fountain in separate vent; see Martin et al. (2018) for more detailed explanation Only observed at the east end of Windscoop Bluff; probably the deposit described by Wright-Grassham (1987, p. 42) as 'massive, matrix-rich, obsidian-bearing lapilli tuff containing lobes of trachytic lava, with spalling obsidian rinds' within her unit MS5 and which she interpreted as a trachytic hyaloclastite associated with local dyke intrusion into ice or water; reinterpreted by Martin et al.  Wright-Grassham, 1987, p. 40) in her unit MS2 and attributed by her to deposition from base surges in a tuff cone

Massive fine breccia and tuff breccia
Variably matrix-to (mainly) clast-supported monomict breccia with scarce large flattened bombs; mostly composed of lapilli-size clasts but locally composed of numerous (to 15%) angular blocks of foliated (pilotaxitic) trachyte up to 1 m in diameter; matrix is fine to coarse ash, 5-10%; lapilli are creamy white, non-vesicular and angular but some show corner abrasion; generally scarce bombs are up to 60 cm in length and have very variable concentrations up to 40%; they have deformed fluidal shapes with flattened vesicles and broken terminations Massive monomict nature and textural and compositional similarity to associated intrusions suggests the breccias are broken & disrupted intrusions that were emplaced prior to caldera collapse; ash-matrixed variants may have undergone transport as mass flows; deposits with pumices and pumiceous bombs indicate dynamic mixing with unconsolidated ignimbrite deposits

Massive coarse breccia
Monomict coarse breccia with angular platy clasts 2-22 cm in diameter composed of multi-coloured (pale grey, cream, rusty) weakly foliated trachyte lava; also dispersed blocks of the same up to 70% set in c. 15% to rarely 30% granule-grade matrix; locally rich in fine ash (generally c. 25; to 70%) with floating blocks; bombs absent; also occurs rarely as lenses with erosive bases within fine lapilli tuff lithofacies at Windscoop Bluff Mainly formed by in situ disintegration of coeval pre-caldera intrusions, disaggregated during caldera collapse; often grade laterally into jigsaw-fractured and unbroken intrusions; less common ash-matrixed breccias probably transported as mass flows but no orginal bedding observed Most widely developed at Anniversary Bluff; essentially a coarser version of massive fine breccia and tuff breccia lithofacies but lacking admixed pumiceous ignimbrites No material from pumice within the ignimbrites was suitable given their highly altered condition. The range of ages is 12.62-11.19 Ma, similar to the age range reported by Wright-Grassham (1987) and . The new ages suggest that most of the intrusions (c. 60%) were emplaced between 12.62 and 12.13 Ma. The intrusive period was followed by a short lull before renewed activity between 11.97 and 11.19 Ma. The intrusions in MSP1 are mainly trachytic but a few intrusions are less evolved (e.g., alkali basalt, hawaiite) and less altered. All the new ages are from samples of trachyte intrusions except for a single alkali basalt intrusion that yielded an age of 11.33 ± 0.03 Ma. Unit MSP5 (water-cooled basanite lava), which unconformably overlies MSP1, is 12.1 ± 0.1 Ma, and 11.4 ± 0.1 Ma was obtained for a sample of lava in MSP6. Samples from two of the three outcrops of MSP7 (basanitic pyroclastic cones) were dated and yielded 40 Ar/ 39 Ar ages of 10.5 ± 0.03 and 10.02 ± 0.05 Ma. Unit MSP8 (trachyte dome) has an age of 6.03 ± 0.01 Ma.
Only a single age was obtained for outcrops of unit MSP9 (basanitic spatter and water-cooled lavas). It was determined on a sample of clastogenic lava within scoria and agglutinate and yielded an age of 0.94 ± 0.03 Ma. Unit MSP10 has yielded a single age of 0.366 ± 0.017 Ma, consistent with the relatively fresh appearance of some of the scoria cones that form MSP10, in which craters are preserved in at least two outcrops.
Finally, the middle lava-fed delta unit at Helms Bluff is 4.92 ± 0.04 Ma. Similar lithofacies at Helms Deep gave an age of 0.043 ± 0.0052 Ma.

Outcrops with bracketed ages
From the field relations and ages of dated units, units MSP2, MSP3, and MSP4 are younger than unit MSP1 and older than unit MSP5 (12.1 ± 0.1 Ma). MSP2 is older than MSP4 but otherwise the relative order of superimposition of the three undated units is uncertain. It is notable that the wide range of ages for MSP1 (12.62-11.19 Ma; all for intrusions) overlaps with that inferred for the three undated units and also units MSP5 and MSP6 (12.1 ± 0.1 and 11.4 ± 0.1 Ma, respectively), all of which unconformably overlie MSP1, a problem that is discussed later.

Unit MSP1
MSP1, the > 800 m-thick basal unit at Mason Spur, is dominated by ignimbrite and breccia deposits and numerous hypabyssal intrusions (MSP1). The ignimbrites are strongly lithified due to the pervasive clay alteration of the volcanic glass (Tables 1, 2; Supplementary Information S3). They are fines-rich, pumiceous, and are overwhelmingly massive but they include isolated relicts of diffuse stratification. In overall characteristics, they resemble the deposits of both low-and high-concentration pyroclastic density currents (Branney and Kokelaar 2002). The stratification has varied attitudes (generally steeply-dipping [> 60°]; Fig. 5b) and there are rare thixotropic textures, which suggest that the ignimbrites were more pervasively stratified originally but lost their stratification while poorly lithified, possibly through one or more deformation events (described below). Obsidian is also locally prominent, particularly in a spectacular deposit at Windscoop Bluff. That deposit has been interpreted as a spatter-rich pyroclastic density current deposit, which formed in unusual circumstances, i.e., co-eruption of ignimbrite and trachytic spatter in separate vents and mingling of the different products during transport (Martin et al. 2018).
