Pityusa Patera is the southernmost of four paterae in the 1.2 × 106 km2 wrinkle-ridged plains-dominated Malea Planum region of Mars. Based on their texture, morphology, and uniqueness to Pityusa Patera, we interpret layered, folded massifs as pyroclastic deposits emplaced during patera formation as a collapse caldera. Such deposits would not be expected in a previously suggested scenario of patera formation by subsidence from lithospheric loading. Our structural measurements and modeling indicate that the folding and high relief of the massifs resulted from ~1.3%–6.9% of shortening, which we show to be a reasonable value for a central plug sagging down into an assumed piston-type caldera. According to a previously published axisymmetric finite-element model, the extent of shortening structures on a caldera floor relative to its total diameter is controlled by the roof depth of the collapsed magma chamber beneath it, which would imply Pityusa Patera formed above a chamber at 57.5–69 km depth. We interpret this value to indicate a magma chamber at the crust-mantle interface, which is in agreement with crust-penetrating ring fractures and mantle flows expected from the formation of the Hellas basin. As such, the folded massifs in Pityusa Patera, which are partially superposed by ca. 3.8 Ga wrinkle-ridged plains, should consist of primordial mantle material, a theory that might be assessed by future hyperspectral observations. In conclusion, we do not favor a formation by load-induced lithospheric subsidence but suggest Pityusa Patera to be one of the oldest extant volcanic landforms on Mars and one of the largest calderas in the solar system, which makes the folded, likely mantle-derived deposits on its floor a prime target for future exploration.

The Malea Planum region is an ~1.2 × 106 km2 physiographic domain defined by Noachian to early Hesperian wrinkle-ridged plains (including Malea Planum) located in Mars’ southern hemisphere, southwest of the Hellas basin (Fig. 1A; Peterson, 1978; Tanaka and Scott, 1987; Williams et al., 2009, 2010a, 2010b; Tanaka et al., 2014). The Malea Planum region hosts four large and morphologically distinct “paterae”, i.e., irregularly to round-shaped, rimless, flat-floored depressions. The second-largest and southernmost, Pityusa Patera, is centered at 37.37°E, 67.17°S and has been ascribed a maximum depth of ~1.5 km and variable diameters ranging from ~170 to ~400 km due to its very gently sloping inner walls (Figs. 1B and 1C) (Head and Pratt, 2001; Plescia, 2003; Williams et al., 2009). Crumpler et al. (1991, 1996) suggested lithospheric subsidence caused by sill intrusions to form paterae with such muted topography (so-called “Arsia-type calderas”). Larson (2007) then suggested such subsidence from loading, e.g., by mid-crustal intrusions akin to Idaho’s Snake River Plain (northwestern United States), to have formed Pityusa Patera and cited its large diameter, lack of discernible bounding scarps or an edifice-like rise, and the apparent absence of caldera-typical volcanic flows and/or vents as reasons. However, recent mapping by Bernhardt and Williams (2021) showed that Pityusa Patera is superposed by the younger, ubiquitous wrinkle-ridged plains characterizing the Malea Planum region (Fig. 1D, unit Npr—Noachian ridged plains), thus explaining the apparent lack of tectonic and volcanic landforms expected for calderas. Nevertheless, while the formation of Pityusa Patera as a caldera reflecting vertical motion of the surface in response to volume changes at depth by a giant, Yellowstone-like magma chamber collapse was tentatively favored by some previous investigations (Peterson, 1978; Head and Pratt, 2001; Williams et al., 2009), no indicative observations were made in support of such a scenario. Because Pityusa Patera is potentially the oldest and one of the largest extant caldera on Mars (Bernhardt and Williams, 2021), further assessment of its origin is crucial to understanding Mars’ early evolution and inform future exploration.

We present new observations including the analyses of previously undescribed massifs of folded, potentially pyroclastic deposits as well as structural investigations based on modeling by Zuber and Mouginis-Mark (1992) that indicate Pityusa Patera to have formed not by loading-induced lithospheric subsidence, but as an actual volcanic mega-caldera from collapse of a magma chamber, potentially at the crust-mantle interface.

