The Arabian harrats¹ 1

“Ḩarrat” is the Arabic possessive noun referring to a particular expansive region of dark, rubbly volcanic rocks; “harrah” or “harra” is the general or non-possessive noun.

Harrat volcanism also does not fit comfortably into commonly used classifications of basaltic volcanoes and volcanic systems. The small scoria cone clusters readily meet expectations of monogenetic volcanism, but the harrats consist mostly of fields of overlapping lava flows that erupted intermittently from approximately linear vent arrays over periods of a million or more years in a particular area. Their persistent activity in focused areas is a defining characteristic not captured by the possibly monogenetic character of any particular vent or eruption. As will be shown, this prolonged spatial localization produces intermediate and evolved magmas that clearly are not monogenetic, among other consequences. Owing to their broad areas, the harrats have been described as medium-sized flood basalts (Walker, 1993), but they lack the volumes and brevities typical of flood eruptions. Their prolonged activity in particular locations shares aspects of central volcanoes, but in most cases their magma supply systems are insufficiently localized and sustained to have produced summit calderas. Their long-term growth by repeated eruptions along linear vent chains constructs elongated edifices that resemble, at least in landform, the rift zones of large shield volcanoes. However, evidence is absent, or has not been recognized, for appreciable along-rift transport of magma from a central location of supply, unlike the rift-zone eruptions familiar from Hawaiian shield volcanoes. Too repeatedly active and long-lived for monogenetic to be their meaningful descriptor, but insufficiently active to support the focused feeder systems of shield and central volcanoes, they are perhaps best characterized simply as sustained basaltic volcanic fields, and “harrats” might be adopted more widely as a general term for such volcanic systems.

This report’s goal is to introduce and summarize the overall characteristics of the Cenozoic distributed volcanism of the Arabia Plate. Our familiarity with this area stems from a collaborative geologic and geophysical project between the Saudi Geological Survey (SGS) and the U.S. Geological Survey (USGS) studying the volcanic history and hazards of the northern portion of Ḩarrat Rahat adjacent to the city of Al Madīnah al Munawwarah (hereafter, Al Madīnah), Kingdom of Saudi Arabia. Results of that work are available as journal articles, cited where appropriate in this chapter, a geologic map (Downs and others, 2019; Robinson and Downs, 2023), and compiled in USGS Professional Paper 1862 (Sisson and others, 2023a, and chapters therein). Other key sources of information are geologic maps of some major harrats, with explanatory pamphlets, produced by Vic Camp and John Roobol, released through the SGS or its institutional predecessors (Camp and Roobol, 1991a; Roobol and Camp, 1991a,b) and associated interpretive journal articles (Camp and others, 1987, 1991, 1992; Camp and Roobol, 1989, 1991b, 1992). Also important are USGS summaries of harrat volcanism by Coleman and others (1983), Brown and others (1989), and Coleman (1993). The detailed volcanic geology of northern Ḩarrat Rahat was first investigated and interpreted by Moufti (1985), and all subsequent studies have adopted the names therein for places and most volcanic products. Moufti and Németh (2016) provide photographs, place names, and basic interpretations for noteworthy volcanic landforms in some Saudi Arabian harrats. Al-Onaizan and others (2017) list, name, and locate Saudi Arabian volcanic landforms exhaustively (in Arabic). A joint effort between King Abdulaziz University and the University of Auckland called the Volcanic Risks in Saudi Arabia (VORiSA) Project was carried out in 2011–2014 and focused principally on the physical volcanology of northern Ḩarrat Rahat and how this informs assessments of hazards (Lindsay and Moufti, 2014). Summaries of Cenozoic tectonics are presented by Bohannon and others (1989), Coleman (1993), Bosworth and others (2005), Bosworth (2015), and Reilinger and others (2015). Withstanding the test of time is Powers and others’ (1966) authoritative description of the Arabian Platform sedimentary sequence. Stern and Johnson (2010) and Johnson and Kattan (2012) compiled and interpreted the geologic history of the Precambrian Arabian Shield.

