The geologic record of a planetary body is reflective of its dominant rock-forming processes, preservation mechanisms, and degrees of erosion and exposure. In the sense that Earth and Mars are both “terrestrial” planets and the laws of physics are universal, analogs are useful because they are amenable to detailed study that can be applied to other environments. However, in many cases, only “partial” analogs are appropriate given the vastly different physical, chemical, and geological conditions between planets. Just as the suite of metamorphic facies on Earth resulting from plate tectonics is rare or absent within Mars’ contiguous lithosphere, unique processes in the Martian environment have produced rocks with no direct terrestrial analog. As will be shown in this paper, the vast quantities of dust on Mars have likely formed “duststone.” Although analogous to terrestrial loess in its mode of deposition and primary particle size and to yardangs in its erosion style, it stands on its own as a probable newly recognized rock type in the Solar System. Its geologic extent may be as significant as sandstone is on Earth. Major questions concern its genesis, evolution, and longevity. The intent of this paper is to show that duststones exist on Mars and that they are a component of that planet’s global dust cycle. By integrating observational- and theoretical-based arguments, we discuss duststone formation and compare it to analogous processes on Earth. This paper is partly a review of the duststone hypothesis proposed by Bridges et al. (2010), but with the addition of a detailed comparison to the dust cycle and deposits on Earth.

To properly frame our discussion and arguments, definitions must be established. For the purposes of this paper, dust is defined as any relatively fine-grained particle capable of being suspended in a planet’s atmosphere for a considerable length of time. Theoretically, the maximum size for suspension can be derived by balancing the threshold friction speed (u*_t) against the terminal fall velocity, with larger-size material predicted to saltate (Greeley and Iversen 1985). In practice, this predicts suspended material grain sizes that are too large. For example, the saltation/suspension boundary for basalt on Mars should be 175 μm (Bridges et al. 2010), despite the fact that eolian ripples, features formed by saltation and creep processes, contain particles less than 100 μm in diameter, which theory predicts should be suspended (Sullivan et al. 2008). This discrepancy between theory and observation may exist because of the lower atmospheric density on Mars compared to Earth, which causes particles to respond more slowly to turbulent eddies that drive suspension (Sullivan et al. 2005, 2011). The size range of suspended dust on Mars is estimated at 2 to 5 μm (Kahn et al. 1992, Lemmon et al. 2004), which is on the lesser side of the distribution of sizes found for loess, but similar to dust aerosols on Earth (Fig. 1). As will be discussed more comprehensively later, dust on the Martian surface commonly aggregates into larger particles and reaches sizes such that it can only move by saltation, or, upon further growth, it becomes immobile. Therefore, we define “dust” as particles that are ones to tens of micrometers in diameter that are suspended or deposited and “dust aggregates” as particles ∼100 μm and larger that are on the surface and derived from dust. “Sand” is defined as grains ∼100 μm and larger that undergo saltation and are the direct by-product of disintegrated rock.

Although there has been considerable research comparing Martian and terrestrial dunes, the ways in which Martian dust and its depositional character contrast to that on Earth have received less attention. The general lack of geomorphic expression in terrestrial dust and loess deposits is at least part of the reason such analogs have not been extensively recognized or discussed. With increased sophistication in imaging of the Martian surface and atmosphere, it is now possible to make at least some comparisons between fine-grained eolian deposits on the two planets. In this section, we review dust and loess deposits on Earth, including their genesis, geologic record, and relation to climate forcing through time.

Fine-grained eolian deposits on Earth are classified as “loess” and “aerosolic dust,” unlike the simpler classification of all fine-grained material on Mars as “dust” (see previous). On Earth, these fine-grained materials can be viewed as intermediate members of a continuum of eolian deposits that range from sand to very-fine-grained (<1-μm-diameter) particles that are transported in the stratosphere (Fig. 1). Loess is dominated by silt-sized particles (2–50 μm particle diameters) and is coarser than the material referred to as aerosolic dust (particles <10 μm diameter), which is capable of long-range transport. Typically, loess contains 60–90% silt-sized particles and smaller amounts of sand and clay (<2μm); it may cover as much as 5–10% of Earth’s land surface (Pye 1987), including small areas of Africa and the Middle East, attesting to non-glacial origins for some deposits.

Loess deposits are expressed as distinctive sedimentary bodies. Thickness is highly variable and can range from a few centimeters to several hundred meters, with thickness and mean particle size decreasing downwind away from a source (Pye 1987; Muhs and Bettis 2000, 2003; Muhs et al. 2004, 2008). The deposits commonly have little or no geomorphic expression (Pye 1987), being draped over preexisting landforms as a mantle, with thickest accumulations in protected, low-lying areas or on flat, stable upland divides and thinnest accumulations occurring on narrow, rounded hillcrests (Fig. 2). One exception to the general rule of little geomorphic expression of loess may be “pahas,” which are streamlined loess landforms aligned with paleowinds that may be yardang-like features (Flemal et al. 1972). However, similar morphologies also have been tied to primary deposition (Lewis 1960), so, at best, geomorphic expressions of eolian erosion of loess are limited.