Based on data from the Mars Express High Resolution Stereo Camera (HRSC; e.g., Neukum et al., 2004; Gwinner et al., 2009) and by the Mars Orbiter Laser Altimeter aboard Mars Global Surveyor (Smith et al., 2001), Pityusa Patera has a best-fit radius of ~115 km as defined by the largest long-wavelength (25 km) slope change (Fig. 1C) around it. Mars’s largest unambiguous volcanic caldera outside of the Malea Planum region is that on the summit of Arsia Mons and has a best-fit radius of ~65 km (e.g., Crumpler and Aubele, 1978). However, the larger depression flanked by Nili and Meroe Paterae in the center of Syrtis Major Planum likely also represents a collapse caldera that was later superposed by wrinkle-ridged plains and, with a best-fit radius of ~120 km, is even larger than Pityusa Patera (Kiefer, 2004). Furthermore, the largest potential volcanic calderas on Earth and Venus are Apolaki caldera in the West Philippine Basin and Sacajawea Patera on Lakshmi Planum, which have best-fit radii of ~80 km and ~100 km, respectively (Roberts and Head, 1990; Barretto et al., 2020), thus making them only ~30% and ~13% smaller than Pityusa Patera. We therefore submit that Pityusa Patera’s size is still comparable to that of other examples and should not be considered as an argument against an origin as a collapse caldera.

Massifs as much as ~1.2 km high with lobate lineations (Fig. 1D, unit Nml—Noachian lobately lineated massifs) cover 6030 km2 (~17%) of Pityusa Patera’s topographically defined floor. Pityusa Patera is the only caldera-like depression on Mars to host such kind of material. Unit Nml is embayed by, and therefore older than, all adjacent units including the wrinkle-ridged plains (unit Npr), for which Bernhardt and Williams (2021) derived a model age of ca. 3.8 Ga. The massifs are characterized by lineations at an average spacing of ~150 m formed by ridges that are as much as several tens of meters high and wide (Fig. 2). Some ridges expose meter-scale blocks and commonly form extensive, subparallel, lobate patterns, including (semi)circular arrangements as much as ~3 km wide (Fig. 2B, two black arrows). Patterns formed by the ridges are neither predominantly perpendicular nor parallel to slopes, as would be expected if the ridges resulted from gravity-driven surface processes. Only four impact craters with widths >2 km occur on unit Nml, two of which show lineation ridges traversing their rims without visible deviation (Fig. 2B, white arrow). Based on these observations, we interpret the small ridges to be surface expressions of truncated, folded layers and used the LayerTools add-in for ArcMap software (https://desktop.arcgis.com/en/arcmap/; Kneissl et al., 2010) to obtain 58 bedding attitudes (strikes and dips). To measure a bedding attitude, a layer was traced with five to 27 points depending on its exposed length, thereby mitigating errors in point placement and georeferencing (Fig. 2A; see Table S1 in the Supplemental Material1). The estimated average vertical error of the HRSC digital terrain model (DTM) of ~10 m (Gwinner et al., 2009) should result in dip errors of no more than 1.1° given that all measurements extend across >100 m of elevation and >1 km of baseline distance. We were able to measure antithetic dips along three fold trains, i.e., adjacent anti- and synforms ("north", “center”, and “south” in Fig. 2). Because the resolution of the DTM prevented an exact determination of each fold limb’s length, we weighted each fold in a train equally based on the relatively equal spacing between apparent fold axes (Fig. 2B, black lines with arrows). Assuming we were able to include the entire lengths of the three fold trains and neglecting potential tilts and deformations of fold axes, the estimated average horizontal shortening strain for unit Nml, ε(Nml), would range between −1.3% and −6.9% (accounting for measurement/fitting/data errors; see also Table S1).