Since the earliest conceptions of continental drift (Lartet, 1869; Wegener, 1915), the Red Sea, Gulf of Aden, and their flanking landmasses were recognized as the archetype of continental rifting, and the region has since figured prominently in developing and refining theories of plate tectonics (Baker, 1970; McKenzie and others, 1970; Burke and Dewey, 1973; Coleman 1974). Cenozoic tectonic events that influenced the region (fig. 1) included (1) collision of the Afro-Arabia Plate’s northeastern continental margin against Eurasia thereby creating the Bitlis-Zagros suture as subduction consumed the last of the plate’s leading oceanic edge and closed the Neotethys Ocean, (2) the Afar mantle plume impinging beneath the (now) Afar depression at the north end of the (now) East African Rift, (3) opening of the Red Sea and Gulf of Aden rifts separating the Arabia and Africa Plates, and (4) the Gulf of Aqaba-Dead Sea Transform fault system separating the Sinai block as the Arabia Plate continued to converge beneath Eurasia.

Associated volcanism on the continental portion of the Arabia Plate is divisible broadly into four stages, not wholly correlative with the aforementioned tectonic events: (1) eruptions of plume-impingement flood basalts around 30 million years ago (Ma) that blanketed what are now the Ethiopian highlands, southern Sudan, and western Yemen prior to the openings of the Red Sea, Gulf of Aden, and the Afar depression; (2) early harrat basaltic volcanism that immediately followed, forming a belt of centers scattered northward across the Arabia Plate as far as Ḩarrat ash Shamā (also known as Ḩarrat ash Shaam, Ḩarrat Shama, or Ḩarrat al Ḩarrah) in Syria, Jordan, and northwestern Saudi Arabia, or perhaps to Karaca Dağ volcanic field (Turkey); (3) regional tholeiitic basaltic diking in the period 24–20 Ma during initial Red Sea rifting, exposed mainly as rift-parallel dikes and small gabbroic intrusions on the Red Sea’s subaerial eastern flank and as scattered basaltic volcanic fields including in the Cairo basin (Egypt) subsurface, with diminished volcanism continuing to about 15 Ma; and (4) dominantly mafic alkalic volcanism commencing around 13–12 Ma, and continuing to the present, that produced harrats and scattered smaller centers with well-preserved constructional volcanic relief and young-appearing volcanic landforms (Coleman and others, 1983; Bohannon, 1986; Bohannon and others, 1989; Camp and Roobol, 1989; Coleman, 1993; Ilani and others, 2001; Bosworth and others, 2005; Bosworth, 2015; Bosworth and Stockli, 2016). There may have been a hiatus, or nearly so, between the third and fourth stages (Ilani and others, 2001), but few efforts have been undertaken to date older harrat rocks, and their concealment beneath younger lavas hinders detailed studies and interpretations. Off-continent volcanism entailed Neogene and Quaternary submarine spreading-ridge and associated magmatism in the Red Sea and Gulf of Aden. This volcanism continues to the present and constructed the Arabia Plate’s western and southern oceanic crustal margins locally building the emergent and historically active volcanic islands of Jazā’ir az Zubayr and Jabāl aţ Ţayr in the southern Red Sea (fig. 1, no. 28).

Physiographic developments correlate with these tectono-magmatic phases. Most of the Arabian Shield now stands about 1,000 meters (m) above sea level but is covered unconformably to the north, east, and south by the gently outward-dipping Arabian Platform sedimentary sequence (fig. 1) that records marine passive-margin sedimentation spanning from the Cambrian, or perhaps latest Proterozoic, through the Neogene (Powers and others, 1966). Marine sedimentary rocks of this sequence crop out at elevations as great as 500 to 800 m above sea level, and in some areas as high as 1,200 m, indicating substantial post-depositional uplift but without short wavelength folding indicative of strong, localized compression, and, likewise, with few extensional faults; the shield appears to have uplifted en masse, or nearly so. Long-term development of the passive-margin sedimentary sequence is not relevant to this chapter, but a regional unconformity or disconformity high in the sequence places lower Miocene and younger clastic rocks atop upper Eocene marine carbonates (Powers and others, 1966; Swift and others, 1998). An unconformity at this level also cuts out most to all Oligocene strata in the Gulf of Suez and northern Red Sea, and this unconformity incises as deeply as Lower Cretaceous strata in the southern Red Sea (Bosworth, 2015). A late Oligocene plate-scale event of modest uplift and erosion, or non-deposition, followed by a regional change from carbonate to clastic sedimentation, accords with reconstructed plate motions indicating collision between the Afro-Arabian and Eurasian continental land masses commencing near the end of the Oligocene. This collision divided the Neotethys Ocean by early Miocene time (McQuarrie and others, 2003; Agard and others, 2011; McQuarrie and van Hinsbergen, 2013; Torsvik and Cocks, 2017).