Unlike eolian sand or fluvial–marine sediments, primary structures in loess are subtle or absent altogether. Faint, horizontal laminations and cross-bedding are only rarely apparent (Fig. 3). Rather, most loess deposits are characterized by a massive (as opposed to structured) condition. Interparticle binding by phyllosilicate clay minerals or secondary calcite accumulations commonly result in a significant amount of material strength, allowing for the formation of vertical faces along riverbanks or stream banks and road cuts. Secondary structures in loess are more common than primary structures, and these consist of fractures, burrows, rhizoliths (root casts composed of Fe-oxides or carbonate), carbonate nodules or concretions, oxidation or reduction streaks or bands, and paleosols. Lithification or cementation of loess particles to form siltstone is unknown in Quaternary (past 2.6 Ma) loess deposits on Earth, but it has been hypothesized to explain the origin of some Paleozoic and Mesozoic siltstones (Johnson 1989; Soreghan et al. 1997, 2002; Chan 1999).

Loess has a mineralogy that usually includes quartz, plagioclase, K-feldspar, mica, calcite (and sometimes dolomite), and phyllosilicate clay minerals (smectite, chlorite, mica, and kaolinite) (Pye 1987). Heavy minerals are usually present, but in small amounts. Bulk geochemical studies show that the dominant constituent in loess is SiO₂, ranging from ∼45% to 75%, but more typically in the range 55% to 65% (Pye 1987, Muhs and Bettis 2003, Muhs et al. 2008). The high SiO₂ content of loess deposits reflects a dominance of quartz, but smaller amounts of feldspars and clay minerals also contribute to this value. Clay mineralogy in loess shows considerable variability. In North America, for example, clay minerals of loess are a direct reflection of source sediment. In Alaska, loess deposits are ultimately derived from metamorphic rocks in major mountain ranges and are rich in mica and chlorite (Muhs and Budahn 2006). In the midcontinent, loess in the Mississippi River drainage basin is derived to a great extent from Paleozoic carbonate rocks (explaining the high calcite and dolomite) and shales, which explain the high mica content in the clay fraction (Grimley 2000). Loess in the Great Plains region is rich in smectite, derived mainly from Tertiary volcaniclastic siltstone that is also rich in smectite (Aleinikoff et al. 1999, 2008; Muhs et al. 1999, 2008).

The “glacial” model of loess formation is a traditional view that silt-sized particles are produced dominantly by glacial grinding of rocks, deposition in till, reworking by fluvial processes as outwash, and finally entrainment, transportation, and deposition by wind (Fig. 4a; see also discussion in Muhs and Bettis 2003). This model has led to the view that loess deposits are primarily markers of global glacial periods. Support for the classical glacial model of silt production and later eolian entrainment as loess comes from modern observations in regions such as Alaska, Canada, and Iceland. In these regions and elsewhere, glaciers exist now, and it is possible to observe loess entrainment from rivers that drain glaciated terrain (e.g., Muhs et al. 2004).

There is also strong evidence for nonglacial loess (Wright et al. 1998; Wright 2001a, 2001b; Smith et al. 2002; Crouvi et al. 2010). “Desert” loess is a term used loosely to describe eolian silt generated in and derived from arid or semiarid regions that were not glaciated. The debate on desert vs. glacial loess centers on whether silt-sized particles can be produced by mechanisms other than glacial grinding, particularly in deserts. This has particular applicability for Mars, where glacial activity has been at most a minor contributor to dust-size particle production over the history of the planet (see following). A variety of mechanisms can, in principle, produce silt-sized particles in arid regions, and these are summarized in a highly simplified model (Fig. 4b). These processes include frost shattering at high altitudes, comminution (particle size reduction by crushing or grinding) by fluvial and mass-movement transport, chemical weathering, salt weathering, eolian abrasion, and ballistic impacts (Wright et al. 1998; Wright 2001a, 2001b; Smith et al. 2002; Crouvi et al. 2010). As with the glacial environment, it is possible to observe silt entrainment in modern desert environments. A good example is the yearly transport of fine silts and clays out of Africa by wind, derived from the Sahara and Sahel regions, where glaciation has played no role in silt particle formation. Indeed, although fine particle transport by wind is highly seasonal in Africa, this continent is, overall, the largest source of dust on Earth today.

