Desiccation cracks provide evidence of lake drying on Mars, Sutton Island member, Murray formation, Gale Crater

Reconstructions of ancient habitable environments on Mars increasingly depend on detailed analysis of sedimentary facies recording aqueous environments. Over the past decade, the Mars Exploration Rover Opportunity encountered ancient eolian, fluvial, and lacustrine environments deposited in hypersaline, acidic, sulfate- and hematite-enriched playas formed in interdune depressions at Meridiani Planum (e.g. Grotzinger et al., 2005). This setting contrasts with the clay- and magnetite-bearing, moderate pH, perennial lacustrine facies in Gale crater (Grotzinger et al., 2014, 2015; Hurowitz et al., 2017; Rampe et al., 2017). Suites of sedimentary structures, facies associations, and authigenic and diagenetic mineral assemblages were essential to recognize these paleoenvironmental settings. Previously, potential martian desiccation cracks were identified in multiple sedimentary deposits from orbit (e.g., El-Maarry et al., 2014) and in situ by rovers (Grotzinger et al., 2005, 2014).

The kilometers-thick sedimentary succession in Gale crater provides an opportunity to observe changes in surface environments over extended periods in martian history. Studies of basal strata in the informally named Murray formation demonstrated the presence of long-lived perennial lakes in Gale crater at ca. 3.6–3.2 Ga (Grotzinger et al., 2014, 2015; Hurowitz et al., 2017). Recent facies observations at higher stratigraphic levels (Fedo et al., 2017) may record an evolution of the environment over time. Here we present in situ evidence for lithified desiccation cracks in the Murray formation, indicating that the lakes may have partially dried in its younger history.

During Sols 1555–1571, Curiosity investigated a series of distinctive centimeter-scale reticulate ridges on the surfaces of several slabs of rock that expose bedding planes in the Sutton Island member of the Murray formation. Their morphology and composition is characterized to determine if they formed via desiccation and to examine implications for the deposition of associated strata.

As of Sol 1700, the Curiosity rover has explored more than 200 m of strata consisting of fluvial, deltaic, lacustrine, and eolian sediments (Williams et al., 2013; Grotzinger et al., 2014, 2015; Banham et al., 2016; Edgar et al., 2017) represented by the Bradbury group, the interfingering and overlying Murray formation (Mount Sharp group), and the unconformably overlying Stimson formation (Siccar Point group) (Fig. 1). The first ∼25-m-thick Murray interval consists dominantly of finely laminated mudstones with minor siltstones and sandstones of lacustrine origin (Grotzinger et al., 2015). It is overlain by an ∼25-m-thick interval with decimeter- to meter-scale cross-stratification that suggests sediment transport as large bedforms or in channels (Fedo et al., 2017), followed by >30 m of finely laminated red/purple-hued mudstone with intervals of very fine sandstone, consistent with sediment accumulation in subaqueous lacustrine environments (Grotzinger et al., 2015; Fedo et al., 2017). These younger strata, comprising the Sutton Island member of the Murray formation, expose broken and tilted slabs of bedrock, including finely laminated red mudstones, centimeter-scale ripple cross-laminated mudstone, decimeter-scale cross-stratification, and massively bedded intervals of siltstone (Fedo et al. 2017).

The focus of the investigation is an ∼80-cm-long, 40-cm-wide rock slab called “Old Soaker” (OS) that exposes a bedding plane with a red surface marked by a network of ridges that form polygons (Fig. 2A). The red mudstone is ∼1 cm thick and overlies a gray sandstone bed containing bedding-parallel seams of calcium sulfate (CaSO₄). OS and a similar nearby slab called “Squid Cove” (SC) were imaged with the Mast Camera (Mastcam) and the Mars Hand Lens Imager (MAHLI) to characterize the geometry and fill of the ridges. Their elemental compositions were examined with the rover’s ChemCam Laser Induced Breakdown Spectrometer (LIBS) and Alpha-Particle X-Ray Spectrometer (APXS).

The geometries of the polygonal ridges were determined using MAHLI images to evaluate whether their shape is consistent with desiccation. Images of ridges and their junctions were traced to calculate vertex angle distributions, widths of ridges and the polygons they form, and ridge surface area. A three-dimensional (3-D) model of OS was generated from 76 MAHLI images processed using photogrammetry software. The grain sizes of the red and gray beds were measured with ∼16 µm/pixel MAHLI images.