Through improvements in dating, flood volcanism associated with the Afar mantle plume is now known to have taken place mainly in a brief pulse at 31–30 Ma, continuing sporadically to about 26.5 Ma, ending in some areas with voluminous eruptions of rhyolite (Bosworth and Stockli, 2016). In western Yemen, the flood basalts and succeeding rhyolites accumulated to thicknesses of 1–2.5 km and constitute about 10 percent of the ~600,000 km² area and greater than (>) 350,000 cubic kilometer (km³) volume of the combined Ethiopia-Sudan-Yemen Traps (Baker and others, 1996; Menzies and others, 1997). Evidence around the Ethiopian highlands points to plume-impingement uplift perhaps commencing as much as 10 million years prior to flood volcanism (Sembroni and others, 2016; Faccenna and others, 2013, 2019), although interstratification of basal flood-lavas with shallow marine sedimentary rocks shows that some areas were near sea level at the onset of flood eruptions (Coleman, 1993). Most uplift associated with the Afar plume, however, followed emplacement of flood lavas (Menzies and others, 1997; Sembroni and others, 2016; Faccenna and others, 2013, 2019) and is apparent today as the high topography surrounding the Afar depression, including across the Red Sea in western Yemen and southwestern Saudi Arabia (fig. 1).

Much or all of the pre-rift Nubian-Arabian Shield was at elevations close to sea level in the Oligocene, as shown by marine mudstone and limestone intercalated with basal lavas of Ḩarrat Hadan (28–10 Ma) (Madden and others, 1980; Sebai and others, 1991) and lateritic paleosols or chert-pebble coquina exposed beneath basal lavas of other old harrats (Coleman, 1993). Lateritic paleosols and fossiliferous mudstones also underlie the flood basalts of the Yemen Traps, but ⁸⁷Sr/⁸⁶Sr isotopic values of fossils in those underlying rocks are unlike Oligocene seawater, suggesting lacustrine, rather than marine, paleoenvironments (Menzies and others, 1997). Geukens (1966) also identifies fossils in sedimentary strata within the Yemen Traps as fresh-water gastropods, lamellibranchs, and ostracods. The Red Sea rift basin had developed by the early Miocene into a distinctly bounded bathymetric low, as recorded by marine shales, sandstones, and carbonates that transgressed atop red beds. By the late Miocene, the basin was sufficiently deep and confined to accumulate great thicknesses of evaporitic salt (Coleman, 1993; Bosworth, 2015). Rift initiation along the Red Sea and Gulf of Aden appears to have been contemporaneous at 22±3 Ma, with the Red Sea initially opening orthogonal to the rift axis, but with the spreading rate approximately doubling at 11±2 Ma and the spreading direction becoming more north-northeasterly (Reilinger and others, 2015). This increase in spreading rate and change in spreading direction correspond, within uncertainty, with the onset of Gulf of Aqaba-Dead Sea Transform faulting and with the establishment of well-organized seafloor-spreading volcanism along the length of the Gulf of Aden (Bohannon and others, 1989; Reilinger and others, 2015; Szymanski and others, 2016). Rift-flank uplift near this time produced the Great Escarpment—the steep, west-facing mountain front along which the Ḩijāz highlands, from Makkah north to the Gulf of Aqaba, and the ʻAsīr highlands, to the south into Yemen, overlook the narrow Red Sea coastal plain (Tihāmat al Ḩijāz, Tihāmat ʻAsīr, and Tihāmat al Yemen). Harrat volcanism also became widespread and vigorous at about 12–10 Ma (Camp and Roobol, 1989; Coleman, 1993), with some large lava flows draping the Great Escarpment and reaching the Red Sea coastal plain, clearly showing that voluminous effusions continued after pronounced uplift (Camp and Roobol, 1989, 1991a; Ball and others, 2023). Extensional faults cut old harrats locally (Szymanski and others, 2016), but few such faults displace the younger harrats and only by small amounts; similarly, the Arabian Shield, although elevated, is low-relief and largely free of fault-bounded horst and graben topography, collectively indicating small-to-negligible amounts of crustal extension and thinning to the east of the Great Escarpment. The coincidence of events around and across the Arabia Plate at about the time of the middle to late Miocene transition is suggestive of a common cause for the diverse tectonic, geomorphic, and magmatic changes—perhaps widespread loss of the lower lithosphere of the Arabia Plate was focused beneath the shield and its capping belt of harrats (McGuire and Bohannon, 1989; Calvert and Sisson, 2023; Sisson and others, 2023b).