In the glacial loess-versus-desert loess debate, there is generally little consideration of silt particle inheritance from sedimentary rocks or volcanic tephra. Notable exceptions include sites in Argentina, where an Andean volcaniclastic source has been recognized (Zárate and Blasi 1993), and Australia, where siltstones may be the ultimate source of much of the silt-sized dust in arid basins (McTainsh 1989). Silt is abundant in the sedimentary rock record, and estimates show that fully half of the detrital quartz in the world’s sedimentary rocks is composed of silt-sized particles (Blatt 1987). In the Great Plains of North America, sedimentary rock (volcaniclastic siltstone) is the most important source of silt-sized particles in Peoria Loess of last-glacial age (Aleinikoff et al. 1999, 2008; Muhs et al. 1999, 2008). Tephra is commonly composed of silt-sized particles, and where there are active volcanoes, silt-sized tephra is one of the components of loess, such as is the case in Holocene loess in southern Alaska (Muhs et al., 2004). Thus, in both the “glacial” and “desert” models of silt particle formation, inheritance from tephra or siltstone can be added as processes (Fig. 4).

It is therefore likely that loess in many regions has origins from both glacial and nonglacial processes. There is no question that glaciers are efficient producers of silt. Drainage basins in Alaska that have glaciers have significantly higher sediment yield than nearby basins that lack glaciers (Hallet et al. 1996). In Iceland, several meters of loess, mostly from glaciogenic sources, have been produced entirely within Holocene time. On the other hand, laboratory experiments demonstrate that eolian abrasion and fluvial comminution are very efficient silt producers (Wright 2001b), and it has been shown that at least some component of loess in deserts is derived from abrasion of sand-sized particles by wind to silt-sized particles (Crouvi et al. 2010). This issue is relevant for studies of Martian dust and dust deposits because some processes of silt production on Earth do not occur now on Mars and perhaps only operated billions of years ago.

The geologic record of loess on Earth consists of sequences of relatively unaltered loess with intercalated buried soils, or paleosols. Soil formation takes place when vegetation stabilizes the surface, usually because of a decrease in sedimentation rate. Such conditions may arise because of a decrease in wind strength, elimination of the loess–sediment source, or both. In Europe and much of North America, where loess has a glaciogenic origin, it is no surprise that loess in the stratigraphic record represents glacial periods, and paleosols represent interglacial periods (Muhs and Bettis 2003, Muhs 2007). In China, although the immediate sources of loess are desert basins upwind from the Loess Plateau, the loess is actually of glaciogenic origin, derived from mountain glaciers that surround the desert basins (Sun 2002). Thus, in China as well as in regions of Europe and North America near glaciated terrain, loess represents glacial periods, and paleosols represent interglacial periods that can be correlated with the deep-sea oxygen isotope record of glacial-interglacial cycles. Elsewhere in North America (e.g., the Great Plains), where loess is nonglaciogenic, ages of loess also correspond to glacial periods (Muhs et al. 1999, 2008; Roberts et al. 2003). Therefore, although loess in this region is not dependent on glacial sources, climatic conditions during glacial periods are apparently favorable for loess entrainment, transportation, and deposition.

Geologic records of aerosolic dust flux occur in a number of depositional settings for sediments, including deep-ocean basins, lakes, large ice sheets, and soils. Deep-sea sediments are one of the great archives of glacial-interglacial cycles of the Quaternary. Foraminifera in deep-sea cores faithfully record the ocean temperature changes and fluxes in the oxygen isotope composition of seawater that accompany the shifts between glacial and interglacial climates. The same deep-sea cores may also retain long-range-transported aerosolic dust, if fluvial transport and other means of sediment origin can be ruled out or accounted for. Several decades of study have shown that both the Pacific and Atlantic Oceans have long records of eolian sediment transport from Asia and Africa, respectively (Rea 2007). The coincidence of dust flux maxima with glacial periods indicates that glacial-climate conditions favored greater dust production and transport.

Thus, loess records from both glacial and nonglacial sources, the eolian record in deep-sea sediments, and the eolian record in polar ice caps all indicate that glacial periods on Earth correspond to times of greater dust flux. It is likely that stronger winds, lower vegetation density, drier climates in dust source areas, and a decreased intensity of the hydrological cycle all combined to enhance dust flux repeatedly in glacial periods of the Quaternary (Mahowald et al. 1999). During such times, which approach conditions on Mars, Earth was a far dustier planet than it is today. Quaternary glacial-interglacial cycles on Earth are controlled, to a great extent, by astronomical cycles of precession and tilt of Earth’s axis and, to a lesser extent, orbital eccentricity (Berger 1992). These cycles are particularly critical for the distribution of solar insolation in summer and therefore affect ice-sheet growth and decay in the Northern Hemisphere of Earth. As discussed later herein, astronomical cycles also exert controls on Martian dust cycles.