Fine and coarse breccias are often prevalent in MSP1. They are formed of monomict, dense, fine, and aphanitic trachyte lava clasts that are texturally and compositionally identical to the massive hypabyssal intrusions (Martin et al. 2018). There is also a textural gradation from jigsaw-fractured trachyte to orthobreccia or rare parabreccia, sometimes within the same intrusion although, in general, large outcrop areas are simply formed of massive breccia. Although mingling of pumice with trachyte clasts derived from associated breccias occurs, scoriaceous trachyte lava clinkers and oxidised clinkers, characteristic of subaerial lavas, are absent. Moreover, traced up the outcrop, the individual breccia outcrops are tens of metres thick, implying that they would always have to be derived from lavas of a similar thickness, which would be unusual. Despite the brecciation, the clasts in the breccias are crystalline and do not have glassy margins, and there is no clear evidence for water chilling. The fragmentation is therefore thought to be due to deformation rather than rapid cooling. We suggest that the breccias represent progressively broken and disrupted hypabyssal intrusions that were affected by one or more volcano-tectonic (caldera-forming) deformation events, presumably the same one(s) that affected the ignimbrites. As such, they can be called collapse breccia deposits.
More clearly intrusive lithofacies are also abundant within unit MSP1. They are of two types based on morphology: (1) those with fluidal shapes; and (2) planar sheet-like intrusions. Category (1) intrusions are distinctive, with their irregular fluidal shapes signifying intrusion into an unconsolidated host. The category (1) intrusions and the massive brecciated intrusions just discussed are also extensively intruded by clastic dykes. Category (2) intrusions form tabular sheets with conspicuous fine-grained chilled margins.
They appear to be unaffected by tectonism and they are not intruded by clastic dykes, but their trachytic compositions indicate a compositional affinity with the earlier trachytic magmatic activity.

Units MSP2 to MSP7
Units MSP2 to MSP7 all represent short-lived small-volume eruptions. Unit MSP2 comprises at least two subaerial trachyte lavas that lack evidence for interaction with water and were erupted subaerially. They crop out at significantly different elevations and locations and hence probably represent two separate emplacement events. MSP3 is one of the most highly distinctive units at Mason Spur. It is dominated by epiclastic sedimentary rocks, comprising coarsegrained mass flow deposits and sandstones. An outlier of MSP3 also contains a massive, thick (> 60 m), polymict mass flow deposit with some clasts > 20 m in diameter and a > 50 m-thick monomict bed composed of fines-poor pumice and blocky lava clasts that is probably a block-and-ashflow deposit related to a coeval trachyte dome that is either not exposed or no longer preserved. Most of the deposits within MSP3 have the characteristics of lahar deposits, particularly the distinctive mixture of angular and abraded clasts (hence their description as breccio-conglomerates). The mixture was probably caused by bulking up of juvenile debris (angular) with alluvial detritus (abraded) scoured from the surface over which they passed (Pierson et al. 1990;Thouret 1990;Pierson and Waitt 1999a), although shear and associated clast-clast and clast-substrate collisions in the moving flow probably also contributed to abrasion (Scott 1988). The abundance of monomict, fine, and aphanitic (juvenile?) trachyte clasts in the coarse beds also suggests a plausible link with a lava dome undergoing gravitational collapses, that transformed downslope to lahars (cf. Trabant et al. 1994;Pierson and Waitt 1999a, b). The distinctive sandy beds with faint planar lamination and spectacular dewatering structures are interpreted as products of hyperconcentrated flows also linked to lahars (i.e., lahar runout deposits ;Scott 1988;Pierson and Scott 1985;Scott et al. 1995;Procter et al. 2021). Better-sorted interbeds with prominent channels and trough cross bedding represent more normal streamflow conditions recurring between lahar events and involving migrating irregular bars similar to highgradient, gravel-dominated braided stream or river deposits (e.g., Miall 1977).
Despite the absence of any primary landform, the features of MSP4 (e.g., abundant variably vesicular blocky-angular sideromelane, fine ash matrix, lack of bedrock clasts) suggest that the deposit is hydrovolcanic in origin and is probably a relict tuff cone (e.g. Sohn 1996;Sohn et al. 2012;Brand and Brož 2015). The combination of bedding orientations and rare directional bomb sags in the ridge summit outcrops, together with a proximal spatter-rich bed and associated bedding dipping steeply (32-50°) to the southwest at the main cliff edge, implies that the vent responsible for MSP4 was probably situated somewhere beyond the cliff edge and is now eroded. MSP4 is overlain by unit MSP5, which is dominated by a thick (> 30 m) basanite lava showing prominent basal blocky and pseudopillow joints characteristic of water-cooling. The lava overlies c. 2 m of stratified pebbly sandstone and sandstone into which parts of the lava have sunk and interacted, deforming the stratification. The sediments are well sorted, nearly monomict and probably fluvial. They also contain numerous conspicuous white pumice clasts probably derived from underlying unit MSP4. Massive breccio-conglomerate > 4 m thick underlies the stratified section and may be a mass flow deposit. Unit MSP6, representing an effusive episode of substantial basanite lavas at c. 11.4 Ma, overlies MSP3 and is also associated with mass flow deposits broadly similar to the coarse lahar deposits of unit MSP3 but finer, sandier, and polymict. The two units are separated by an unconformity.
Unit MSP7 comprises deeply eroded pyroclastic cone remnants dated at 10.5 and 10.02 Ma. Reasoning similar to that used for MSP4 suggests that they were also tuff cones. Lapilli tuffs in the MSP7 outcrop above "The Island" are generally massive or show faint deformed stratification. Some of the disturbance was possibly caused by intrusion of a coeval hypabyssal neck present nearby, but the predominantly massive nature of the deposit is unusual for tuff cones, which are typically well stratified. The lapilli tuffs may thus represent a within-vent deposit homogenised by multiple explosive episodes and passage of upwardmoving debris jets (White and Ross 2011;Smellie et al. 2018). However, a thin sequence of stratified tephra beds with bomb and impact sags is prominent in mid-elevation

Units MSP8, MSP9, and MSP10
Unit MSP8 is a small trachyte lava dome. It is associated with thick beds alternately dominated by pumice or fine-grained crystalline lava blocks interpreted as explosively-generated block-and-ash-flow deposits derived by collapses of "The dome." The oxidized surfaces shown by some of the pyroclastic beds and the surfaces of bombs and some breccia masses in the block-and-ash-flow deposits are consistent with emplacement at high temperatures.
