Planet Mercury: Volcanism in a theatre of global contraction, with examples from the Hokusai quadrangle

Mercury's geological history has been dominated by global contraction caused by secular cooling of the planet's interior. This cooling has had a profound effect on the expression of the planet's volcanism and tectonism, and the expressions of these two surface evolutionary processes are deeply intertwined. Here, we use case studies from the Hokusai quadrangle of Mercury to gain insight into the interplay between Mercury's volcanism and tectonism, which we review throughout this paper. We perform the first crater size–frequency analysis of the southernmost extent of Borealis Planitia, Mercury's largest expanse of volcanic plains, and find that it formed ~3.8–3.7 Ga. We discuss the importance of “intermediate plains”, a widespread unit in the Hokusai quadrangle, as the manifestation of relatively low-volume effusions with an uncertain stratigraphic relationship with Borealis Planitia. Finally, we detail the formation of the Suge Facula pitted ground during the geological history of Rachmaninoff crater, and hypothesise that such textures probably formed more widely on Mercury but have often either been buried by thick lava flows or otherwise obscured. Unanswered questions in this work can be used to drive the next phase of Mercury exploration and research with the arrival of the BepiColombo mission.


Introduction
Results from the MErcury Surface, Space ENvironment, GEochemistry, and Ranging (MESSENGER) mission provided many answers about the history of volcanism on the Solar System's innermost planet, yet many questions remain. With the launch of the BepiColombo mission (Benkhoff et al., 2010;Rothery et al., 2020a), set to return data from Mercury from 2026, we are in an appropriate interval to review our understanding of the planet's volcanism, which is strongly tied to its global tectonic regime. Where appropriate, we illustrate the interplay between tectonism and volcanism on Mercury with case studies from the planet's Hokusai quadrangle (H05: Fig. 1: Wright et al., 2019), which encapsulates many aspects of the key constituent processes. For ease of comparison with Wright et al. (2019), in which the geological mapping of the quadrangle was presented, here we show several figures of the whole or parts of H05 in the same Lambert Conformal Conic (LCC) projection (central meridian, 45°E; standard parallels, 30°N and 58°N), which we refer to as the "default" LCC projection of H05.  Davies et al., 1978). The Hokusai quadrangle is illustrated with the geological map of Wright et al. (2019). The base map is the 250 m/pixel MESSENGER MDIS mosaic in an orthographic projection, centered on 45°E, 45°N.

Fig. 2.
An unprojected mosaic of MDIS NAC images captured during MESSENGER's third Mercury flyby. Southeast H05, which MESSENGER imaged for the first time. The 305 km-diameter peak ring impact basin named Rachmaninoff occupies the center of this view and is surrounded by its radially textured impact ejecta. Rachmaninoff ejecta superposes relatively uncratered "smooth plains" in the upper-and lower-right of this view, demonstrating that the Rachmaninoff impact postdates emplacement of the local smooth plains. Smooth plains within Rachmaninoff itself must either be contemporaneous with, or younger than, the impact. The spatial resolution of this mosaic approaches~500 m/pixel toward the lower-right corner of this figure. Credit: NASA/Johns Hopkins University Applied Physics Laboratory/Carnegie Institute of Washington.

Background: volcanism on Mercury
Mercury has a prolonged history of both effusive and explosive volcanism (Byrne et al., 2018b), concentrated largely within the first billion years of its planetary history but with a potentially long coda.

Effusive volcanism
Almost the entire crustal surface of the planet was produced volcanically, although thereafter it was intensely battered and fragmented, especially in older regions, by impacts. This volcanism is recorded by "plains" that have sustained different degrees of impact cratering since their emplacement (Fig. 3).
1.2.1.1. Smooth plains. Over 27% of Mercury surface is morphologically classified as "smooth plains" (Figs. 3a,b and 4: Denevi et al., 2013). These plains, chiefly characterized by their relatively smooth surfaces between impact craters, rather than impact craters themselves, are indicative of some form of planetary resurfacing. Plains volcanism, equivalent to lunar mare volcanism, was quickly postulated as the origin of the smooth plains on the basis of Mariner 10 observations (Murray et al., 1974), although an alternative origin as fluidized impact basin ejecta/melt was suggested (Wilhelms, 1976) by analogy with Apollo 16 results for the Cayley Formation on the Moon (Eggleton and Schaber, 1972). Some resurfacing by impact basin ejecta emplacement has undoubtedly occurred on Mercury, but the distributions and volumes of most smooth plains are incompatible with an ejecta origin. Except for the circum-Caloris plains, smooth plains are generally not found around impact basins that could have deposited the plains as ejecta as adjudged by superposition relations.
Photogeological evidence for a volcanic plains origin for most of Mercury's smooth plains includes: embayment relations (Fig. 5a); the presence of completely flooded "ghost craters" (Fig. 5b); and spectral contrasts between smooth plains and their surroundings (Fig. 5c). On the basis of such evidence, >65% of smooth plains on Mercury are confidently judged to be volcanic in origin . However, no unambiguously volcanic features, such as flow units/fronts, source vents, or pāhoehoe/ʻaʻā lava surface textures, have been unequivocally identified within the smooth plains (Byrne et al., 2018b). Smooth plains have probably buried their source vents/fissures (although see Byrne et al. (2013) for one possible preserved example), as is often the case in expanses of basaltic lavas on Earth and other terrestrial bodies (such as large igneous provinces: Byrne et al., 2018b), and finer-scale textures on multi-billion-year-old lava flows will almost certainly have been destroyed by subsequent impacts and the associated development and dispersal of regolith. Mercury's regolith is perhaps 25-40 m thick on the smooth plains (Kreslavsky and Head, 2015), meaning the original lava textures are probably deeply buried or pulverized-factors that improved imaging resolution would not be able to overcome. Consequently, planetary geologists must rely on other observations when interpreting images of Mercury's surface.
Major smooth plains deposits on Mercury (Fig. 4) include: Borealis Planitia (formerly known as Mercury's "northern smooth plains" or the "northern volcanic plains": Head et al., 2011;Ostrach et al., 2015); Caloris Planitia (the smooth plains within the Caloris Basin, the largest well-preserved impact basin on Mercury: Murchie et al., 2008;Fassett et al., 2009), and the circum-Caloris smooth plains that almost entirely surround the basin rim . As an example, Borealis Planitia, the largest contiguous area of smooth plains on Mercury, has an estimated minimum volume of 4 × 10 6 km 3 (Ostrach et al., 2015), which is comparable to the volume of the Siberian Traps large igneous province on Earth (Fedorenko et al., 2000).