The red surfaces of OS and SC are covered by networks of arcuate ridges with up to 5 mm of positive relief that define predominantly four-sided and some five-sided, 0.5–3.5-cm-wide polygons (Figs. 2B and 3A). Red surfaces of adjacent slabs also show raised ridges spanning an area of a few square meters. The ridges range in length from a few centimeters to ∼0.3 m and mostly meet orthogonally, forming T-junctions (Fig. 3B). The ridges are made of red-to-gray sediment similar in color to the surrounding bed (Figs. 2B and 2C) and comprise ∼20% of OS’s surface. No grains in the ridges or surrounding surface are resolved in MAHLI images (Fig. 2C), indicating a maximum grain size of coarse silt. CaSO₄ veins distinct from ridge material follow most, but not all, of the ridges (Figs. 2B and 2C) and in some cases cross-cut the ridges (e.g., Fig. 2E). Sub-millimeter-wide fractures occur within the polygons (Fig. 2C). Gray, semi-circular, millimeter-scale patches dot the red beds on OS and SC. They can show raised relief and in places are cross-cut by veins (Figs. 2B and 2D).

Some very fine sand grains and millimeter-scale concretions or embedded grains are visible in MAHLI images of the gray bed at OS (Fig. 2F). The ridges taper off within millimeter-scale depressions in the red mudstone at OS (Fig. 2G). Fractures associated with the ridges of the SC slab penetrate the red mudstone and terminate at the boundary with the underlying gray sandstone (Fig. 4). The gray beds appear to lack ridges (Figs. 2A and 2F).

ChemCam analysis of OS identified three distinct bed compositions (Fig. 2A): (1) a lowermost bright sandstone with no ridges and a composition consistent with cementation of sandstones by calcium sulfates; (2) a gray bed with comparatively high K₂O abundance relative to the bright sandstone (1.5–2.5 wt%); and (3) an overlying red mudstone compositionally similar to other Murray mudstones (Table 1) (Mangold et al., 2017). APXS measurements of OS show that the red mudstone bed is similar in composition to average Murray bedrock, but is two to three times richer in Cl (2–3 wt%) and Br (1150–1430 ppm). The gray bed (target “Fresh Meadow”) is distinct from the overlying red bed, with relatively enriched K₂O, SO₃, Na₂O, and FeO_T and depleted TiO₂, SiO₂, and Al₂O₃ (Table 2).

ChemCam observation points on the ridges validate that their composition is distinct from CaSO₄ vein fill and close to that of the gray bed, with lower Al₂O₃ and SiO₂, high H emission lines, and higher K₂O abundances than the red bed. The presence of strong H lines on the ridges indicates the presence of a significant component of hydrous phases absent from the red layer. The dark patches (target “Gilley Field”; Table 1) in the red bed are enriched in FeO (up to 27 wt%) and MnO (0.7 wt%) relative to the surrounding rock. The bright veins are similar to CaSO₄ veins encountered since the beginning of the mission (Table 1) (Nachon et al., 2017).

Proposed formation mechanisms for the ridges must account for several observations: (1) ridges form polygonal networks with T-junctions and continuous arcuate shapes; (2) the ridges in the red mudstone beds correspond to fractures that penetrate those beds; (3) the fractures are restricted to the red beds and terminate at the boundary with coarser underlying material; (4) the fractures are filled with very fine-grained sediment; (5) CaSO₄ veins run along many but not all of the ridges, in some cases cross-cut the ridges, and, unlike the ridges, cut all beds in exposed cross sections; and (6) the ridges are compositionally similar to the underlying gray bed. The most likely fracturing mechanisms include desiccation, synaeresis, and hydraulic fracturing.

Shrinkage cracks form in response to tensile stresses within sediment that result from contraction due to moisture or heat loss (Shorlin et al., 2000). When stress exceeds local tensile strength, materials fracture and cracks begin to grow orthogonal to the direction of maximum tensile stress, typically resulting in a polygonal pattern (Sletten et al., 2003). In uniform material, new cracks will turn to converge with other cracks orthogonally, resulting in junctions mostly near 90° (Shorlin et al., 2000) as observed at OS and SC (Fig. 3). Abundant T-junctions show that sediments dried to completion, possibly in a single event, rather than undergoing multiple wetting and drying cycles that tend to form 120° junctions (Goehring et al., 2010).