The eruption of unit MSP9 began with effusion of basanitic lavas with distinctive joint patterns indicative of water cooling. However, other products of MSP9 and all of MSP10 consist of scoria cones and associated agglutinate and clastogenic lavas. MSP9 is dated as 0.94 Ma. Although Wright-Grassham (1987) obtained a K-Ar age of 11.7 ± 0.4 Ma apparently within the MSP9 outcrop to the south-east of The Dome (MSP8), our mapping suggests that it probably dates a small isolated outlier of MSP7 basanitic spatter, several of which crop out on this slope above the main MSP7 outcrop (Fig. 4). The locations of the vents responsible for unit MSP9 are unknown. Some were probably situated on the Mason Spur ridge crest, although others may have been fed by the scoria and spatter-filled dykes which cut MSP1 (see Fig. 7 and S4-20-22). It is also conceivable that some vents are represented by scoria cones assigned to MSP10. However, most of the MSP9 outcrops are > 1 km from known MSP10 vents, which is probably too far for accumulation of spatter and agglutinate erupted from those centres. Scoria cones in MSP10 have an east-northeast-west-southwest alignment and were probably erupted from fissures.

Un-numbered units
Additional basanite outcrops, mainly massive breccias, occur at Helms Bluff, Snow Petrel Peak (4.9 Ma) and Helms Deep (0.04 Ma). The thickness and lithology of the breccia deposits, together with their abundant irregular lava masses with blocky jointing characteristic of water cooling, strongly resemble the chaotic lava-lobe-bearing breccia lithofacies that characterises the basal subaqueously-deposited lithosome of glaciovolcanic 'a'ā lava-fed deltas (Fig. 8e, f; Smel lie et al. 2011a, b, 2013Smellie and Edwards 2016). The breccias at Helms Deep lack any capping lavas (presumably eroded) but pāhoehoe lavas capping poorly exposed breccia(?) deposits are seen in the cliff close to the Helms Deep outcrop and probably represent an underlying pāhoehoe lava-fed delta. The outcrops at Snow Petrel Peak and Helms Bluff pass up into capping 'a'ā lavas and are 'a'ā lava-fed deltas. Fig. 7 View of stratified unit MSP3 and its stratigraphical relationship with unit MSP1, looking east from the west end of Anniversary Bluff. MSP3 deposits fill a deep ravine cut in MSP1 and are overlain by basanite lavas and pale khaki-colored mass flow deposits of MSP6. The contact between MSP3 and MSP6 is erosional. White trachytic ignimbrites that cap unit MSP3 in the background are missing in the foreground. Note the prominent dip to the left (north-west) of stratification within unit MSP3 indicating that deformation has occurred. The true dip is less than that shown and is over-emphasized due to camera foreshortening (cf. Fig. 6a) ; the hills in the background are more-degraded examples of scoria cones assigned to MSP10. (E) Chaotic basanitic pillow lava masses and aphanitic breccia at the base of the middle sequence at Helms Bluff; the deposit resembles the subaqueous lithosome of an 'a'ā lava-fed delta; the orange-coloured deposit is a lapilli tuff probably reworked from some very large rock masses carried by the delta as it advanced (cf. Smellie and Rocchi 2021, figs. 8, 12); ice axe (ringed) gives scale. (F) Chaotic blue-grey and dark grey basanite lava masses and rust-coloured aphanitic breccia, interpreted as the subaqueous lithosome of an 'a'ā lava-fed delta; the hammer near centre of the photograph (ringed) is 25 cm in length; Helms Deep (Dunbar Head)

Discussion
This study combines the first modern in-depth assessment of the volcanic history of Mason Spur together with an improved chronology. The study is facilitated by the deep erosion of the outcrop, which has exposed a 1000-m-thick section of a volcanic complex. The current assessment is based on interpretations of the lithofacies and numerous new isotopic ages. It is currently the most detailed description of an ancient volcanic complex within the WARS. The sequence at Mason Spur is constructed from multiple volcanic and volcaniclastic eruptive units together with epiclastic sequences, which we have designated MSP1-MSP10. The activity is summarized as follows: the volcanism was mainly trachytic during Miocene time (between c. 13 and 12 Ma), represented by units MSP1 (the most voluminous  Smellie and Martin, 2021). At least four major eruptive stages are recognized and are described below (Fig. 9). This stage is represented by massive ignimbrites and lithic breccias. They are the oldest deposits exposed at Mason Spur and they represent the most voluminous eruptive episode of the volcanic complex. The abundant hypabyssal intrusions (see next section) and pervasive hydrothermal alteration in the ignimbrites and breccias might suggest that MSP1 is a subvolcanic (hypabyssal) complex. However, the association of abundant clastic dykes, deformation of (unconsolidated) ignimbrites and widespread massive brecciated intrusions, together with the large volume implied by the extensive thick outcrop and its pervasive alteration, indicate that unit MSP1 is probably a caldera fill (assumed by Martin et al. 2018). The deformation resulted from one or more caldera collapse events, and is also evidenced by the variable, often very steep dips (> 60°) of relict stratified Fig. 9 (A) Diagram showing the stratigraphical succession exposed at Mason Spur based on this study. The descriptive eruptive stages used in this paper are also shown (see text and Table 1  The apparent absence of extra-caldera ignimbrites near Mason Spur may be due to the location of the volcanic centre as an island within the Ross Sea Embayment. If deposition of the extra-caldera products took place on a Miocene Ross Ice Shelf and outlet glaciers of the Transantarctic Mountains, the deposits would have been transported by ice to the ocean and effectively lost to the geological record. Alternatively, in an ice-poor or ice-free period when an ice shelf was absent, the tephras would have been deposited mainly within the Ross