Salters and others (2023) and Sisson and others (2023b) address the origins of northern Ḩarrat Rahat’s basalts from the standpoints of radiogenic isotopes and trace and major elements. Northern Ḩarrat Rahat’s basalts prove to be the closest among the Arabian harrats to northern hemisphere MORB with respect to their lead (Pb), neodymium (Nd), hafnium (Hf), and strontium (Sr) isotopic values but are offset from such MORB consistent with incorporation of as much as 30 weight percent of material from (or like) the Afar mantle plume. Neogene and Quaternary (post-trap) volcanic rocks of Yemen and the Gulf of Aden that erupted closer to the Afar depression show greater isotopic influences of the Afar plume, and the signature of the Dupal³ 3

A region of the Indian and south Atlantic Oceans where midocean ridge basalts have elevated ²⁰⁸Pb/²⁰⁴Pb and ²⁰⁷Pb/²⁰⁴Pb values relative to midocean ridge basalts of the northern hemisphere and Pacific Ocean.

Collectively, these results support a relatively simple scenario wherein chemically depleted asthenospheric upper mantle peridotite, mixed with subordinate materials from (or like) the Afar mantle plume, upwells beneath the axis of the harrat belt and melts to small extents as it passes from the garnet to spinel stability fields. Melting does not continue to shallow depths in the spinel field, unlike MORB, so most liquids remain alkalic. That melting ceases deep in the spinel field is consistent with upwelling having impinged on the base of the Arabia Plate lithosphere. Segregation pressures deep in the spinel field of 1.8–2.0 gigapascals (GPa) (Sisson and others 2023b) would place this at depths of approximately 60–70 km, consistent with detection of the lithosphere-asthenosphere boundary in that depth range beneath the main harrat belt both by magnetotellurics (Bedrosian and others, 2019; Peacock and others, 2023) and by seismic receiver functions (Blanchette and others, 2018). This is, however, considerably shallower than the depth expected for the base of Neoproterozoic lithosphere (Artemieva, 2006), a topic that will be addressed subsequently.

The origin of northern Ḩarrat Rahat’s basalt-hawaiite-mugearite-benmoreite-trachyte compositional spectrum is relatively simple. The magmas erupt through the Neoproterozoic Arabian Shield, so crustal assimilation is to be expected. This is confirmed by Pb isotopes that shift toward the common-Pb values of exposed shield rocks as magnesium oxide (MgO) concentrations or Mg numbers (Mg# = molar Mg/[Mg+Fe²⁺]) of the volcanic rocks decrease (Salters and others, 2023). The shifts are, however, small (fig. 10) and can be accounted for by no more than 5 weight percent assimilation upon attaining trachytic magma compositions. Low extents of assimilation are also indicated by trace element ratios that fail to shift toward continental crustal values with progressive differentiation (fig. 11). Instead, crystallization differentiation, broadly defined, is the dominant differentiation process, accompanied by mixing between more and less evolved magmas and by interactions between earlier and later magmatic intrusions (Sisson and others, 2023b).