These large smooth plains units all have similar CSFDs, which suggests that they formed during a relatively short window Ostrach et al., 2015;Fassett et al., 2009: Table 1). This conclusion is consistent with observations of several superposition relationships of opposite senses between the smooth plains interior and exterior to the Caloris Basin, indicating effectively contemporaneous emplacement . Byrne et al. (2016) measured CSFDs of several smaller (though still expansive) smooth plains deposits, finding that most such distributions are similar to those of the more extensive smooth plains. These authors determined the smooth plains with the lowest N(10) to have a value of 29 ± 21, to which they ascribed an  Denevi et al. (2013) incorporating updates by those original authors (Byrne et al., 2018b). Smooth plains cover~27% of Mercury's surface. Abundant, quasi-circular patches of smooth plains indicate deposits within impact craters in most cases, at least some of which are volcanic. H05 is outlined in red. This figure shows the MESSENGER 250 m/pixel global monochrome mosaic in a Robinson projection centered on Mercury's equator at 90°E. absolute model age of 3.5 Ga (see Byrne et al., 2016 for details and assumptions of impact crater production function and chronology). Younger model ages for smooth plains deposits have been reported elsewhere (Prockter et al., 2010;Fegan et al., 2017), but for much smaller, local deposits confined mostly to impact crater floors, rather than corresponding to regional-scale resurfacing. Therefore, it seems that large-volume effusive volcanism on Mercury ceased~3.5 Ga (Byrne et al., 2016).
1.2.1.2. Intercrater plains. The most spatially extensive geological unit on Mercury is the "intercrater plains" (Trask and Guest, 1975), which covers >70% of the planet's surface . Intercrater plains, originally described as "level to gently rolling ground between and around large craters and basins [with a] high density of superposed small craters in the size range 5-10 km" (Trask and Guest, 1975) are everywhere embayed by smooth plains and are more heavily cratered, although the amount of cratering varies substantially across Mercury. Superposition, as well as areal crater density, demonstrates that the intercrater plains are older than the smooth plains; most intercrater plains were emplaced between~4.1 and~3.8 Ga (Marchi et al., 2013).
Several lines of evidence support a volcanic origin for most of the intercrater plains. The widespread occurrence of intercrater plains across the planet (most of the uncolored portions of Fig. 4 are intercrater plains), and the lack of large impact basins that could have supplied the necessary volume of ejecta material to produce such plains, argue for a large igneous province-style plains emplacement. Furthermore, it has been shown that smooth plains can be modified by overlapping fields of secondary craters from two or more relatively young primary impacts into a morphological texture indistinguishable from intercrater plains (Whitten et al., 2014). This observation supports the conclusion that intercrater plains were originally emplaced similarly to smooth plains, but at an earlier stage in Mercury's history such that these older plains have become more heavily cratered.
As further corroboration of this view, CSFDs for the most densely cratered smooth plains overlap within error with CSFDs for the least densely cratered intercrater plains (Whitten et al., 2014;Byrne et al., 2016). Furthermore, that no surface appears to have an absolute model age greater than~4.1 Ga suggests that there was intense, continual resurfacing early in Mercury's history (Marchi et al., 2013) driven by the more energetic planetary and space environment in the early Solar System. For example, young Mercury's volcanism would have been enhanced while the planet's interior was hotter, and more intense impact bombardment during this time might have facilitated or even induced additional volcanism (Ivanov and Melosh, 2003;Elkins-Tanton and Hager, 2005).
XRS data have been used to create major element ratio maps of Mg/ Si, Al/Si, S/Si, Ca/Si, and Fe/Si (Weider et al., 2015;Nittler et al., 2020). These maps provide evidence for "geochemical terranes" on Mercury, which in some instances spatially coincide with morphological units (Weider et al., 2015). GRNS-derived maps of Na and K abundances   Denevi et al. (2013) 92 ± 16~3.7-3.9 f also broadly support the concept of geochemical terranes (Peplowski et al., , 2014. In general, smooth plains have lower Mg/Si and higher Al/Si than more heavily cratered terrains . However, there is geochemical variation within the smooth plains, consistent with compositionally heterogeneous mantle sources. For example, southern Borealis Planitia in H05 (see Fig. 9) has an intermediate Mg/Si ratio (0.3-0.6), whereas most of the rest of Borealis Planitia has a lower such ratio (<0.3: Weider et al., 2015). Poleward of 80°N, Borealis Planitia has the highest K abundance (~2000 ppm: Peplowski et al., 2012) and Na content (~5 wt%: Peplowski et al., 2014) on the planet, whereas alkali content of the intercrater plains is generally lower. These observations support an evolution from more ultramafic lavas forming the intercrater plains in Mercury's early eruptive history toward feldspar-dominated lavas by the time of smooth plains emplacement (Namur and Charlier, 2017).
The higher spatial resolution of XRS data over Borealis Planitia allowed for a bulk composition for these plains to be determined with a lowered prospect that the composition would be contaminated by XRS pixels crossing into adjacent terrains (Vander Kaaden and McCubbin, 2016  , and allowed those melts to crystalise under a variety of Mercury environmental conditions (oxygen fugacity =~2-7 log 10 units below the iron-wüstite buffer, pressure = 0.57-5 GPa, temperature = 1100-2000°C: see Vander Kaaden and McCubbin (2016) for full details of their experiments). These authors' experiments show that Borealis Planitia lavas should contain Na-rich plagioclase, Mg-rich olivine, and low-Ca, low-Fe pyroxene with sulfides of Fe, Cr, Mn, Ti, Mg, and Ca. Thus, Vander Kaaden et al. (2017) determine that the smooth plains on Mercury have an alkali-and sulfide-rich boninite-or komatiite-like composition.
Such lavas have high eruption temperatures, with rheologies similar to Hawai'ian basalts (Sehlke and Whittington, 2015;Vander Kaaden et al., 2017), which is in turn consistent with observations of lavaeroded channels and kipukas on Mercury . Such Mercurian lavas probably did not have ultra-low viscosities, however.