Sea. The only potential source of information on coeval marine sedimentation in the area is from drill cores, AND-2A and AND-1B, obtained at sites c. 90 km north-east of Mason Spur (Fig. 1). MSP1 was probably erupted during a period represented by an erosional hiatus in AND-2A (Fig. 10). However, the hiatus was succeeded by the deposition of Geochemical Unit 4 (123-225 m below sea floor (mbsf); Panter et al. 2008;Di Vincenzo et al. 2010). Geochemical Unit 4 is largely composed of diamictite (c. 80% by volume, estimated) with subordinate conglomerate, sandstone, and siltstone. Linked petrographical and geochemical studies of the unit indicated that the clasts are dominated by volcanic rocks, and continuous XRF scanning of the core shows a simultaneous marked increase in Si; the proportion of fresh volcanic glass also shows a marked increase in Geochemical Unit 4 (Panter et al. 2008; see also Nyland et al. 2013). The clasts consist of a mixture of mafic, intermediate, and felsic compositions, including black scoria and brown (probably mafic) glass, light gray to golden brown pumice (including long-tube pumice) and intermediate and felsic clasts with pilotaxitic textures; one of the few clasts Fig. 10 Chronostratigraphical framework for strata recovered by AND-1B and AND-2A drill cores (blue bars represent relatively well dated strata; the gray bars indicate intervals for which ages are not well constrained; after Levy et al. 2021) compared with the age of volcanic episodes at Mason Spur identified in this paper (red bars). MSP1a-ignimbrites; MSP1b-intrusions. The positions of sedimentary deposits rich in volcanic detritus possibly derived from eruptions at Mason Spur (i.e., AND-2A Geochemical Unit 4 and AND-1B Lithostratigraphical Unit 6) are also shown; see text for details). The erosional hiatuses indicated represent regional seismic unconformities encountered in Ross Sea drill cores, interpreted as periods of multiple ice advances and retreats  ▸ analyzed was shown to have a peralkaline trachyte composition. The abundance of volcanic clasts increases markedly above 225 mbsf, with a prominent peak between 182 and 174 mbsf, and continues up to 30 mbsf. The available information suggests that the volcanic material in AND-2A, particularly its sudden influx, may have been strongly influenced by trachytic and subordinate mafic eruptions at Mason Spur associated with MSP1, including caldera collapse. Either the exposed edifice contributed eroded material for a long time or much of the debris was reworked up-section into younger sediments during successive glacial advances that were probably responsible for the depositional hiatus. Eruption of MSP1 ignimbrites also probably corresponds to an erosional unconformity in AND-1B  ; Fig. 10). The closest succeeding units are two sedimentary sequences designated LSU5 and LSU6 (McKay et al. 2009); LSU5 is significantly younger than LSU6 (c. 6.5 Ma; Wilson et al. 2012) and is not relevant to MSP1. The basal part of LSU6 (c. 1250-1067 mbsf) is dominated by diamictites (> 90% by volume, estimated) with minor sandstone, whereas the upper part (c. 1067-762 mbsf) comprises mainly mudstones and sandstones with lesser diamictites (c. 25% by volume). Volcanic debris is present throughout LSU6 although it is not unusually abundant (Pompilio et al. 2007). It comprises mafic, intermediate, and felsic compositions but intermediate-felsic compositions with trachytic textures are dominant. Glass is also present and its abundance increases markedly above c. 850 mbsf. It is mainly mafic but ranges to trachyte in composition. The lower volume of volcanic detritus and the abundance of mafic glass in AND-1B (LSU6) compared with AND-2A may be due to different sediment transport paths to the two sites, which are 50 km apart, or else different provenances. Since most of other large volcanic centres in the area (Mount Discovery, Brown Peninsula, White Island, Ross Island; Smellie and Martin 2021) are younger than c. 5 Ma and would not have been present, the palaeotopography was undoubtedly very different at the time. Overall, a link with eruptions at Mason Spur is less obvious in AND-1B than for AND-2A but is still permissible.
Stage 1b: post-caldera intrusive activity (MSP1b; c. 12.9-11.2 Ma) The abundant hypabyssal intrusions within unit MSP1 can be assigned to at least two intrusive episodes: (1) early postcollapse of the caldera; and (2) late post-collapse. By contrast, the massive breccias described for stage 1a are linked to the caldera collapse event, i.e., created by pervasive brecciation of pre-collapse intrusions. They are thus pre-caldera in age. The category (1) intrusions with fluidal margins intrude the ignimbrites and brecciated intrusions indicating that their emplacement post-dated the collapse, although a few show minor faults affecting their margins. Only one of our dated samples is from a category (1) intrusion. It has the oldest age in our new dataset (12.62 ± 0.01 Ma) and is almost within error of the oldest previously published age (12.9 ± 0.2 Ma; Table 3). The other dated samples were obtained on category (2) intrusions (tabular sheets with planar margins), which intrude ignimbrites, breccias and fluidal-margined intrusions. The lack of brecciation of category (1) and (2) intrusions signifies a post-collapse age. Thus, the age of the collapse event is bracketed by the age of the brecciated intrusions (pre-collapse; age unknown) and the fluidal intrusions (early post-collapse; c. 12.9 Ma). Here we take c. 13 Ma as the likeliest time for the caldera-forming event. However, the ages of category (2) intrusions extend up to 11.2 Ma (Table 3). This implies a lifetime for the basal trachyte volcano of at least 1.5 Myr and the possibility of associated trachytic ignimbrite eruptions, or small trachytic centres now removed by erosion. A lifetime of > 1.5 Myr is not unusual for a large polygenetic stratovolcano (Panter et al. 1994) but other explanations are possible (see later).