The first geologic process discovered on Mars is also its most pervasive: Global dust storms were first observed telescopically in the nineteenth century and tend to occur annually (Martin et al. 1992, Newman 2001). Spacecraft observations show that smaller-scale dust storms and vortices are also frequent. It is estimated that 2.9 × 10 kg yr⁻¹ of dust is exchanged between the surface and atmosphere (Pollack et al. 1977). If evenly coated over the surface, this would form a layer just a few micrometers thick. However, as on Earth, the distribution and thickness of dust are nonuniform. Based on radar measurements, accumulations of dust in high-albedo, low-thermal inertia regions are estimated at up to 1 to 2 m from radar measurements (Christensen 1986), and in Arabia, thicknesses of 20 m and greater are based on the lack of small impact craters (Mangold et al. 2009) (a map showing Arabia and other Mars locations presented in the text is shown in Fig. 5). The polar layered deposits at both poles, with a thickness of a few kilometers, have a significant dust component (Kahn et al. 1992, Thomas et al. 1992, Smith et al. 1999, Phillips et al. 2008). Fundamental questions include: How did so much dust form, and how is it preserved on the surface today?

Based on remote sensing and in situ measurements, the composition of Martian dust is dominated by framework silicates, mostly feldspar, with lesser amounts of sulfate, pyroxene, olivine, amorphous phases, hematite, and magnetite—a mineralogy consistent with the mechanical weathering of basalt (Hamilton et al. 2005). This is distinct from the dominantly granitic (or rocks derived from granites) source for the majority of terrestrial loess. The presence of olivine in Martian dust in particular indicates that water did not play a dominant role in dust formation, with genesis occurring during dry periods of Martian history (Goetz et al. 2005).

There are several processes that can potentially produce Martian dust (Fig. 6). Mars has abundant volcanoes and lava flows, some of which have a morphology consistent with pyroclastic eruptions, such that ancient tephra was probably a significant contributor. Impacts have pulverized the surface, releasing large quantities of ejecta, including dust-size material. Meteoritic infall provides a steady source of particulate materials. Based on the impact cratering record, ∼8.3 cm of material has been added to Mars, if distributed evenly over the surface, since the early-mid Noachian (>4 Ga) (Yen et al. 2006). The flux of nonmelted meteoritic material that reaches the Martian surface is estimated at 8.6 ×10⁶ kg yr⁻¹ (Flynn 1996), yielding ∼8 cm of material (assuming basalt density), consistent with the cratering-derived estimate. Erosion associated with outflow channels, valley networks, other fluvial features, and possibly glacial grinding has disaggregated rocks and coarse soil components.

All of these processes were largely confined to the Noachian and Hesperian Periods before ca. 3.5 to 1.8 Ga (Tanaka et al. 1992). These high-energy processes (except for minor meteoritic infall) do not occur today, such that current rates of dust production are considerably lower. Probably the most significant dust-forming mechanism in the modern era is the abrasion and deflation of the relatively soft rock that constitutes yardangs and eolian mantles over much of Mars. Much of this material is likely duststone, the erosion of which recycles dust back into the Martian eolian system. More minor processes include eolian abrasion of rocks and soil by sand rounding and disintegration of this sand in saltation and abrasion, thermal cycling of rocks, and possibly salt weathering (Wells and Zimbelman 1997). On Earth, precipitation cleanses dust from the atmosphere, and oceans and other water bodies serve as permanent sinks. On land, vegetation acts to stabilize dust and form soils. With Mars lacking these characteristics, dust is not easily removed from the eolian system.

Numerous wind tunnel and theoretical studies have demonstrated that the threshold for lofting dust that is ones to tens of micrometers in size by wind, regardless of the planetary surface, exceeds that needed for fine sand (Greeley and Iversen 1985). This is because dust is more easily sheltered by roughness elements, it resides partially within a laminar sublayer, and its low weight makes it more susceptible to electrostatic and other cohesive forces (Iversen et al. 1976, Shao and Lu 2000). Friction speeds needed to raise dust on Mars are on the order of 10 m s⁻¹ which is about an order of magnitude greater than on Earth (Greeley and Iversen 1985). These speeds correspond to free-stream winds over 100ms⁻¹ up to several kilometers per second—conditions rarely, if ever, achieved. So, under standard conditions, linear winds on their own cannot raise dust, and other mechanisms must be considered to account for the significant load of suspended dust in the atmosphere.