Differentiation of basalts through hawaiites has been investigated thoroughly because those magmas dominate erupted volumes (Sisson and others, 2023b). Magnesian alkali basalts (to nearly 13 weight percent MgO) differentiate by olivine fractionation to approximately 10 weight percent MgO whereupon olivine gabbroic fractionation commences; this continues to approximately 6–7 weight percent MgO, at which point titanomagnetite joins the crystallizing assemblage. Concentrations of SiO₂ of bulk magmas (approximately that of the melt) decrease modestly during the initial olivine gabbroic crystallization interval, but SiO₂ concentrations then increase once titanomagnetite joins the crystallizing assemblage. Crystallization of the olivine+titanomagnetite gabbroic assemblage continues through the hawaiites. Most of the basalts are mixtures, however, obscuring simple liquid lines of descent (fig. 12). Using enrichments of the highly incompatible elements thorium (Th), tantalum (Ta), uranium (U), and niobium (Nb) as indices of differentiation, and referencing to the median alkali basalt of northern Ḩarrat Rahat with Mg#>0.64 as the parent (Downs, 2019; Sisson and others, 2023b), titanomagnetite appears after about 60 weight percent crystallization, and the median hawaiite would result from about 75 weight percent crystallization. Apatite joins the crystallizing assemblage within the mugearites, shown by its appearance as microphenocrysts, and alkali feldspar joins and comes to dominate the crystallizing assemblage near the benmoreite-trachyte transition. Using the same parent and enrichments of highly incompatible elements, the extent of solidification to form the median mugearite is estimated as 83 weight percent, the median benmoreite as 86 weight percent, and the median trachyte as 95 weight percent. Some mugearites, hawaiites, and low-MgO basalts of northern Ḩarrat Rahat have P₂O₅/K₂O values appreciably greater than those of high-MgO basalts, inconsistent with simple, progressive crystallization-differentiation, but the anomalously high P₂O₅/K₂O values can result from those magmas having assimilated apatite-bearing cumulates produced from more evolved members of the harrat magmatic system (Sisson and others, 2023b). Re-assimilation of apatite-free cumulates is probably also common but cannot be recognized as readily.

Differentiation depths are not closely defined, but results of thermodynamic modeling are consistent with magmas mainly lodging and differentiating in the mid-crust or modestly deeper (Sisson and others, 2023b), and certainly not in the upper crust, in accordance with the absence of fumaroles or other surface geothermal features and a historical lack of upper crustal seismicity. Repetitive eruptive sequences of basalt through mugearite or benmoreite to trachyte of the Matan volcanic center of northern Ḩarrat Rahat point to events of progressive differentiation and provide estimates for their durations (Stelten and others, 2018). However, rocks of the nearby Al Efairia volcanic center do not show similar temporal-compositional sequences (Downs and others, 2023) indicating more complex differentiation processes. Consistent across all the major harrats, however, is that intermediate and evolved magmas have mainly erupted from the areas with the greatest aggregate constructional volcanic relief. The high relief of such areas is evidence that those are the sites of the greatest long-term magmatic fluxes. Such areas are expected to be underlain by the most numerous and voluminous associated crustal intrusions that yield intermediate and evolved melts both by direct, progressive crystallization-differentiation and by partial remelting upon magmatic replenishment (Stelten and others, 2020, 2023). Magnetotelluric and seismic imaging have failed to detect magma reservoirs at any depth in the crust beneath Ḩarrat Rahat (Bedrosian and others, 2019; Civilini and others, 2019; Peacock and others, 2023), indicating that such intrusions are too poor in melt, too small, or both, to be resolved by the applied methods.

The Arabia Plate consists mainly of Neoproterozoic rocks that assembled and stabilized by 750 Ma in its eastern portion (now largely concealed by the Arabian Platform) to 550 Ma in western parts where they are exposed in the Arabian Shield (Stern and Johnson, 2010). Applying the simple and area-weighted regressions between lithospheric thickness and terrane age of Artemieva (2006) yields expected lithospheric thicknesses in the east of at least 124 km (simple regression) or 132 km (weighted), and in the west of either at least 116 km or at least 126 km. Consistent with these is an estimate of 150 km to the base of the lithosphere in the southeastern portion of the Arabia Plate based on heat flow measurements in Oman, Saudi Arabia, and central Yemen (Rolandone and others, 2013). Using synthetic receiver functions, Hansen and others (2007) obtained similar depths to the lithosphere-asthenosphere boundary of 134–162 km beneath the Arabian Platform (median 144 km), but 69–123 km beneath the Arabian Shield including the major harrats (median 101 km), 52–78 km beneath the coastal plain and Great Escarpment (median 62 km), and 54–67 km beneath the Gulf of Aqaba-Dead Sea Transform fault system (median 63 km). Inversion of teleseismic wave travel times shows a belt of anomalously slow seismic velocity in the asthenospheric upper mantle centered beneath the most active harrats (Chang and others, 2011; Hansen and others, 2012; Yao and others, 2017), broadly underlying the pronounced shallowing of the lithosphere-asthenosphere boundary beneath the Arabian Shield (figs. 13 and 14). Magnetotelluric measurements independently detect the lithosphere-asthenosphere boundary as only at 70 km depth beneath northern Ḩarrat Rahat (Bedrosian and others 2019; Peacock and others, 2023). Thus, the mantle lithosphere beneath the belt of the most active harrats is thinner by approximately 50–80 km than would be expected for its age, defining an arch along the lithosphere’s basal contact with the underlying asthenosphere. Notably, this arch and its overlying belt of voluminous harrats do not match the locations and orientations of terrane boundaries or other structures exposed on the Arabian Shield (Stern and Johnson, 2010; Johnson and Kattan, 2012).