Stage 2: uplift and eruption of multiple small-volume trachyte and basanite monogenetic centers (MSP2-MSP7; c. 12.9-10 Ma)
Unit MSP1 underwent significant erosion prior to the eruption of the other stratigraphical units. It may imply a change in the local base level due to uplift of the volcano. There is no definitive evidence for uplift in the Mason Spur sequence but we note that unit MSP3, at least, has been back-tilted slightly (Fig. 7), which suggests that some deformation has occurred. Moreover, neptunian dykes filled by white ignimbrite have infilled fractures in the top few metres of the sedimentary rocks of unit MSP3 (Fig. S4-23), also consistent with deformation associated with vertical displacement(s). Mason Spur is situated on the eastern flank of the Transantarctic Mountains (Fig. 1). Different parts of the Transantarctic Mountains appear to have acted independently to phases of uplift, although there is little direct information for the period after c. 30 Ma (Fitzgerald 2002;Olivetti et al. 2018). Miller et al. (2010) suggested that uplift of c. 300-800 m may have affected parts of the Transantarctic Mountains since c. 14 Ma, based on thermochronology and geomorphology. Moreover, post-Middle Miocene Climate Transition (i.e., after 14.8-13.8 Ma) isostatic uplift of a few hundred metres might also have occurred in response to valley erosion by glaciers (Halberstadt et al. 2021;Sauli et al. 2021). Mason Spur is also situated on the margins of the so-called Discovery accommodation zone, a prominent tectonic feature that crosses the Transantarctic Mountains and separates two discrete structural blocks (Wilson 1999).
The zone acts as a natural focus for major trunk glaciers crossing the mountains and is likely to have experienced episodes of vertical movement, although poorly quantified (Wilson 1999;Olivetti et al. 2018). Mason Spur also lies within a southern extension of the Terror Rift, one of the youngest parts of the WARS (Fig. 1). The Terror Rift formed either during the Middle Miocene epoch after 13 Ma (Fielding et al. 2006;Fielding 2018) or else as a result of a change in relative plate motion forces, from orthogonal to dextral strike-slip, after 26 Ma (Granot and Dyment 2018). Faulting and associated changes in elevation are also known to have continued sporadically in the Terror Rift during Pliocene, Pleistocene, and Recent times (Fielding et al. 2006;Paulsen and Wilson 2009;Sauli et al. 2021), so changes in base level are likely to have affected Mason Spur on more than one occasion.
The erosion recorded at Mason Spur probably commenced soon after caldera collapse and recurred frequently thereafter (e.g., prior to eruption of each of the unconformably overlying units MSP2-MSP4, formed between 13 and 12 Ma), resulting in a cumulative erosional relief of hundreds of metres. The presence of intrusions with even younger ages (to 11.2 Ma) that were exhumed prior to eruption of units MSP5 and MSP6 suggests that significant erosion also occurred after 12 Ma. It seems likely that many of the category (2) intrusions (both trachytic and rarer basanites) may be feeder dykes for some of the younger units present as well as other units that have been completely eroded; the intrusions are included within MSP1 simply as a convenience because they occur within the MSP1 outcrop and are truncated by erosion.
Following eruption and caldera collapse of the large trachyte ignimbrite volcano (MSP1), trachyte and minor basanite magmas fed activity at several small centres resulting in effusion of trachyte (MSP2) and basanite (MSP5) lava flows and white trachytic ignimbrite deposits (top of MSP3), then further basanite lava flows (MSP6) and trachytic and basanitic pyroclastic cones (MSP4, MSP7). Unit MSP2 comprises subaerial trachyte lavas, which were probably erupted during two discrete effusive episodes, both undated but older than units MSP4 and MSP7 (Fig. 4). The thick sedimentary deposits of unit MSP3 infilled a deep ravine cut in unit MSP1 (Figs. 4, 7). Fluvial conditions have been inferred for some of the sediments within unit MSP3, associated with lahar and lahar runout deposits. The interbedded fluvial deposits are subordinate to other sediment types (see Supplementary Information S5). It is generally assumed that the Antarctic climate became steadily more frigid after c. 14 Ma, and rapidly transitioned to a predominantly polar glacial regime (Lewis et al. 2007). However, modelling of the Middle Miocene epoch suggests that episodes of amelioration probably also occurred, consistent with tundra conditions (Halberstadt et al. 2021), when flowing water (rivers, streams) may have developed during summers. Unit MSP6 unconformably overlies unit MSP3 and, as well as basanite lavas, it contains mass flow deposits that probably accumulated within the same ravine as MSP3. The ravine may have been rejuvenated perhaps by further uplift and downcutting. Additional magmas were erupted explosively from at least four small tuff cone centres (MSP4 [trachyte], MSP7 [basanite]), mainly after a depositional or erosional gap lasting c. 1 Myr. Finally, basanite lava flows interacted with surface water (or meltwater) at a high elevation (c. 1000 m) at 12.1 Ma (MSP5). The location of its source vent is unknown but is probably obscured by the outcrop of unit MSP8.

Stage 3: final trachyte activity (dome and block-and-ash-flow units) (MSP8; c. 6 Ma)
Effusion of a small but prominent trachyte dome (MSP8a) and associated block-and-ash-flow deposits (MSP8b) at the summit of Mason Spur marked the final pulse of trachyte volcanism at Mason Spur (Fig. 9). The event was subaerial, lacking any obvious interaction with water, and is dated at 6.03 ± 0.01 Ma, within error of the K-Ar age of 6.13 ± 0.20 Ma obtained by Wright-Grassham (1987) ( Table 3). Following a hiatus in volcanism potentially indicating a repose period of c. 5 Myr, younger activity at Mason Spur was exclusively basanitic and lasted sporadically until late Middle Pleistocene time. Because of its markedly younger age, the younger basanitic activity represented by units MSP9 and MSP10 probably corresponds to a fresh influx of magma. It may be related to a Pleistocene extensional tectonic phase in the Terror Rift that permitted mafic magmas to rise along deep crustal fractures (e.g., Fielding et al. 2006;Rilling et al. 2009;Sauli et al. 2021), similar to basanite magma that was sporadically erupted, in small volumes, during the Miocene stage 2 eruptive phase (Fig. 9).