1.2.1.4. Constructs. Unlike the other silicate volcanic bodies in the Solar System (Earth, the Moon, Venus, Mars, and Io), there is little morphological evidence of constructional volcanic edifices on Mercury (Byrne, 2020). For example, Mercury has no giant volcanoes equivalent to the Tharsis Montes of Mars, Hawai'ian volcanoes on Earth, nor the myriad large shields that dot the surface of Venus.
Long-wavelength topographic undulations of Mercury's surface observed in MLA data , such as Borealis Planitia's "northern rise" (Fig. 6), which impinges on H05, appear not to be large shields. There is no summit complex and the floors of some impact craters on the flanks of this rise are tilted away from the summit, which indicates at least some of the rise's topography was caused by updoming , rather than shield-building. A lack of compositional or temporal distinction between the northern rise and the surrounding smooth plains is also difficult to reconcile with a shield volcanism origin. However, a positive free-air gravity anomaly centered on the rise could be consistent with uplift caused by crustal underplating (James, 2018).
Notably, there is little evidence on Mercury for even small volcanic constructs, which are otherwise abundant throughout the Solar System. A feature in Caloris Planitia was suggested to be a candidate shield volcano on the basis of concentric tones apparent on fly-by images, interpreted as being the result of a subtle topographic rise surrounding an irregular depression, thus regarded as a vent . However, orbital topographic data show the rise to have flank slopes of only~0.14° (Rothery et al., 2014), although the central depression is widely accepted as an explosive volcanic vent (which also explains the concentric tonal pattern: see Section 1.2.2). One global survey for small constructs found two candidate volcanoes, but could not be conclusive about their origin, volcanic or otherwise, based on the available MESSENGER image resolution . In any case, small volcanic edifices formed by effusive eruptions are rare if not entirely absent on Mercury, suggesting that some aspect of the planet's typical eruptive mode prohibited their formation (e.g., high effusion rates), that evidence for their existence has been efficiently erased by impacts, or both . Ultra-low-viscosity lavas, which might be expected to build constructs less efficiently, appear to be inconsistent with analog experimental (Sehlke and Whittington, 2015;Vander Kaaden and McCubbin, 2016) and petrological models (Vander Kaaden et al., 2017) of Mercury composition lavas.

Explosive volcanism
Being the innermost planet, Mercury was predicted to be depleted in volatile elements compared with the other rocky planets, because solar heating is expected to have driven volatiles out of that part of the protoplanetary disk (Wetherill, 1994). One of the most surprising results from the MESSENGER mission was that Mercury's surface is rich in many moderately volatile elements (Ebel and Stewart, 2018;Nittler et al., 2018). Consistent with this finding is the discovery of bright, diffuse spots that appear to be surficial materials draping topography and grading into their surroundings ( Fig. 7a: Blewett et al., 2009). Most such deposits encompass one or several irregular pits near their mid-points, such as the example mentioned in the previous section. Such pits cannot be impact craters because of their non-circularity and lack of raised rims (Rothery et al., 2014).
Instead, these depressions are interpreted as explosively excavated volcanic craters, with their surrounding material having been deposited as ballistically emplaced pyroclasts (Kerber et al., 2009;Rothery et al., 2014;Jozwiak et al., 2018). In addition to their diffuse, gradational contacts with their surroundings (Fig. 7a,c), these deposits are characterized by high spectral reflectance and relatively steep slopes in plots of reflectance versus wavelength, making them relatively red compared with average terrains on Mercury. Originally, these deposits were informally numbered as "red spots" ), but are now known as "faculae" (sing. "facula"), the International Astronomical Union-approved descriptor term for "bright spot". The formally named examples have an individual name accompanied by the descriptor term facula/faculae.
On Earth, explosive volcanism is driven by the exsolution of volatile species from ascending magma (Cashman, 2004), and so surface observations of explosive volcanism on Mercury strongly suggest a source of volatiles within the planet's subsurface. Nearly 200 sites of putative explosive volcanism have been cataloged on Mercury Goudge et al., 2014;Thomas et al., 2014a;Jozwiak et al., 2018). Some examples have been studied in detail. For example, Agwo Facula (formerly "red spot 3"; Blewett et al., 2009;Kerber et al., 2011) is a 60 km-diameter deposit in southwest Caloris Planitia (which corresponds to the now-discounted constructional feature described in Section 1.2.1.4: Fig. 7a). The deposit is centered on an irregular pit interpreted as an explosively excavated volcanic crater  that is >1 km deep (Rothery et al., 2014). Internal crosscutting relationships of septa within the pit indicate that the pit constitutes at least nine overlapping volcanic vents, making the example within Agwo Facula a compound vent (Rothery et al., 2014). The shapes and mutual relationships between vents associated with each pit have no resemblance to the expected morphological results of subsidence along caldera-bounding ring faults, but are consistent with explosive excavation, likely modified by subsequent collapse into evacuated conduits. Recently, Pegg et al. (2021) showed that 70% of explosive volcanic vents on Mercury are compound, so that multiple eruptions at each vent site are the norm-at least in those examples preserved on the surface.
Another important example is Nathair Facula (previously referred to informally as NE Rachmaninoff: e.g., Kerber et al., 2011;Weider et al., 2016), which is the largest putative pyroclastic deposit on Mercury (~140 km in radius; Fig. 7c). Indeed, Nathair Facula hosts the planet's largest vent (which is 31 km across). Kerber et al. (2011) determined that a substantial enrichment in magmatic volatiles would be required for erupted material to achieve the minimum ejection velocity to transport pyroclasts to the observed edge of the deposit. MESSENGER XRS data show a depletion of S in Nathair Facula compared with its surroundings (Weider et al., 2016). Similarly, MESSENGER NS data suggest that C is 1-2 wt% less abundant in Nathair Facula than in its surroundings .