Some of the highest wind speeds on Mars are attained by katabatic flow off the retreating seasonal polar cap, combined with diurnal winds flowing from high elevations (Lee et al. 1982, Magalhaes and Gierasch 1982, Kahn et al. 1992). The magnitude of these winds is enhanced in the Southern Hemisphere, where the proximity of perihelion in late spring to the summer solstice produces strong gradients (Wells and Zimbelman 1997). These winds can exceed the threshold friction speeds of ∼2 m s⁻¹ necessary to saltate sand. The flux of dust lofted by saltating sand is proportional to (u*/u*_t)³ (White 1979, Newman et al. 2002) or (u*/u*_t)⁴ (Westphal et al. 1987, Haberle et al. 2003), such that even winds slightly above sand threshold can cause large dust fluxes. Vortices induced by surface–atmosphere thermal instabilities can also saltate sand and provide enhanced lift to raise dust, producing dust devils (Greeley et al. 2004). More minor contributors to dust suspension include removal of dust on pebbles and rocks, which, being higher in the wind speed profile and not sheltered by nearby roughness elements, are subjected to greater shear stress (White et al. 1997), and small-scale surface disruptions such as dust-rich landslides (Sullivan et al. 2001). On the other hand, dust aggregates with sizes equivalent to sand are easily moved (Fig. 7). This likely accounts for the presence of ripples in dust-rich regions (Bridges et al. 2010) and is a component of the duststone model that will be presented herein.

In local events, such as dust devils, dust is transported only a few kilometers. However, global storms can raise dust as high as 30 to 40 km or more, where it is subjected to transport by Hadley cells from the southern latitudes in the summer to the middle and temperate northern latitudes (Toon et al. 1980, Pollack et al. 1981, Magalhaes 1987, Haberle et al. 1993). In the latter stages of global storms, the high-altitude dust acts to dampen the lapse rate (temperature change with altitude), decreasing convective turbulence and causing dust to fall out (Haberle 1986, Wells and Zimbelman 1997). Grains can also serve as condensation nuclei for CO₂ and H₂O, enhancing dust deposition in polar regions (Kahn et al. 1992).

On Mars today, the coincidence of perihelion with southern spring results in global storms originating in the Southern Hemisphere and net dust deposition in the equatorial and northern latitudes via the cross-equatorial Hadley cell. Dust is spread nonuniformly, with enhanced deposition in preexisting high-albedo, low-thermal-inertia, low-rock-abundance regions, specifically, Tharsis, Elysium, and Arabia (Fig. 5). Why these regions serve as dust sinks is not completely understood. One hypothesis is that they may have initially formed in areas of low wind velocity, allowing dust to bury sand and rocks, thereby removing sand saltation triggers and roughness elements that could enhance turbulence to otherwise drive dust into suspension, thereby increasing the net accumulation rate (Christensen 1988).

The argument of perihelion has a period of ∼51,000 years, such that the hemisphere where global storms originate changes every 25,000 years or so, with alternate epochs perhaps having a net north–south transport (Christensen 1986, Kahn et al. 1992, Kieffer and Zent 1992). Under this model, net dust sources and sinks alternate in hemispheric polarity in response to orbital precession. If true, then remnant, thick dust/duststone deposits may exist over much of Mars. These and other hypotheses of Martian dust removal and deposition are simplifications, as terrestrial experience and detailed models show that erosion and deposition are fundamentally driven by the mass balance per unit volume of surface (Exner 1925, Kubatko and Westerink 2007). In detail, it may be that some areas of “dust sinks” on Mars undergo periods of erosion and abrasion, and some “sources” may locally accumulate sediment. In our study of Mars, we are presently limited to observations that reflect broad-scale processes integrated over long time periods.

Three classes of observations provide compelling evidence for the existence of duststone on Mars. Those are outlined next, along with interpretations based on these observations.

Cohesion and geochemical properties are correlated in Martian soil, with cohesive cloddy materials and surface crusts enriched in sulfur and chlorine (Clark et al. 1982; Rieder et al. 1997, 2004; Moore et al. 1999). The formation of this “duricrust” is believed to occur via the mobilization of S and Cl by water-vapor diffusion through the soil (Jakosky and Christensen 1986a) and may account for variations in thermal inertia outside the lowest inertia areas (Jakosky and Christensen 1986a, 1986b; Christensen and Moore 1992). Mapped at coarse scale by the Thermal Emission Spectrometer (TES), duricrust is not exposed everywhere, but rather it makes up one of four distinct thermophysical units, with the corresponding “Unit C” having moderate to high thermal inertia and intermediate albedo (Mellon et al. 2000). This unit consistently borders “Unit A,” in which low inertia and high albedo indicate a dust cover. Tharsis is relatively rich in Cl and depleted in Si and Fe in its upper ∼50 cm, as indicated by Gamma Ray Spectrometer (GRS) data (Karunatillake et al. 2009).

The thermophysical properties of Unit C, combined with its association with Unit A, are consistent with thick cemented dust (Mellon et al. 2000). In the dusty Unit A regions, a dust cover thicker than a couple of centimeters (the diurnal skin depth) can mask any underlying indurated material. Therefore, cemented dust could exist in these regions as well, or be exposed as outcrops at spatial scales finer than the TES data (∼3 km per pixel). In support of this hypothesis, Tharsis, within Mellon et al.’s Unit A, exhibits large seasonal variability and diurnal differences in the apparent thermal inertia that can be attributed to a layered surface, such as loose dust covering an indurated layer, although horizontal mixing and subpixel slope effects are other possible causes (Putzig and Mellon 2007a, 2007b). Finally, in Tharsis, the chlorine-rich composition detected by GRS is consistent with an indurated soil composition.