Basanite lava-fed deltas sourced from neighbouring volcanoes
It has been suggested that the sequence of 'a'ā lava-fed deltas at Helms Bluff was probably sourced from a vent in the vicinity of Mount Discovery (Smellie and Martin 2021). Our new age for the middle sequence at Helms Bluff (4.92 ± 0.04 Ma) improves on the K-Ar age (4.51 ± 0.31 Ma) by Wright-Grassham (1987 ; Table 3). The Helms Bluff and Snow Petrel Peak sequences are remarkably similar, with subhorizontal capping lavas and associated breccias each > 200 m thick. We suggest that both sequences were erupted from vent(s) on Mount Discovery and may be broadly the same age. 'A'ā and likely pāhoehoe lava-fed deltas also occur at Helms Deep. Despite the 'a'ā delta outcrop being significantly eroded, with its coeval capping subaerial 'a'ā lavas entirely removed, paradoxically it yielded the youngest age in our study. Both outcrops at the locality appear to dip gently to the southeast, suggesting that they drape the Mason Spur succession and issued from vents on Mount Morning. Because all of the lava-fed delta outcrops examined issued from vents extrinsic to the Mason Spur volcano, they presumably represent discrete basanite magma batches unrelated to each other and to those that erupted at Mason Spur.

What caused the longevity of volcanism at Mason Spur?
Our study has highlighted how volcanism at Mason Spur extended between c. 13 Ma and present, which would be unusually long if they were associated with a single evolving polygenetic centre. However, Mason Spur is situated within the WARS close to the rift margin. It sits above relatively thick crust adjacent to the Transantarctic Mountain front, a major lithospheric discontinuity (Wright-Grassham 1987;Wilson et al. 2007;Martin et al. 2015Martin et al. , 2021b. It is also situated at the southern extension of the Terror Rift proposed by Horgan et al. (2005; Fig. 1) and on the southern margin of the Discovery accommodation zone. According to Granot and Dyment (2018), extension caused by rifting in the WARS ended at c. 11 Ma. Younger fault activity associated with the Terror Rift was considered to be related to intraplate deformation, perhaps associated with dextral strike-slip activity (Sauli et al. (2021). The younger faults are deep structures clearly imaged on seismic profiles (Sauli et al. 2021) and, along with a driver such as edge-driven mantle flow (Panter et al. 2018), would have facilitated the uprise of mafic magmas during extensional episodes (Rilling et al. 2009). While the history of the Discovery accommodation zone is unclear, the Terror Rift has experienced repeated extension since its initiation, continuing into the Pliocene and Pleistocene, and the rift may still be active today (Paulsen and Wilson 2009;Sauli et al. 2021). Both zones of structural weakness, and the presence of a translithospheric fault promoting edgedriven mantle flow (Faccenna et al. 2008, Panter et al. 2018Panter and Martin, 2021), may be responsible for the eruption of mafic (mainly basanitic) magmas at Mason Spur after the initial burst of trachytic activity at 13-12 Ma. Thus, we envisage basanite magma being emplaced into thick, riftmarginal crust where it fractionated to yield the trachytes of unit MSP1 (Wright-Grassham, 1987). Later basanite magmas were smaller in volume and were linked to lesser extensional events. They traversed the crust via deep faults linked to the Terror Rift and the mafic compositions suggest that they probably did not stall in shallow crustal magma chambers. The magmatic longevity seen at Mason Spur is therefore a consequence of magma production within an evolving rift experiencing multiple episodes of extension over several million years.
What triggered the major caldera-forming eruption at Mason Spur (unit MSP1)?
Significant caldera eruptions, like that which occurred at Mason Spur, are associated with eruption rates at the middle to upper end of the range preserved in the rock record (Costa and Martí 2016;Giordano and Cas 2021). Some of the largest are related to felsic caldera eruptions, e.g., Oruanui, New Zealand . The common occurrence of felsic batholiths in the rock record shows that accumulation of evolved magma in the upper crust is not unusual but the infrequency of significant felsic caldera eruptions compared with felsic batholith occurrences indicates that specific conditions must be met in order to trigger those eruptions (Jellinek and DePaolo 2003). Felsic caldera eruptions from large magma chambers typically occur in settings characterised by low rates of magma production (Jellinek and DePaolo 2003). Other factors may be important (e.g., far-field tectonic forces, magma crystal content, lateral magma transport, effective viscosity of wall rocks, thick continental crust, and deglaciation or ice cap retreat; Jellinek and DePaolo 2003;Costa and Martí 2016;Sulpizio and Massaro 2017). The eruptions also involve shallow (4-10 km) magma chambers and they occur in island arc, continental rift or extensional settings (Jellinek and DePaolo 2003;Costa et al. 2011;Gregg et al. 2015;Costa and Martí 2016). Rates of magma production in the WARS are roughly 10 −7 km 3 year −1 , assuming a maximum volume of 10 6 km 3 erupted over 50 Ma ( Behrendt et al. 1994;Rocchi et al. 2002). This overlaps with the range of magma production of other, similarly sized continental rifts (Panter 2021), e.g., the Basin and Range province (USA) at 10 −5 to 10 −7 km 3 year −1 (Crowe 1986). At these slow magma production rates, the relatively large magma chamber needed to feed the > 85 km 3 Mason Spur eruption would have required a long period of accumulation without eruption, probably 10 5 years or more (Jellinek and DePaolo 2003).