Under the strongly reducing conditions of Mercury's interior (McCubbin et al., 2012), S should be soluble in magma. Weider et al. (2016) suggested that S and C depletions at Nathair Facula were caused by the oxidation of magma-soluble S and C by oxides, such as SiO 2 and Facula's pit, it appears that the irregular outline of Nathair Facula's pit could not have been created by a single eruption (Rothery et al., 2020b;Pegg et al., 2021). Its variable floor texture freshness gives some indication of the migration of the eruption center (from southwest to northeast), but there are no cross-cutting septa to corroborate this hypothesis. This panel shows NAC images EN0224508427M, EN1003815055M, EN1003843856M, and EN1993843866M (~26 m/pixel) in the default LCC projection of H05.
FeO, to form volatile species such as SO 2 and CO. These S-and C-bearing oxides might have volatilized in the magma, driven the explosive eruptions that formed Nathair Facula and subsequently been lost to space, thus leaving a reduced spectral signature for S and C in the Nathair Facula deposit. Furthermore, the loss of C that was previously in the form of graphite, held to be an important darkening agent stable at Mercury's surface , could help to explain the relatively high spectral reflectance of Nathair Facula and other putative pyroclastic deposits (Weider et al., 2016).
As with effusive volcanism, there is no evidence for pyroclastic constructs beyond subtle slopes associated with some faculae, which, as surficial deposits, are generally too thin to hide pre-existing impact craters (Rothery et al., 2014;Thomas et al., 2014b). Numerical simulations under Mercury surface conditions (i.e., a vacuum, and surface gravity of 3.7 ms −2 ) suggest that pyroclastic cones, such as those found on Earth (e.g., Riedel et al., 2003), and interpreted to have formed on the Moon (e.g. Lawrence et al., 2013) and Mars (e.g., Brož and Hauber, 2012), cannot form to dimensions that are observable with present data (Brož et al., 2018).
Faculae superpose the smooth plains of Caloris Planitia Rothery et al., 2014), indicating that explosive volcanism locally outlasted effusive volcanism. There have been no observations of smooth plains embaying a facula anywhere on Mercury. This observation suggests that explosive volcanism generally outlasted effusive volcanism on Mercury, with the possible exception of highly localized, post-impact, late-stage effusive volcanism (Prockter et al., 2010;Chapman et al., 2012). Although large-volume effusive volcanism appears to have ceased~3.5 Ga (Byrne et al., 2016), candidate pyroclastic vents have been observed within craters that retain bright albedo rays (Thomas et al., 2014a;Jozwiak et al., 2018). Ray-bearing craters have a size-frequency distribution consistent with them having formed between 280 ± 60 Ma and the present day (Banks et al., 2017), which means explosive volcanism potential persisted into Mercury's most recent geological history. Of note, there is no geological evidence preserved on Mercury to indicate either way whether explosive eruptions occurred during the phase of major effusive activity. Landform assemblages associated with explosive volcanism on Mercury appear to be better preserved than those expected of effusive volcanism (e.g. source vents/fissures, flow fields) because explosive volcanism persisted be-yond~3.5 Ga, when burial of pits and faculae by plains became unlikely and impact erosion was less efficient.

Background: global contraction on Mercury
The dominant form of tectonic deformation on Mercury is crustal shortening. The most prominent expression of that deformation is the planet's population of "lobate scarps" (Fig. 8). Although interpreted as the surface manifestations of thrust faults based on Mariner 10 observations (Murray et al., 1974), it was not until MESSENGER that their global ubiquity was revealed (Byrne et al., 2014). By analogy with Earth, these structures are displacement gradient folds, and likely fault-propagation folds, such that the landforms termed "lobate scarps" (which is not a geological term: Klimczak et al., 2019) are hanging-wall anticlines atop blind or surface-breaking thrusts.
Extensional tectonics are not nearly so prevalent on Mercury, and are overwhelmingly restricted to volcanically infilled impact features (Byrne et al., 2018a). Patterns in thrust orientation and spatial density may be controlled by some combination of tidal despinning (Melosh and Dzurisin, 1978;Dombard and Hauck, 2008), crustal thickness variations (Galluzzi et al., 2019), and thermal differences (Beuthe, 2010), but the overall horizontally compressive stress regime that brought about global thrust faulting was probably caused in the main by secular cooling of the planet's interior and the consequent global contraction (Byrne et al., 2016). Secular cooling and global contraction have been predicted to have a strong influence on the volcanic evolution of single-plate planets (Solomon, 1978).
Several studies have investigated the onset, duration, and cessation of lobate scarp activity. Banks et al. (2015) examined the superposition relationships between impact craters in various states of degradation along the lengths of several lobate scarps. Crater degradation was used as a proxy for crater age (Kinczyk et al., 2020), and craters with certain characteristics of degradation (e.g. the presence or absence of bright crater rays, preserved terraces, and central peak, etc.) have been used to update absolute model ages for the time-stratigraphic systems on Mercury (Banks et al., 2017). A lobate scarp segment that cuts a crater must have undergone at least some of its total fault movement after the formation of that crater (Banks et al., 2015). In such a case, it is difficult to know how much fault displacement, if any, occurred before the impact crater formed or how much time passed between crater formation and the latest increment of slip occurring. Therefore, the age of the cross-cut impact crater (inferred from its degradation classification) qualitatively offers a maximum age for the last slip event of the crosscutting segment of a lobate scarp.
Conversely, a lobate scarp segment that is superposed by an impact crater indicates that resolvable movement on the superposed scarp segment ceased before that crater formed (Banks et al., 2015). In this case, it is difficult to know how much time passed between the cessation of faulting and the formation of the impact crater. Therefore, the age of the superposing impact crater qualitatively provides a minimum age for the local cessation of fault activity (Banks et al., 2015). Based on their observations of impact crater-lobate scarp superposition relationships, Banks et al. (2015) inferred that the surface manifestation of global contraction had initiated at the time of the Calorian system (~3.9 Ga). Some lobate scarp segments superpose Mansurian and Kuiperian craters, indicating that at least some faulting locally continued through these periods (Banks et al., 2015).
Buffered crater statistics have been used to estimate absolute model ages for several of the largest lobate scarp systems on Mercury (Giacomini et al., 2015(Giacomini et al., , 2020. These studies estimate that activity on the largest lobate scarps ceased at around~3.8-3.6 Ga. Very small, fresh lobate scarps suggest that new scarp formation continued at least until Mercury's recent geological past, and might be ongoing today (Watters et al., 2016), although there so far has been no substantive assessment of how these smaller, fresh scarps are kinematically linked to the older, larger scarps across Mercury.