Images from the Microscopic Imager on the Mars Exploration Rovers (MERs) show dust aggregates that reach sizes of several hundred micrometers (Herkenhoff et al. 2004, 2006, 2008; Sullivan et al. 2008), equivalent to that of sand. The much thinner and drier atmosphere on Mars makes dust much more prone to electrostatic forces than in the Earth environment (Greeley 1979, Krinsley and Leach 1981), such that aggregates and “cloddy material” (as classified at the Viking landing sites) can form (Shorthill et al. 1976, Greeley 1979). In addition to those at the landing sites, aggregates may make up low-thermal-inertia dunes and materials of the north polar erg (Herkenhoff and Vasavada 1999), a region also proposed to contain “loess” deposits (Tanaka et al. 2008). Layered deposits in the Hellas Basin and nearby Terby Crater have been proposed to form from finegrained material such as dust or loess (Wilson et al. 2007). It has also been conjectured that “parna sheets,” analogous to clay aggregate lunette duneforms on Earth, may exist on Mars, although none have been definitively identified (Greeley and Williams 1994).

Dust aggregates are common on Mars, especially in areas for which remote-sensing observations indicate dust-rich compositions. Aggregates will continue to grow until they are disrupted by wind forces. In other words, dust that is ones to tens of micrometers in size that is found in the atmosphere is not likely to remain in this state on the surface, but rather to clump together with other dust particles and aggregates.

High-resolution orbital images from the High Resolution Imaging Science Experiment (HiRISE) camera show that regions on Mars classified as dusty–Unit A (e.g., Tharsis, Arabia, and Elysium) contain suites of fine-scale reticulate bedforms intermixed with wind-sculpted, bright bedrock (Bridges et al. 2010) (Figs. 8, 9). The bedrock is commonly composed of blunt-to-tapered outcrops for which plan-view shape and texture are analogous to terrestrial yardangs. Earth-based radar measurements restrict the depth of a cover of fines to 1 to 2 m on average (Christensen 1986), consistent with a dusty surface above a more indurated substrate. In Tharsis, images of stratigraphic cross sections showing contacts with underlying basalt demonstrate that the light-toned, presumably dust-rich, bedform surfaces sit atop a mantle at least several meters thick (Keszthelyi et al. 2008) (Fig. 10). In Arabia, thickness of dusty material is estimated at 20 m or more (Mangold et al. 2009).

Because the thermophysical properties of these areas indicate a dust-rich composition, and particles with diameter of ones to tens of micrometers are suspended by winds above threshold, Bridges et al. (2010) proposed that the reticulate bedforms were formed from the saltation of dust aggregates hundreds of micrometers in size (Fig. 7b). Furthermore, because aggregates approaching a millimeter in size are below threshold, it was also suggested that continued growth would also act to stabilize the particles. This, combined with the formation of duricrust and subsequent burial by younger dust deposits, would form a semidurable “duststone” that, when exposed, would be subjected to wind abrasion processes.

The putative duststone rock may exist in other areas of Mars (Fig. 11). Buried and exhumed terrain is common on the planet (Malin and Edgett 2001, Frey et al. 2002, Edgett 2005), and much of the middle latitudes is covered by a mantle composed of ice-permeated dust that is many meters thick (Mustard et al. 2001). In some craters that were probably wholly or partially exhumed, light-toned rock masses are apparent. This includes Pollack Crater (containing “White Rock”; Fig.11a) and the upper rock masses of the Gale Crater mound (Fig. 11b), both of which have remote-sensing properties consistent with indurated dust (Ruff et al. 2001, Milliken et al. 2010). A light-toned outcrop in Becquerel Crater and the upper Gale mound also contain nonconformable, rhythmically bedded rock masses with stratigraphic bundles consistent with astronomically-forced mechanisms (Lewis et al. 2008, Milliken et al. 2010) such as is expected for cyclical air-fall dust.

The large volume of dust on Mars and the unique environment of its deposition compared to Earth account for the formation of duststone on that planet as opposed to our own. We present a hypothesis, modified after Bridges et al. (2010), for the formation of duststone that we believe is consistent with current data and suggest several measurements that can be performed by future missions to test it. The major stages of formation are:

1. (1)

   Dust is deposited on the Martian surface from atmospheric fallout. In “sinks,” accumulation exceeds removal, and thick deposits form. Dust clumps into aggregates via electrostatic processes. It eventually reaches a large enough size that it can no longer be suspended when subjected to high-speed winds and instead saltates and forms small-scale ripples, such as those currently seen in the dusty Tharsis, Elysium, and Arabia regions. Many aggregates will be destroyed and comminuted to smaller particles. Depending on the pore space of the growing aggregates, others may remain above the threshold curve and be removed by the winds. However, a fraction, those with great enough density and those not destroyed, will grow to sufficient size such that they are below the threshold curve, resulting in a stable surface. This stable population increases over time (Fig. 7b).