To trigger a significant caldera eruption requires (1) magmatic overpressure or (2) external forces, such as far-field stresses or foundering of the magma chamber roof (e.g., Wilson et al. 2006;Costa et al. 2011;Allan et al. 2012;de Silva and Gregg 2014;Gregg et al. 2015). The former involves a triggering pressure within the reservoir that exceeds the lithostatic confining pressure (Gregg et al. 2015). It may occur during magma recharge leading to variation in buoyancy, crystal cargo or volatile content (Wark et al. 2007;Caricchi et al. 2014;Malfait et al. 2014). Gregg et al. (2015) calculated that a rheological threshold for an eruption is surpassed once the magma chamber volume exceeds around 100 km 3 . The role of far-field stresses is often harder to determine given that the trigger may not be preserved in the rock record, something Allan et al. (2012) referred to as the invisible hand. In the following paragraphs, we outline the conditions relevant to Mason Spur unit MSP1 that may lead to a significant eruption, including our preferred hypothesis, without implicitly demonstrating causation.
The WARS is conventionally considered to have stalled at c. 26 Ma (DeMets and Merkouriev 2016; Sauli et al. 2021), which implies that there was no far-field stress driver. However, motion between West and East Antarctica may have continued until Middle Neogene time, c. 11 Ma (Granot and Dyment 2018). This includes accommodation of motion in the Terror Rift offshore of southern Victoria Land. The revised middle Neogene age of rifting coincides with the significant felsic caldera eruption at Mason Spur (MSP1) and a causal link is plausible.
Magmatism in the Mount Morning Volcanic Field is considered to have commenced by at least 25 Ma (Martin et al. 2021a, b). McMurdo Volcanic Group magmatism is generally thought to have anomalously heated the lithosphere both in Victoria Land and in the western Ross Sea (Berg et al. 1989;Wörner and Zipfel 1996;Martin et al. 2014;Shen et al. 2020;Wiens et al. 2021). Thus, the MSP1 ignimbrite eruptions at Mason Spur were preceded by more than 10 Ma of magmatism, during which the crust was thermally preconditioned (Jellinek and DePaolo 2003).
Deglaciation might be an additional factor (Jellinek et al. 2004;Geyer and Bindeman 2011;Nyland et al. 2013;Sulpizio and Massaro 2017). For example, a retreating ice body less than a kilometre thick may depressurize shallow magma systems, whilst larger ice masses may even influence melt generation down to mantle depths (Gudmundsson 1986;Sigmundsson et al. 2010). A potential contributing factor triggering the eruption of MSP1 at Mason Spur might thus be significant fluctuations in ice thickness during and immediately following the Middle Miocene Climate Transition, a period of significant climate instability (Halberstadt et al. 2021;Levy et al. 2021).
In summary, major conditions favorable for a significant felsic caldera eruption at Mason Spur involved (1) a rift setting (WARS/Terror Rift), and (2) motion between East and West Antarctica continuing to the middle Neogene. Additional factors that probably contributed include (3) a low rate of magma production, (4) thermal pre-conditioning of the lithosphere, and possibly (5) ice unloading. All of these circumstances may have acted in unison.

Comparison of Mason Spur with other volcanoes in the West Antarctic Rift System
As a result of this study, the Mason Spur volcanic complex is now one of the best described and dated centres in the WARS. The only comparable study is by Panter et al. (1994) for Mount Sidley (Marie Byrd Land). However, information is also available for a few other well-exposed centres such as Crary Mountains and Mount Murphy (Marie Byrd Land), and Minna Hook and Hallett Coast volcanoes (Victoria Land;LeMasurier et al. 1994;Wilch and McIntosh 2002;Smellie et al. 2011a;Smellie and Martin 2021;Smellie and Rocchi 2021;Wilch et al. 2021). None closely resembles Mason Spur. The principal distinguishing features of the Mason Spur volcanic complex are (1) abundant ignimbrites (unit MSP1 and top of unit MSP3); (2) conspicuous epiclastic rocks (mainly unit MSP3); (3) several tuff cones (MSP4 and MSP7); and (4) presence of lava domes (unit MSP8 and implicated in unit MSP3). Moreover, the volcanism at Mason Spur appears to have occurred intermittently and frequently over a period of c. 13 Myr, longer than any other WARS volcanic centre documented so far, apart from Mount Morning (Martin et al. 2021a;Wilch et al. 2021).
Lava domes are scarce in the WARS but are present on several of the large volcanoes in the Erebus Volcanic Province (Wright-Grassham 1987;Smellie and Martin 2021). Mount Sidley is principally a phonolite-trachyte volcano with a 1.5 Myr life span. It formed from the overlapping products of three major centers (Panter et al., 1994). Two of the centers experienced caldera collapses but they lack tephras ejected during the caldera-forming events. Ignimbrites are present on one centre and were probably erupted from a flank vent. They are volumetrically minor and the bulk of the erupted products are 'a'ā lavas; one center also experienced a significant trachytic airfall eruption. The evolution of all the other WARS centers are alike, and they are dominated by effusion of mafic glaciovolcanic 'a'ā lava-fed deltas (Smellie 2021). Apart from numerous small scoria cones and a few tuff cones, tephras are minor and epiclastic deposits are virtually absent (e.g., glacial diamicts at Mount Murphy; Wilch and McIntosh, 2002). Overlapping centers also occur at Crary Mountains and along the Hallett Coast, the latter linked to eruptions from major coast-parallel fissures (Wilch and McIntosh 2002;Smellie and Rocchi 2021;Wilch et al. 2021). Calderas are present at several of the Hallett Coast centres and elsewhere in the Erebus Volcanic Province but they lack associated caldera-related tephras. Ignimbrites are absent. Like Mason Spur, the lifetimes of the major edifice-building periods for volcanoes within the WARS are typically about 1-2 Myr. Much younger activity, sometimes separated by as much as 8 Myr, often occurred but is represented mainly by scoria cones (Wilch et al. 2021) whereas tuff cones are a prominent feature of the Mason Spur volcanic complex.