Influence of Mercury's tectonics on timing of effusive volcanism
The most important influence of Mercury's tectonic regime on its volcanism is the apparent cessation of large-volume effusions~3.5 Ga. Secular cooling of the planet's interior led to global contraction, which induced a global horizontally compressive stress regime in the lithosphere (Solomon, 1978;Byrne et al., 2016), evidenced by a global distribution of thrust fault-related landforms superposing volcanic plains (Byrne et al., 2014). Horizontally compressive stresses in the lithosphere inhibit magma ascent when they exceed the ability of magma to keep a conduit open (Byrne, 2020). This condition favors lateral magma transport within subsurface sills, rather than dyke propagation to the surface (Menand et al., 2010). The cessation of large-volume effusive volcanism in response to global contraction, which seems to have occurred on the Moon as well (Solomon and Head, 1980), appears to have been geologically abrupt on Mercury, as shown by the clustering of smooth plains absolute model ages 3.5-3.7 Ga (Byrne et al., 2016).
Here, we report for the first time new CSFDs for that portion of Borealis Planitia situated within H05, which corroborate this trend. We acquired crater statistics for our mapped original extent of the Borealis Planitia smooth plains within H05 (Fig. 9). We counted all craters >5 km in diameter that superpose the plains (e.g. crater ejecta/rim unembayed by the plains: Wright et al., 2019), but we base our interpretations mainly on craters ≥10 km in diameter to avoid including secondary impact craters. Primary impact craters form when a space object impacts a planetary body at hypervelocity. Material ejected from primary craters can in turn impact the planetary surface and form secondary craters, which can be as large as~10 km in diameter on Mercury, because of the high average primary impact velocity Byrne et al., 2016). If such secondary craters were included in the crater count, they would alter the crater size-frequency distribution of the surface being counted, which could return an incorrectly old age, or the distribution might not fit on an isochron at all. We also divided Borealis Planitia within H05 into three spectrally distinct subregions-north (relatively spectrally red), southwest (relatively blue), and southeast (relatively red)-to test if these spectral reflectance differences correspond to different CSFDs that might indicate lavas of varying ages and/or compositions. Our calculated N(10) values for Borealis Planitia within H05 (Table 2) agree well with those determined by Ostrach et al. (2015) for Borealis Planitia north of 50°N (Table 1).
We plotted cumulative frequency plots for our crater counts using CraterStats2 (Fig. 10). We derived all absolute model ages using the production and chronology functions (for porous scaling) of Le Feuvre and Wieczorek (2011). We have plotted isochrons for absolute model ages with two decimal places to highlight different crater size-frequency populations-but consider only the first two figures of these ages to be significant. Broadly, these plots show that the uppermost smooth plains of Borealis Planitia within H05 have absolute model ages of~3.7-3.8 Ga, consistent with previous measurements (Ostrach et al., 2015). The overall Borealis Planitia cumulative frequency curve does not plot on a single isochron, indicative of more than one crater population preserved here.
Specifically, craters >40 km in diameter plot approximately on the 3.8 Ga isochron, whereas craters below this diameter threshold plot beneath this isochron. This finding means that the smooth plains within H05 either have a surplus of craters >40 km in diameter or are deficient in craters <40 km in diameter. A surplus of craters >40 km in diameter could be explained by the inclusion in our counts of large craters that were buried by (and thus predate) the smooth plains. This possibility is not likely, since Wright et al. (2019) mapped the ejecta of craters >20 km in diameter, and such ejecta is resolvably embayed by smooth plains in many cases; thus, it is straightforward to identify those large craters superposed by and thus older than these smooth plains.
A deficit of craters <40 km in diameter could be explained by a partial resurfacing event within Borealis Planitia, which preferentially hid such craters. Two candidate partial resurfacing mechanisms applicable to H05 include by lavas of limited spatial extent, or by ejecta from large craters superposing the smooth plains. Multiple resurfacing events within Borealis Planitia are already evidenced by its closely collocated populations of large and small ghost craters (Ostrach et al., 2015). Young, large craters such as Rustaveli, Sousa, and the Hokusai crater  (2011) with porous target scaling.  (Michael and Neukum, 2010). Isochrons (gray curves), which represent the shapes of ideal CSFDs of the labelled absolute age, are plotted for reference. All absolute model ages were calculated using the chronology and production functions of Le Feuvre and Wieczorek (2011) with porous target material scaling to reflect Mercury's fractured megaregolith. Craters <5 km in diameter were not used for fitting isochrons because our counts are not complete below this diameter threshold. The CSFDs shown in panels (a-c) do not lie along a single isochron (red and blue components), perhaps indicating a resurfacing event(s) that preferentially removed craters below a certain diameter (see text). Points on the spatial randomness plots show on the y-axis the number of standard deviations by which the mean second-closest neighbour distance (M2CND) of 3000 simulated impact crater populations deviates from the M2CND of the measured impact crater population at a given crater diameter (x-axis). In all cases, the M2CNDs plot within the gray bands, indicating that these values suggest the crater populations are spatially random, rather than ordered or clustered formations indicative of secondary cratering (Michael et al., 2012). itself, which have extensive ejecta blankets draped atop Borealis Planitia, conceivably buried many craters superposing the smooth plains such that we could not recognise and so include them in our counts.
The cumulative frequency curves for northern and southwestern H05 also plot on two isochrons corresponding to~3.7-3.8 Ga, which suggests that these regions contain the partial resurfacing event(s) that manifest in the CSFD of the whole of Borealis Planitia in H05. Conversely, the southeastern H05 count area cumulative frequency curve plots on a single isochron of~3.7 Ga, suggesting that this is a chronostratigraphically homogeneous region at the scale of our crater count (craters >5 km in diameter). This result suggests that the best estimate absolute model age for the smooth plains of Borealis Planitia in H05 is~3.7 Ga, a finding consistent with that of the entire Borealis smooth plains unit (Ostrach et al., 2015).