Restoring the Arabia Plate and the Sinai and Somali blocks to their positions relative to the Africa Plate prior to openings of the Red Sea and Gulfs of Aden, Aqaba, and Sinai places the south end of the belt of anomalously slow upper mantle seismic velocities adjacent to the (now) Afar depression at the (now) north end of the East African Rift system (fig. 14). This is consistent with these features collectively reflecting a formerly continuous structure, such as an arch in the base of the sub-continental lithosphere of east Africa and Arabia. Ebinger and Sleep (1998) recognized that because of substantial relief on the base of the Africa Plate, upwelling plume material could be channelized as it encountered and spread buoyantly beneath the lithosphere, producing belts of intraplate magmatism not obviously related to crustal terranes, active tectonics, or plate boundaries. This may have been the case for the Arabian harrat belt with portions of the Afar mantle plume, or adjacent ambient asthenosphere prompted into motion by the plume, spreading preferentially beneath a preexisting arch in the base of the lithosphere (Sisson and others, 2023b). Shear-wave speeds for the region are anisotropic with fast directions oriented generally north-south to about 20° west of north similar to the orientation of the harrats and the low-velocity belt in the upper mantle (Hansen and others, 2006; Chen and others, 2015); this fast direction can be interpreted as the direction of mantle flow. That the Red Sea did not rift along that belt may result from the Red Sea’s orientation being approximately perpendicular to the direction of “slab pull” in the early Miocene (Reilinger and others, 2015) with the Red Sea effectively being a product of passive rifting.

The opening of the Red Sea and Gulf of Aden would have severed any formerly continuous arch in the base of the continental lithosphere, interrupting simple channelized flow from the Afar mantle plume; yet, harrat volcanism continued on the Arabia Plate possibly owing to at least two processes. Elkins-Tanton (2005) proposed a feedback mechanism wherein magmas produced by localized mantle upwelling solidify as eclogite upon intruding the base of the sub-continental lithosphere, increasing the density of the intruded region and fostering its foundering. As foundering takes place, asthenosphere upwells to replace the masses of sinking lithosphere, and as it does so, the upwelling asthenosphere undergoes decompression melting; those melts then intrude the deep lithosphere and also solidify as eclogites, prompting further foundering, and thereby continuing the process. Once commenced, the phenomenon might erode some distance up through the mantle lithosphere in a localized region, with associated continued magmatism, although the long-term persistence of the process has not been assessed by modeling. Alternatively, and independent of plumes, arches in the base of the sub-continental lithosphere exist for some reason, the simplest being that those areas are hotter than is typical for the terrane’s age, as would be so if they are situated above persistent upwelling zones of the convecting ambient asthenosphere. Arrival of a plume might then accentuate upwelling already established beneath the thin lithosphere regions.

Harrat magmatism is a member of a cascade of phenomena that affected the Arabia Plate commencing in the Oligocene but accelerating in the Miocene. These include (1) the opening of the Red Sea and Gulf of Aden leading to organized seafloor-spreading magmatism, (2) the uplift of the Arabian Shield from near sea level to its high elevations with associated outward tilting of the Arabian Platform sedimentary sequence and creation of the Great Escarpment, and (3) the severance of the Arabia Plate from the Sinai block along the Gulf of Aqaba-Dead Sea Transform fault system. The confluence of these events across a region much broader than the harrat belt points to a regional weakening of the Arabia Plate, with its widespread, coherent uplift consistent with loss of mantle lithosphere across a region more extensive than the zone of notably reduced seismic velocities. Potentially, arrival of the Afar mantle plume, or of associated smaller plumes to its north, stimulated asthenospheric convection sufficient to thin and weaken the continental plate from below to the point of its failure across distances of thousands of kilometers (Calvert and Sisson, 2023).