Desiccation cracks form at the sediment-air interface and are preserved in the rock record through sediment infill from overlying strata (Plummer and Gostin, 1981). The compositional and color similarity of the ridges to the average Murray formation, which is predominantly comprised of silt-sized grains or smaller, suggests that the ridges are comprised of sediment. Ridge-forming sediment at OS and SC is indistinguishable from the surrounding bed based on grain size alone, so this observation is not definitive evidence for sediment infill from an overlying bed.

Sulfate-mineralized fractures attributed to hydraulic fracturing are prevalent throughout the Murray formation (Grotzinger et al., 2014, 2015; Caswell and Milliken 2017; Young and Chan, 2017), and CaSO₄-filled veins also run along most of the OS ridges, so burial-related hydraulic fracturing may be considered a potential mechanism for the origin of the ridges. However, cross-cutting relationships indicate that the ridges and their infilling materials were lithified prior to the formation of sulfate-filled fractures; sulfate-filled fractures cross-cut some ridges and are not visible along all ridges (Figs. 2C and 2E). Moreover, hydraulic fracturing should yield relatively consistent fracture orientations (Hubbert and Willis, 1972), which are not observed at OS or SC. Zones of weakness created by early sediment-filled fractures were likely overprinted by burial-related stresses (Caswell and Milliken, 2017), followed by precipitation of calcium sulfates.

The ridges are restricted to the red surfaces and their associated fractures terminate at the boundary with the underlying sandstone, consistent with desiccation of a thin mud layer. This style of termination is inconsistent with synaeresis cracking or hydraulic fracturing, which in the latter case would also be expected to cross-cut the bedding planes (Hubbert and Willis, 1972; Plummer and Gostin, 1981; Young and Chan, 2017). The 0.5–3.5 cm length scale of the polygons on OS and SC is consistent with a millimeter- to centimeter-thick deformable layer, similar to the observed thickness of the red mudstone bed and analogous to terrestrial experiments (e.g. Shorlin et al., 2000). Variation in polygon size across slabs may be due to changes in basal friction, bed thickness, or impurities. The cracks are parallel-sided (do not taper downward) in profile (Fig. 4), which can occur if the coupling between the desiccated and underlying beds is low enough to not affect the fractures (Shorlin et al., 2000).

The ridges likely formed via desiccation of a surficial mud layer and filling by sediment sourced from an overlying bed. Although the ridges and underlying gray bed are compositionally similar, the occurrence of interstratified gray and red beds suggests that a gray mudstone originally covered the cracked red mudstone and acted as a source of fracture fill material. Diagenesis associated with later fluids may account for their current similar compositions. Sulfates and other salts may have formed during desiccation as evaporites and/or after the fractures lithified. After deposition, the Murray formation was buried under up to several kilometers of sediment that likely provided sufficient overburden pressure to generate hydraulic fractures (Caswell and Milliken, 2017; Young and Chan, 2017). These fractures likely propagated along pathways of reduced strength produced by the desiccation cracks, and in some cases cross-cut polygons. These fracture networks then acted as conduits for fluid precipitation of CaSO₄ cements.

Recognition of distinct suites of sedimentary structures is a powerful tool in interpreting Mars paleoenvironmental history as it has been for Earth. The Murray formation is interpreted to record a transition from long-lived (∼10⁵–10⁷ yr) perennial lacustrine conditions observed in the basal Murray (Grotzinger et al., 2015; Hurowitz et al., 2017) to episodically exposed conditions recorded by the desiccation-cracked surfaces in the Sutton Island member of the Murray. The predominance of T-junctions at OS and SC indicates a single drying event rather than multiple cycles of wetting and drying. The identification of desiccation cracks is an important facies attribute of the Murray formation that suggests a history of oscillating lake levels that led to intermittent exposure and possible evaporative diagenesis. The drier conditions may represent a temporal transition from deeper lacustrine facies dominated by suspension fallout to shallower lakes with more common traction deposition and desiccation. The Sutton Island member is ∼70 m stratigraphically higher than the Pahrump Hills member, but is also >1 km closer to the center of the Gale lake basin. This suggests the exposed lake facies are not basin margin facies, but rather lowstand facies which represent lake level oscillations.

We are indebted to the Mars Science Laboratory (MSL) project engineering and science teams, and MSL team members who participated in tactical and strategic operations, for their efforts that were vital in collecting the data presented. Thanks to S. Kattenhorn, K. Benison, and K. Herkenhoff for reviews that improved this manuscript. Some of this research was performed at the Jet Propulsion Laboratory, California Institute of Technology (USA), under a contract with NASA. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Grant No. DGE-1144469. These data are archived in the Planetary Data System (pds.nasa.gov).