There are smooth plains in H05 that are stratigraphically younger than Mercury's major regions of smooth plains, but these are relatively minor volumes and they are generally restricted to impact crater interiors (Wright et al., 2016;Byrne et al., 2018b). Impact cratering results in several temporary and permanent effects on a planetary surface that are conducive to the extraction of melt from within the planet. The temporary effects include the deposition of heat (Davison et al., 2012) and, in the case of Mercury, the resetting of the global horizontally compressive stress regime at least within the crater (Byrne, 2020). More long-lived consequences of impacts pertinent to melt extraction include the opening of fractures as favourable magma ascent pathways (Klimczak, 2015), and the removal of overburden (i.e. crust and perhaps mantle: Byrne, 2020). It has been suggested that impacts can induce mantle decompression melting directly on Earth (Elkins-Tanton and Hager, 2005), but this mechanism seems unlikely on Mercury given its smaller size (i.e. its interior is cooler than Earth's at any given time) and the rarity of impact basins of the size required to generate sufficient decompression (>300 km-diameter) younger than 3.5 Ga (Ivanov and Melosh, 2003).
It would be useful to know for what period of time impact craters facilitated late-stage effusive volcanism after~3.5 Ga. The small areal extents of smooth plains that are stratigraphically younger than their more extensive counterparts are generally insufficiently large for statistically robust CSFDs, making estimating their absolute model ages difficult. The best example of stratigraphically young smooth plains is within Rachmaninoff crater (Figs. 2 and 5c), which superposes, and thus postdates, Borealis Planitia (~3.7 Ga). Based on its relatively fresh geomorphology, Rachmaninoff probably formed around the transition between Mercury's Calorian and Mansurian systems (Banks et al., 2017;Wright et al., 2019;Kinczyk et al., 2020). Smooth plains within Rachmaninoff that bury parts of the peak ring suggest a depth (and hence volume) too great to be impact melt, and, furthermore, these plains are of a color supportive of their being volcanic, rather than impact melt (Prockter et al., 2010;Chapman et al., 2012). Depending on the timing of the formation of the Rachmaninoff basin, this suggests that impact craters could facilitate late-stage effusions on Mercury for at least a few hundred million years following the formation of the bulk of Borealis Planitia, and, by extension, the other lava units on the planet.

Influence of tectonism on Mercury's expression of volcanism
As global contraction caused large-volume effusive volcanism to wane, the decrease in eruptive volumes compared with the larger regions of smooth plains led to different morphological expressions of plains volcanism on Mercury. As described earlier, Mercury has two plains units of global import: smooth plains and intercrater plains Whitten et al., 2014). However, several quadrangle geological mappers (Galluzzi et al., 2016;Mancinelli et al., 2016;Guzzetta et al., 2017;Wright et al., 2019) have found photogeological textures that are notable at the quadrangle scale but that are not adequately described by the unit definitions of either smooth plains or intercrater plains, which are terms developed first from hemisphereand then from global-scale mapping.
For example, Wright et al. (2019) dubbed a surface texture in H05 as "intermediate plains," describing the unit as "plains with a roughness intermediate between intercrater and smooth plains" and "hummocky terrain composed of degraded crater rims with intervening low-lying regions with smooth, level surfaces". Crucially, Wright et al. (2019) found that "smooth patches constitute~50% of intermediate plains", meaning by definition that these units cannot be classified as intercrater plains (Whitten et al., 2014). However, at the quadrangle map scale, these intervening smooth patches were generally too finely interspersed with upstanding crater rims to be mapped as smooth plains and meaningfully displayed at the intended publication scale, which led Wright et al. (2019) to classify the intervening rough and smooth textures into "intermediate plains". Wright et al. (2019) interpreted these "intermediate plains" as intercrater plains that have been partially inundated by smooth plains materials. This interpretation is supported by the presence of putative volcanic vents within regions mapped as intermediate plains in H05, indicating that a magma source was present beneath these regions. The same interpretation was made by Byrne et al. (2013) for what those authors termed "graded terrain" adjacent to several valleys filled with smooth plains northwest of the Caloris basin. This graded terrain has a muted topographic expression that transitions to the characteristic rugged texture of the intercrater plains with increasing distance from the valleys filled with smooth plains, interpreted as lava-modified channels . These authors interpreted their graded terrain as evidence that at least some volume of erupted lavas within the region were not channelized but instead overbanked the channels to spread out over the surroundings. The "grading" of the muted intercrater plains into normal intercrater plains with increasing distance from the broad channels suggests that, as the lava flux in the region dropped, overland lava flow retreated until it became channel-confined once more.
It is possible that the intermediate plains in H05 could have formed in a similar manner: H05's intermediate plains are in contact with the smooth plains of Borealis Planitia, so the smooth patches within the intermediate plains could represent overland effusions that retreated before they were able to completely resurface the underlying intercrater plains. This hypothesis implies that the smooth patches within the intermediate plains are contemporaneous with or older than at least some of the smooth plains of Borealis Planitia. Alternatively, the smooth patches within the intermediate plains could represent terminal smooth plains volcanism, which could make them stratigraphically younger than the smooth plains of Borealis Planitia.
It is difficult to tell whether the effusions that formed the intermediate plains in H05 occurred in either the early or late stages of smooth plains emplacement in Borealis Planitia. If the interpretation of the intermediate plains as intercrater plains with a thin infilling of superposing lavas is correct, then age dating by measuring the CSFDs of the intermediate plains would be equivocal, since it would be difficult in many cases to determine if a crater superposes the intermediate plains or is a flooded/buried crater protruding from the underlying intercrater plains. Furthermore, the count areas for the smooth patches in the intermediate plains would be too small, and thus the number of included craters too low, to yield robust crater statistics. MESSENGER data are of insufficient resolution to determine this stratigraphic relationship unequivocally, and so this issue cannot be resolved until data from BepiColombo are returned.

Impact craters and basins
Impacts remove overburden, propagate fractures, deposit heat, and temporarily reset the local stress regime: all of these facilitate the extraction of magma to the surface (Byrne, 2020), thus impacts can modulate the strong influence of Mercury's global horizontally compressive stress regime. Whatever the link between impact-generated heat and partial melting, it is clear that smooth plains, most of which are volcanic, commonly occupy impact structures (e.g., Head et al., 2008: Fig. 4). In H05, the Rachmaninoff and Rustaveli craters are instructive examples (Fig. 11). Ejecta from both these craters superposes smooth plains, which means that both craters probably postdate the cessation of large-volume effusive volcanism in this region.