Given that unit Nml in its current form is older than the wrinkle-ridged plains and unique to Pityusa Patera’s interior (Bernhardt and Williams, 2021), we infer that the compressive stress field that folded unit Nml was confined to the patera and unrelated to wrinkle-ridge formation. Furthermore, we suggest the material of the folded massifs to be genetically related to Pityusa Patera, potentially pyroclastic deposits, as indicated by unit Nml’s low impact-crater retention, eroded appearance, and apparent ability to accommodate strain via folding indicating a comparatively low rock strength. In this case, unit Nml would be the oldest unit of volcanic origin in the Malea Planum region and among the oldest volcanic materials on the surface of Mars (e.g., Xiao et al., 2012; Tanaka et al., 2014). However, due to the widespread superposition by the wrinkle-ridged plains, neither the exact timing of unit Nml emplacement during caldera formation nor unit Nml’s true geographic extent (possibly indicating an uneven distribution from a trapdoor-style collapse, etc.) can be reliably determined with currently available data.

Shortening within calderas (e.g., Fernandina caldera [Earth] or Zeus Patera [Mars]) as indicated by folded unit Nml may be the result of the interior caldera floor sagging down as a single lid or coherent “plug” within a piston-type caldera (Fig. 3; e.g., Zuber and Mouginis-Mark, 1992; Roche et al., 2000; Kusumoto and Takemura, 2003, 2005; Howard, 2010; Holohan et al., 2015). While this inner plug is bound by a vertical or outward-dipping reverse fault forming early during caldera collapse, it causes little to no horizontal strain, whereas subsequently forming inward-dipping normal faults accommodate nearly all of the vertical movement and cause the majority of extension and shortening of the outer and central caldera zones, respectively (e.g., Kusumoto and Takemura, 2003; Hardy, 2008; Howard, 2010). Therefore, assuming that all horizontal compression is accommodated by it, the central plug is shortened according to the dip angle of the normal fault as it is sagging into a narrowing funnel during and after collapse. In a simplified, axisymmetric case, this shortening, S, can be expressed by:
where θ is the dip angle of the normal fault and D is the rim-to-floor depth of the caldera. Experiments and terrestrial observations have shown that θ for piston-type calderas can be as shallow as ~50°, although values >60° are more common (Roche et al., 2000; Kusumoto and Takemura, 2005; Hardy, 2008 ). The current depth of Pityusa Patera is ~1.5 km, although this is a minimum value due to the substantial post-collapse infill, including wrinkle-ridged plains material as much as ~2.7 km thick as indicated by ghost crater and structural wrinkle-ridge analyses. For a range of θ = 50°–70° and DPityusa = 1.5–4.2 km, SPityusa would be ~0.9–7 km. Assuming the geographic extent of unit Nml across ~205 km to be the minimum diameter of the sagged plug in Pityusa Paterae, the calculated shortening value results in a strain of ε(Pityusa plug) = −0.5% to −3.4%. This strain range is compatible with the strain derived from the folded layers of unit Nml within Pityusa Patera (~−1.3% to −6.9%).

Piston-type calderas (Fig. 3) bound by inward-dipping normal faults tend to form above collapsing magma chambers whose roof depth d is larger than their radius (e.g., Lipman, 1997; Roche et al., 2000; Cole et al., 2005; Kusumoto and Takemura, 2005; Hardy, 2008; Howard, 2010). This not only results in the above-mentioned compressive stress affecting the interior plug of the caldera, but also is in agreement with an earlier axisymmetric finite-element model of the stress field in Zeus Patera, the oldest and largest of Olympus Mons’s summit calderas (Zuber and Mouginis-Mark, 1992). The model showed that the change in surface stress indicated by ridges giving way to graben at half of Zeus Patera’s best-fit radius (RC) is not strongly sensitive to the magma chamber’s aspect ratio, internal pressure, or relative stiffness but is highly dependent on the radius and depth of the chamber. Therefore, this stress transition at a certain fraction of RC can inform the depth d to the roof of the magma chamber (assuming the chamber width to be ~0.75× to 1× the caldera width, which is very likely in our piston-type scenario). For Pityusa Patera, the fraction of RC(Pityusa) can be constrained by unit Nml, which our observations indicate to have undergone folding by a stress field that was limited to within the caldera. Unit Nml can be found as much as ~103 km from the caldera center, for which we estimate a physiography-based best-fit radius RC(Pityusa) of ~115 km, thereby resulting in a minimum radius fraction of ~0.9 × RC(Pityusa). According to the model by Zuber and Mouginis-Mark (1992), 0.9 × RC(Pityusa) would correspond to a magma chamber depth d of 0.5–0.6 × RC(Pityusa) = 57.5–69 km. This depth range should be regarded as maximum estimate, given that the magma chamber radius might be smaller than that of the caldera (Roche et al., 2000; Hardy, 2008; Howard, 2010).