2. (2)

   Once the surface is dominated by particles having a size below threshold, it becomes further stabilized through the formation of duricrust through multiple cycles of water-vapor diffusion (Jakosky and Christensen 1986a).

3. (3)

   Some stabilized surfaces will be buried by new dust deposits, which are, in turn, stabilized as well. Compaction of the underlying dust converts the cemented material into harder duststone.

4. (4)

   If subsequently exposed, the duststone will be subjected to wind abrasion, depending on the thickness and extent of local particulate materials. In many settings, this will be variable, as seen in Tharsis today, with some exposed duststone undergoing erosion and other units being sheltered. A similar scenario is envisioned for putative duststones in craters such as Pollack and Gale, although in these cases the nature of the former burying material is less certain as these areas are not currently dust sinks, although they may have been in past epochs. Sand serves as the main abrasive agent where it is present, given the much greater effectiveness of its ability to erode rock (Laity and Bridges 2009). However, in areas like Tharsis, where sand appears lacking, duststone should only be abraded if it is weakly cemented, because the main abrading material is probably dust aggregates.

5. (5)

   Over time, lithified duststone is eroded, forming scalloped yardangs and remnant rock masses. Many of the eroded particulates probably rejoin the eolian system as suspended grains or as saltating sand-size particles that are incorporated into the soil or break apart and go into suspension.

This model does not consider the role of ice. It has been proposed that the desiccated mantle covering much of the Martian midlatitudes is air-fall dust cemented by ice derived from diffused atmospheric water vapor under favorable orbital configurations (Mustard et al. 2001). This is analogous to the polar layered deposits being composed of dust–ice mixtures (Thomas et al. 1992). Such an ice-rich mantle is more like permafrost than bedrock, although desiccation is predicted to produce loose loess-like materials (Mustard et al. 2001), so that remnant masses could be considered duststone.

Finally, not all light-toned rocks on Mars are duststone. For example, the morphology of the wind-eroded Medusae Fossae Formation and its location near the Tharsis and Elysium volcanic constructs are consistent with a pyroclastic origin (Ward 1979, Scott and Tanaka 1982, Bradley et al. 2002, Mandt et al. 2008, de Silva et al. 2010). Fluvial processes have probably contributed to the formation of the lower parts of the light-toned layer mound in Gale Crater (Milliken et al. 2010, Thomson et al. 2011). It is likely that other light-toned rocks on Mars have similarly diverse origins. Our hypothesis is that many, but not all, light-toned rocks, especially those associated with thick dust deposits, are duststone.

The rates of erosion of duststone, and the ways in which these balance against rates of formation, are not known. Given that the dust cycle on Mars seems to have been occurring since the Noachian, it is likely that cycles of deposition, stability, and erosion have been ongoing ever since. Duststones will be protected from erosion for the duration over which they are buried, such that there could be very old formations beneath the surface today or recently exposed formations. Given the low rate of primary dust production on Mars in the current epoch, the rates of duststone formation versus erosion are directly tied into the question of whether dust abundance overall is increasing or decreasing. Answers to this and related questions cannot be found until more data on rock formation and exposure ages are provided by samples or in situ measurements.

With greater confidence we can say that the range of rock types currently forming on Mars is far more limited than it was in the Noachian and Hesperian. Volcanic and fluvial processes that produced suites of igneous and water-based sedimentary rocks are infrequent today. Thus, any preserved Noachian and Hesperian duststones are probably intermixed with strata formed by other geologic processes. Given their air-fall nature, duststones could serve as temporal marker beds analogous to ash and ash-derived clays on Earth. In the modern era, Martian rock production is limited to the movement of sand and the dust cycle, combined with processes to stabilize and lithify these materials. We therefore consider duststone, perhaps with limited sandstone production, as the dominant rock-forming processes in the current era. This idea can be tested by future in situ missions that visit areas believed to contain duststone.

When we define the dust cycle as describing the production of particles with diameters of ones to tens of micrometers, followed by their deposition and removal or incorporation into the geologic record, there are important commonalities and differences between Mars and Earth. Building upon the previous sections, we compare and contrast the dust cycles on the two planets and conclude with some general statements about the role of dust on other planetary bodies. Some of the ideas here are broad generalizations and are presented as working models that, we hope, will spur further research into the role of dust in the planetary context.