The color contrast between the spectrally "red" interior Rachmaninoff smooth plains and the low-reflectance, blue material excavated in the crater's peak ring, rim, and intervening floor (Fig. 11a) suggests that there is a compositional difference between these regions. Impact melt generally does not have a spectral contrast with other crater materials, as the former is essentially a melted and resolidified form of the latter. No differentiation of melt sheets on Mercury by fractional crystallization has been observed. The smooth plains within Rachmaninoff embay the peak ring, rather than drape it, indicative of effusive emplacement by relatively low-viscosity lavas. Crater size-frequency studies of the Rachmaninoff interior plains suggest that these plains have a younger crater retention age than the crater's other materials. This finding strongly suggests that the interior plains were emplaced following an interval of geological time after the impact, which also argues against an impact melt origin for the plains (Prockter et al., 2010;Chapman et al., 2012).
"Pitted ground", present in the form of Suge Facula in southeast Rachmaninoff and other minor faculae south of the Rachmaninoff peak ring, has been suggested to have formed when lavas flowed over volatilebearing substrates on Mercury (Thomas et al., 2014c). The pitted ground surface texture consists of clustered and coalesced pits tens of meters deep with uneven floors and muted rims (Thomas et al., 2014c). Pitted ground is generally surround by a reddish halo, similar to a facula associated with a large central pit, but the multitudinous pits of pitted ground are more consistent with near-surface volatilization, similar to tuff ring formation on Earth (Lorenz, 1986), rather than deep-seated volcanic eruptions. North of Suge Facula, the crater floor between the peak ring and the crater rim is spectrally low in reflectance and is blue, which implies that no smooth plains volcanic materials have reached there. We interpret Suge Facula and the other faculae in Rachmaninoff with the pitted ground texture as the most distal reaches of centrally sourced lava flows in the crater. Under this interpretation, the thickness of the lava diminishes toward the eastern extent of Suge Facula.
For a model in which pitted ground formed when lava flowed over a volatile-bearing substrate, the lava must have been of some specific thickness such that explosive loss of volatiles in the subsurface created pits at the surface. Furthermore, for the pitted ground texture to be present today, it must not have been buried too deeply by subsequent lava flows and/or obliterated by impact gardening. Therefore, the eastern spatial extent of Suge Facula was controlled either by lava flow length or by the presence of sufficient volatiles in the substrate (i.e., pitted ground will have formed only where flowing lava and volatile-bearing substrates coincided). Given that no spectrally "red" smooth plains are identified beyond Suge Facula, it appears that the extent and thickness of the lava controls the extent of Suge Facula, and that the eastern margin of Suge Facula is the approximate location of the flow terminus.
Although the western extent of Suge Facula might simply correspond to the boundary of the volatile-rich substrate, it could alternatively have been controlled by the thickness of the interior lava. Since the lava thickness presumably increases toward the center of Rachmaninoff, flow thickness will be greater in the west of Suge Facula than in the east. The location of the western margin of Suge Facula might therefore reflect a threshold thickness of lava above which subsurface volatile loss did not manifest at the surface. Alternatively, as the supply of lava waned following the creation of an originally more extensive region of pitted ground, subsequent lava flows emanating sources within the Rachmaninoff peak ring might not have reached as far as the proposed flow terminus at the eastern margin of Suge Facula, but were of sufficient thickness to bury any pitted ground texture that developed west of Suge Facula. In any case, the occurrence of pitted ground within impact craters, rather than in Mercury's major regions of smooth plains suggests that relatively small-volume effusive eruptions, such as those that took place in craters postdating the cessation of large-volume plains volcanism on Mercury, are required for pitted ground texture to be preserved.
Rustaveli's peak ring (Fig. 11b) is almost entirely surrounded and embayed by smooth plains, with only the tops of presumably the largest  Fig. 5c). Smooth, high-reflectance spectrally "red" plains, which exhibit concentric grabens (two larger examples indicated by white arrows), are situated within Rachmaninoff's peak ring. These plains have breached the peak ring in the south. The spectrally bright spot in the lower-right of the image is Suge Facula, an is an example of Mercury's pitted ground (Thomas et al., 2014a). More minor pitted ground faculae occur at the terminus of the high-reflectance red plains at the bottom of the image (see labelling in Fig. 12). (b) Rustaveli basin in~665 m/pixel enhanced color. The smooth plains on Rustaveli's floor almost completely bury the crater's peak ring. Pitted ground faculae are located around the peak ring elements protruding through the later smooth plains. Both panels show the default LCC projection of H05. mountain peaks protruding from the plains. Burial of the peak ring of a crater of this size cannot be achieved by the anticipated thickness of impact melt alone (Grieve and Cintala, 1992;Cintala and Grieve, 1998), supporting the view that this basin, too, hosts post-impact volcanic smooth plains.
Several studies of Mercury's record of explosive volcanism show that vents and faculae are commonly co-located with impact craters Goudge et al., 2014;Thomas et al., 2014b;Jozwiak et al., 2018). Indeed, it appears that impact cratering has a statistically robust spatial control on the occurrence of explosive volcanism on Mercury (Klimczak et al., 2018). Furthermore, most (>70%) of putative explosive volcanic vents on Mercury are compound, meaning they were formed by multiple eruptions occurring at essentially the same place (Pegg et al., 2021). This observation supports the view that structurally controlled magma ascent pathways, once established, were continuously utilized by ascending magma because the stress required to keep those conduits open would have been less than that need to form new fractures in the crust-especially under a horizontally compressive tectonic stress state. Therefore, the fractures propagated by impact cratering likely played a key role in enabling the ascent and eruption of magmas on Mercury after the onset of global contraction (Byrne, 2020).
Several vents and faculae are located within the periphery of Caloris Planitia, but none is found more than 100 km inside it Rothery et al., 2014;Jozwiak et al., 2018). This arrangement resembles the occurrence of post-mare-volcanism on the Moon around the edges of the lunar mare basins (McGovern and Litherland, 2011). It is possible that the quasi-circular lithospheric loads represented by the lunar mare cause lithospheric flexure such that an annulus surrounding the load undergoes extension: such a stress state is favourable for magma ascent around this periphery and so could be responsible for localising smallvolume, late-stage lunar volcanism around the edges of lunar mare basins (McGovern and Litherland, 2011). This same process might explain the occurrence of several volcanic vents near the edge of Caloris Planitia (Byrne et al., 2018a). In addition, it might explain the occurrence of Nathair Facula, Neidr Facula, and other smaller putative pyroclastic deposits in H05, along the southern margin of Borealis Planitia (Wright, 2019).