Despite the abundance of large polygenetic volcanoes with prominent summit calderas within the WARS (Panter et al. 1994;Smellie and Martin 2021;Smellie and Rocchi 2021;Wilch et al. 2021), large volumes of ignimbrites are associated only with two poorly known volcanic centres in Victoria Land, at Deception Plateau and possibly Malta Plateau c. 600 km north north-east of Mason Spur (Smellie and Rocchi 2021). Only one volcano in the WARS, at Mount Waesche (Marie Byrd Land), is associated with caldera collapse-related deposits (Dunbar et al. 2021). The Mount Waesche caldera is small in comparison (c. 2 km diameter) and the inferred deposits form a widespread thin (few metres) drape of polymict breccia on the volcano slopes; ignimbrite deposits are absent. Ignimbrites associated with caldera collapses also characteristically contain abundant accessory and accidental clasts derived from deep within the volcano interior and underlying basement (e.g., Druitt 1985;Druitt and Bacon 1986;Allen and Cas 1998). These clasts are absent from MSP1. Caldera collapses in large volcanic systems like Mason Spur, which vented abundant ignimbrites, are normally associated with explosive eruptions and the formation of thick extra-caldera ignimbrite deposits, except in unusual circumstances (e.g., Hildreth and Fierstein 2000;Lipman and McIntosh 2008).
The absence of major ignimbrite deposits preserved elsewhere in the WARS may be due to a combination of factors: (1) insufficient exposure into the deep interiors of other volcanoes; this seems unlikely as a general explanation given the absence or scarcity of ignimbrites from those volcanoes that are deeply eroded (Minna Hook, Hallett Coast volcanoes, Mount Sidley, Mount Murphy and Crary Mountains); or (2) the evidence was removed by ice advection. Smellie (2021) proposed that tephra-producing eruptions, including any ignimbrites, that occur during glacial periods would deposit their tephras in the surrounding ice. Ice movement would carry the tephras to the ice margins and dump the deposits into the sea, thus leaving little or no terrestrial record. Given the history of prolonged glacials in Antarctica (Halberstadt et al. 2021;Levy et al. 2021), this seems the most likely explanation for the scarcity of pyroclastic rocks generally within the WARS.

Conclusions
Mason Spur is the remnant of a polygenetic volcanic complex in southern Victoria Land, part of the Mount Morning Volcanic Field in the Erebus Volcanic Province (McMurdo Volcanic Group). The deep interior of the complex is unusually well exposed due to the coastal location and the combined effects of focussed marine and glacial erosion. The outcrops have been divided into multiple stratigraphical units based on field relationships, the predominant lithofacies, and differences in isotopic age. Together they provide a rare and exceptionally detailed geological history involving eruptions over a particularly prolonged period (c. 13 Myr) within the intracontinental West Antarctic Rift System.
The oldest-exposed and most voluminous eruptive unit (MSP1) consists of indurated, deformed, massive trachytic ignimbrites together with pre-and post-caldera hypabyssal intrusions. The ignimbrites are undated but were probably erupted around c. 13 Ma (Middle Miocene). They were invaded by abundant hypabyssal intrusions, the oldest of which were pervasively brecciated during a major volcanotectonic event (i.e., that which formed the caldera). MSP1 is interpreted as a caldera fill and empirical calculations suggest that a minimum volume of c. 85 km 3 of volcanic products accumulated within the caldera. The caldera must have formed between 13 and 12.6 (possibly 12.9) Ma, the latter being an age obtained on the oldest fluidal-margined intrusion emplaced in the early post-caldera period. Unit MSP1 was profoundly eroded prior to and following the eruption of several younger units. A substantial palaeo-relief was created, possibly triggered by base-level readjustment due to uplift. Ravines hundreds of metres deep were cut and infilled by trachytic coarse mass flow deposits (MSP3). These are interbedded with sandstones representing multiple lahar and lahar runout deposits, and were probably sourced in gravitational collapses of one or more trachyte domes. Unit MSP3 also includes a thin unit of fresh white trachytic ignimbrites. It was succeeded, after further erosion, by effusion of thick basanite lava flows and additional mass flow deposits (MSP6). The intrusive activity represented within MSP1 lasted from 12.9 until 11.2 Ma. The intrusive magmas, which are mainly trachytic but include rare basanites and alkali basalts, probably fed multiple small centres, resulting in eruption of trachytic lava flows (MSP2) and a tuff cone (MSP4), together with basanitic lava flows (MSP5, MSP6) and pyroclastic cones (MSP7). This phase of activity lasted until about 10 Ma.
After a gap of c. 4 Myr, the final episode of trachyte eruption occurred at 6 Ma, creating a small trachyte lava dome and associated hot-emplaced block-and-ash-flow deposits (MSP8) followed by a prolonged pause in activity lasting 5 Myr. The most recent volcanic episode at Mason Spur is represented by two groups of small-volume basanitic magmas from several Pleistocene scoria cones (MSP9, MSP10), erupted at 0.94 and 0.37 Ma. The eruptions were dominantly Strombolian but early-formed lava flows in unit MSP9 were water-cooled, perhaps due to interaction with surface snow. Additionally, substantial basanitic lava-fed deltas of Pliocene and Pleistocene age were erupted from vents located on the Mount Discovery and Mount Morning volcanoes, at 4.9 and 0.04 Ma, respectively. They are preserved as relicts flanking Mason Spur to the north, north-east and south-west.
The eruptive history and construction of the Mason Spur volcanic complex are dissimilar to any other volcanic center within the West Antarctic Rift System described so far. The most striking differences are the prolonged history of eruptions to form the present-day volcanic complex, and the abundance of ignimbrites in the main edifice-building phase. The differences are probably due to the location of the centre within a rift that was active prior to c. 11 Ma, which helped to trigger voluminous trachyte eruptions from a large crustal reservoir followed by a major caldera collapse. Subsequent eruptions were from multiple, small, mainly basanite centres linked to deep crustal faults created by residual intraplate deformation.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.