Some geologic processes that Earth and Mars have in common are those that act to comminute rock into finer material that can subsequently be suspended and deposited into accumulations of many meters or more. Although hydrological and particularly glacial processes have been more prevalent on Earth, with the latter considered one of the main producers of loess, the mechanism by which dust is formed has little impact on its subsequent evolution. In other words, it is sufficient to state that dust is formed in abundance on Earth and Mars, and that the modes of deposition, stabilization, accumulation, and removal are the steps in the process where environmental effects exert the strongest biases.

The deposition of dust on Mars is much more uniform than on Earth, where thicknesses are correlated to distance from source. Because Martian dust can be transported vast distances without being flushed out by precipitation or being affected by ocean-induced global circulation patterns, deposition can occur on global scales. Even the net south–north hemispheric bias in the current epoch is counterbalanced, over the long term, by the Martian precessional cycle.

Processes that remove dust compete against those acting to stabilize it, and these mechanisms are significantly different on Mars and Earth. Wind can obviously remove dust on both planets, but the much higher threshold speeds needed in Mars’ low-density atmosphere make areas of low wind intensity or surface roughness much more prone to serve as dust sinks than is the case on Earth, where dust in virtually all locations can be resuspended. On Earth, fluvial erosion is an additional and significant removal agent and one that is completely absent on Mars. Therefore, unconsolidated dust is much more likely to remain stable on Mars.

On the other hand, dust stabilization on Mars seems to require some combination of aggregation, cementation, and burial. No liquid water is involved. In contrast, the active terrestrial hydrosphere and biosphere acts to alter primary dust and lesser amounts of sand and clay into a weakly cemented sediment, commonly bound by carbonate cement and clays. Such a “loessification” process is not expected on Mars today. Pre-Quaternary loessite, though probably requiring water for its diagenesis, may be a closer analog to Martian dust deposits in that it is better cemented and buried.

The erosion of dust deposits on the two planets is likewise quite different. Although loess can be derived from arid settings on Earth, it rarely accumulates in significant amounts in deserts, tending to get transported to areas where it can be stabilized by vegetation. Subsequently, pedogenesis begins, and soil formation dominates until further loess deposition takes place. Even pre-Quaternary loessite has identifiable paleosols (Soreghan et al. 1997, 2002). As such, the erosional expression of loess on Earth is poorly defined geomorphically, although there is no question that there are stratigraphic records of loess erosion seen as unconformities (see Muhs et al. [2003] for examples in Alaska). This is in marked contrast to Mars, where all dust accumulates in dry settings, and subsequent erosion is confined to eolian abrasion and deflation, forming yardangs. In Earth’s deserts, yardangs can be composed of sandstone (quartz, carbonate, or gypsum-dominated) or fine-grained sediments, such as silts and clays.

The role of dust on Mars is unique in the Solar System. Unlike on Earth, it is almost entirely unaffected by hydrological and biological alteration processes that can accumulate into thick sequences. Once deposited, it becomes part of the sedimentary strata, as dust or duststone, or reenters the eolian system through deflation or abrasion. On Earth, about 23% of the surface is sandstone (Ronov 1983), or about twice the amount of loess coverage. Given the dustiness on Mars, the volume of duststone could be as proportionally significant as sandstone is on Earth. This can be verified or refuted by future missions and data analysis.

The differences between Mars and the other worlds in our Solar System with solid surfaces and significant atmospheres (e.g., Venus and Titan) are even more distinct. Venus lacks a hydrosphere, and dust production from volcanic sources is strongly inhibited by the great atmospheric pressure (93 bars), which reduces volatile exsolution (Head and Wilson 1986). Although Magellan radar shows evidence for wind features associated with dust (Greeley et al. 1992), production of fine particles is probably limited to impact processes. Vast deposits of dust like those on Mars or in loess on Earth are not seen. Titan has dunes that are probably composed of frozen organics (Barnes et al. 2008). Dust-size organics or other materials may be present on the surface, but the methane “hydrological” cycle (Tokano et al. 2006) may act to flush it out. Expanding our horizons to extra-solar planets, Mars-like bodies with thin atmospheres having limited or no hydrologic cycles could also have dust that is continually produced and accumulates over time. These may yet be discovered in future investigations.

Dust formation, transport, deposition, stabilization, and erosion are active geologic processes on Earth and Mars. Dust accumulates where it can be stabilized and will not be subjected to resuspension. On Earth, stabilization is driven by hydrological and biological processes. On Mars, accumulation occurs in dust sinks that may switch hemispheric polarity in response to the precessional cycle. The comparison to loess is therefore informative in understanding the ways in which dust on two terrestrial planets can accumulate and then have divergent evolutionary paths. On Mars, which lacks liquid water and biology, dust is stabilized by growth of micrometer-sized dust into aggregates and cementation. Continual accumulation will compact these materials, forming dust-stone. The role of dust on the two planets is similar in that thick accumulations can form, but the different environments result in divergent evolutionary paths.

Comments from D. Rubin, J. Pigati, and an anonymous reviewer significantly improved this paper.