Lobate scarps
Some studies have postulated that faults might act as pre-existing weaknesses that ascending magmas could exploit to reach the surface Thomas et al., 2015). There are some examples of small patches of smooth plains embaying lobate scarps (Fig. 8:  Malliband et al., 2018). These ponded patches could correspond to localized, late-stage effusions, presumably erupted via fractures kinematically associated with the scarp-producing thrusts, but in some cases an ejecta origin for these plains cannot be ruled out. Some workers observed that explosive vents and lobate scarps are closely spatially collocated (Thomas et al., 2014b), although others argued that they are generally not (Jozwiak et al., 2018). Klimczak et al. (2018) statistically tested the spatial association between lobate scarps and pyroclastic vents. These authors found that shortening structures on Mercury are so widespread that, although there is an apparent spatial association between shortening structures and vents, it is statistically equivocal given the relatively high probability that this co-location could have arisen by chance alone (Klimczak et al., 2018).

Influence of volcanism on tectonism
Mercury's tectonic expression also appears to have been influenced by its volcanic characteristics in some cases. For example, major fault systems partially border a region with relatively high Mg abundances in the Victoria quadrangle, to the west of H05 (Galluzzi et al., 2019). Morphologically, this "high-Mg region" constitutes intercrater plains (Galluzzi et al., 2016), and thus the surface rocks are probably the remains of lavas erupted early in Mercury's history (Whitten et al., 2014). High-Mg magmas form by high degrees of partial melting of undepleted mantle (Charlier et al., 2013). Therefore, it is possible that these high-Mg lavas erupted due to high degrees of partial melting extracted from a chemically heterogeneous mantle . The interior of the high-Mg region has a far lower spatial density of faulting than its edges, and thus it is possible that this region constitutes a mechanically strong crustal block that focused crustal shortening along its edges (Galluzzi et al., 2019).
Mercury's volcanic smooth plains probably represent a more competent surface unit than the intercrater plains, and this strength contrast has effected some control on the expression of Mercury's tectonics (Byrne et al., 2014). For example, as stated previously, there is relatively spatially restricted extensional tectonism on Mercury (Byrne et al., 2018a). However, networks of grabens are found within some volcanic plains inside impact craters, such as Rachmaninoff (Fig. 11a: Blair et al., 2013), or ghost craters such as Copland . Here, the impact crater may have "shielded" the internal volcanic plains from Mercury's prevailing stress regime, possibly through strain partitioning into the perimeter of the impact feature itself. This partitioning might have permitted the interior plains to contract thermally causing "thinskinned" extensional faulting within such volcanic units of sufficient thickness Blair et al., 2013). Some volcanic fills in larger basins have lobate scarps concentrated at their edges, which perhaps formed when the interface between the strong volcanic infill and the basin floor acted as a décollement that restricted shortening strain to the edge of the volcanic fill (Fegan et al., 2017). Mercury. An absolute model timescale is shown at the left. The shapes on the right are schematic representations of the relative intensity and duration of key geological processes on Mercury discussed in this paper. Absolute model ages of relevant geological events are shown at the left. The horizontal line separating the labels "sp" (smooth plains) and "icp" (intercrater plains) indicates the approximate time when plains volcanism changed from being preserved as intercrater plains to smooth plains. Question marks indicate uncertainties about the timings and durations of geological processes.

Summary
Mercury's planetary evolution is summarised in Fig. 13. Impact cratering was a presumably a dominant process in Mercury's early history, and, though greatly diminished, continues into the present. The oldest surfaces on Mercury formed~4.1 Ga, probably by a combination of effusive volcanism and basin formation (Marchi et al., 2013). Effusive volcanism possibly occurred before then, but no record of that time remains. Continual effusive volcanism (4.1-3.75 Ga) is recorded by intercrater plains (Whitten et al., 2014), and subsequently large-volume smooth plains (3.75-3.5 Ga: Head et al., 2011;Denevi et al., 2013;Ostrach et al., 2015;Byrne et al., 2016). Small-volume effusive volcanism occurred sporadically in impact craters (Prockter et al., 2010;Blair et al., 2013;Fegan et al., 2017;Wright et al., 2018) for an uncertain duration following the cessation of large-volume eruptions. Explosive volcanism persisted throughout much of Mercury's history, but when this process began remains mysterious (Thomas et al., 2014b;Jozwiak et al., 2018). While large-volume effusive volcanism was coming to an end, the largest thrust-fault-related landforms formed in response to Mercury's global contraction (Byrne et al., 2014;Banks et al., 2015;Giacomini et al., 2020). There is evidence that small-scale thrust faulting continued into Mercury's recent geological history (Watters et al., 2016). Thus, Mercury's geological evolution has been dominated by a widespread horizontally compressive stress regime imposed by global contraction in response to secular cooling of the planet's interior. The Hokusai quadrangle represents a useful site on Mercury in which to examine the interplay between tectonic and volcanic activity before, during, and after the innermost planet transitioned into a state of global contraction.
Lower-volume flows that form the intermediate plains in H05 could represent the terminal smooth plains volcanism if they formed similarly to the graded terrain of Byrne et al. (2013), but the stratigraphic relationship between Borealis Planitia smooth plains and the intermediate plains remains uncertain. After global contraction caused the cessation of large-volume effusive volcanism on Mercury, impact craters such as Rachmaninoff and Rustaveli hosted relatively minor effusions that allowed the preservation of the small but compositionally distinct pitted ground. There might also have been late-stage, small-volume effusive eruptions that took advantage of fractures associated with lobate scarps (Malliband et al., 2018). The intermediate plains, pitted ground, and small ponds of smooth plains against lobate scarps in H05, and elsewhere, are key targets for the BepiColombo mission to Mercury (Rothery et al., 2020a), with which we might understand further the interwoven history of volcanism and tectonism on the innermost planet.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.