The strain of −1.3% to −6.9% we derived from the folding of the massifs in Pityusa Patera is compatible with the −0.5% to −3.4% that would be created in a central plug sagging into Pityusa Patera assuming a piston-type caldera model. Based on their texture and morphology, the layered, folded massifs lend themselves to an interpretation as volcanic deposits, possibly pyroclastics of a caldera-forming eruption sequence. Because their exclusive location within Pityusa Patera indicates these deposits to be genetically related to patera formation, and because such deposits would not be expected in the previously suggested alternative scenario of patera formation by subsidence from lithospheric loading (Crumpler et al., 1991, 1996; Larson, 2007), we interpret Pityusa Patera as an ancient collapse caldera, potentially the largest and oldest of its kind on Mars.

According to the axisymmetric finite-element model by Zuber and Mouginis-Mark (1992), the extent of shortening structures on a caldera floor relative to its total diameter also informs the roof depth of the collapsed magma chamber beneath it, which would imply Pityusa Patera’s chamber was at a depth of 57.5–69 km. This would correspond to the current crustal thickness, for which models predict a moderate local thinning within Pityusa Patera to ~55–60 km (Parro et al., 2017). Although crustal thickness might have been different during patera formation at ≥3.8 Ga, models suggest it has not changed much since then (e.g., Breuer and Spohn, 2003), therefore putting the Pityusa magma chamber at the crust-mantle boundary. This contrasts with Olympus Mons or comparable volcanoes on Earth such as Kīlauea (Hawai'i), whose magma chambers were estimated to be within the edifice; i.e., at depths of ≤16 km (e.g., Zuber and Mouginis-Mark, 1992). On Venus, crust-mantle–boundary magma chambers have been invoked to explain the diameters and depths of its largest calderas (Head and Wilson, 1986). Terrestrial crust-mantle–boundary magma chambers with attributed surface activity have been suggested for some volcanoes of the Cascade Range in northern California (Elkins Tanton et al., 2001). Here, olivine tholeiites are created at ~36–66 km depth, possibly a result of mantle flows being deflected by the underlying subducting slab of the Farallon plate. As for Pityusa Patera, a magma chamber as much as 57.5–69 km deep collapsing by depressurization (and thus likely feeding surface activity) would be in good agreement with predicted mantle-fed volcanism facilitated by deep ring fractures and mantle upwelling caused by the Hellas impact event (e.g., Peterson, 1978; Williams et al., 2009). If unit Nml represents deposits derived from such a mantle source, hyperspectral observations uncompromised by dust and revealing corresponding signatures (e.g., high-Mg olivine and low-Ca pyroxene) that are spatially associated with units Nml’s ridges, i.e., layers, would be vital to further this discussion, but are not available at present. In conclusion, we suggest Pityusa Patera to be one of the oldest extant volcanic landforms on Mars and one of the largest calderas in the solar system, which makes the folded, likely mantle-derived deposits on its floor a prime target for future exploration.

This work was funded by the German Research Foundation (grant number BE 6457/1-1) and conducted at the Ronald Greeley Center for Planetary Studies at Arizona State University (USA). We would like to extend special thanks to Christian Klimczak from the University of Georgia (USA) for his input concerning our structural measurements. We are also very grateful to Walter S. Kiefer from the Lunar and Planetary Institute, as well as to four anonymous reviewers for their valuable feedback.

1Supplemental Material. Figure S1 (full-size version of Figure 2A), and Table S1 (all 58 measured bedding attitudes, including various error parameters as well as median and average dip values). Please visit https://doi.org/10.1130/GEOL.S.14417636 to access the supplemental material, and contact editing@geosociety.org with any questions.
Gold Open Access: This paper is published under the terms of the CC-BY license.