Title Glacial and gully erosion on Mars: A terrestrial perspective

Authors Conway, SJ; Butcher, FE; de Haas, T; Deijns, AA; M. Grindrod, P; Davis, JM

Description publisher: Elsevier articletitle: Glacial and gully erosion on Mars: A terrestrial perspective journaltitle: Geomorphology articlelink: https://doi.org/10.1016/j.geomorph.2018.05.019 content_type: article copyright: © 2018 Elsevier B.V. All rights reserved.

Date Submitted 2018-08-17 1Glacial and gully erosion on Mars: a terrestrial perspective

2Susan J. Conway1*

3Frances Butcher2

4Tjalling de Haas3,4

5Axel J. Deijns4

6Peter M. Grindrod5,6

7Joel Davis6

81. CNRS, UMR 6112 Laboratoire de Planétologie et Géodynamique, Université de Nantes,

9France.

102. School of Physical Sciences, Open University, Milton Keynes, MK7 6AA, UK.

113. Department of Geography, Durham University, South Road, Durham DH1 3LE, UK.

124. Faculty of Geoscience, Universiteit Utrecht, Heidelberglaan 2, 3584 CS Utrecht, The

13Netherlands.

145. Work done at: Centre for Planetary Sciences at UCL/Birkbeck, University of London,

15Malet Street, Bloomsbury, London WC1E 7HX, UK.

166. Now at: Department of Earth Sciences, The Natural History Museum, Cromwell Road,

17London SW7 5BD, UK.

18*corresponding author: [email protected]

1 19Abstract

20The mid-to-high latitudes of Mars host assemblages of landforms consistent with a receding

21glacial landscape on Earth. It is hypothesised that these landforms are a result of dramatic

22changes in climate brought about by swings in Mars’ orbital obliquity, which can vary

23between 15° and 35° on timescales of ~100,000 years. At the highest obliquities it is thought

24that water ice is driven off the two permanent polar caps and redistributed to lower latitudes,

25and as the obliquity swings to lower values water ice is transported in the opposite sense.

26Here, we report on the relationship in time and space of two suites of landforms: gullies and

27glacial landforms. Gullies are kilometre-scale erosion-deposition systems comprising a

28source alcove, transportation channel and deposition apron or fan. The glacial landforms we

29describe here fall into two categories – extant viscous flow features where ice could still be

30present and relicts of glaciation including arcuate ridges commonly interpreted as moraines.

31Both gullies and glacial landforms are particularly common at the mid-latitudes – hinting at a

32common climatological origin. Here, we measure headwall retreat associated with the glacial

33landforms and date the host-craters to constrain the retreat rates. Our analyses show that

34the phase of glaciation recorded by the youngest suite of glacial landforms had headwall

35retreat rates up to ~103 m Myr-1, equivalent to erosion rates of wet-based glaciers on Earth

36and to headwall retreat rates associated with extensive martian bedrock gully systems. In

37addition, we do not find evidence for repeated cycles of erosion, but rather a single episode

38possibly 5-10 Ma. Gullies seem to postdate this episode and have caused cyclical reworking

39of the terrain on the steepest slopes. The majority of the glacial ice preserved on Mars as

40Viscous Flow Features (VFF) pre-dates this erosional episode. Additional evidence that this

41accelerated rate of glacial erosion was facilitated by small quantities of basal pore water is

42provided by the presence of arcuate ridges consistent with terrestrial glaciotectonic features

43which require liquid water to form; textural alteration of the eroded bedrock surface

44consistent with ice-segregation and frost-shattering; and finally the presence of downslope

45pasted-on terrain, which could represent glacial deposits (till). The pasted-on terrain is

2 46usually considered as a thicker latitude dependant mantle located on sloping terrain formed

47from airfall of ice nucleated on dust, but our reinterpretation suggests the inclusion of more

48debris than previously assumed. Although our results cannot substantiate that gullies are

49produced by meltwater, the discovery of this “wet” glacial event does provide evidence for

50widespread meltwater generation in Mars’ recent history.

51

52Keywords: Mars; martian gullies; Viscous Flow Features; Glacier-like forms; Liquid water

3 531. Introduction

54The martian mid-latitudes are host to a suite of landforms that indicate significant

55geologically recent (10s – 100s Ma) surface-atmosphere exchanges of water ice. This study

56focuses on two of the most common landforms: martian gullies and glacier-like forms. We

57examine the role they have played in landscape evolution over the last ten to hundreds of

58millions of years by using statistical analysis of topographic data. In this introduction, we first

59present a brief overview of the state of knowledge concerning the present and past

60distribution of ice and related landforms on Mars, then specify how martian gullies fit into this

61context and finally present the scope of the present study.

621.1 The distribution of water ice on Mars

63Water ice is stable and exposed at the surface at the two polar caps of Mars, which each

64contain a volume of ice similar to the Greenland ice sheet on Earth – approximately 106 km3

65. Water vapour is contributed to the atmosphere by seasonal sublimation of the north polar

66cap, which has higher summer temperatures than the southern cap, because of its lower

67altitude and higher atmospheric pressure . It also has a larger part of the water ice cap

68exposed at the surface compared to the south, where the rest of the surface is partially

69hidden by a perennial, thin, CO2 ice layer and by surface debris . The water vapour

70contributed to the atmosphere is redistributed across the planet and can be deposited as

71surface frosts down to the mid-latitudes . Theoretical modelling predicts that ground ice on

72Mars should exist in diffusive equilibrium with the atmospheric water vapour – it should be

73cold trapped into the pores of the regolith . This idea is supported by observations from the

74Neutron and Gamma Ray Spectrometers on Mars Global Survey that found abundant

75hydrogen in the top metre of the regolith down to ~50° latitude in both hemispheres, which

76can be explained by an ice content of 4% to >64% in the regolith . This geophysical

77evidence is further supported by the observation of ubiquitous polygonally patterned ground

78in the same latitudinal band consistent with thermal contraction cracks formed in ice-

4 79cemented soil over annual timescales . A trench dug by the Phoenix lander at 68°N found

80both excess and pore ice centimetres below the surface , newly formed impact craters have

81exposed water ice in their ejecta at latitudes down to 39° N and exposures of almost pure

82ice in eight ~hundred metre scarps have been found at ~55° latitude .

83The latitude zone hosting ground ice is also an area that is smooth at scales of a kilometre ,

84as measured from elevation data from the Mars Orbiter Laser Altimeter (MOLA). This

85smoothing is partly attributed to the presence of the latitude dependant mantle (LDM) - a

86deposit thought to consist of the remnants of an airfall deposit of ice nucleated on dust,

87where the dust forms a sublimation lag protecting the remaining deposits from sublimation.

88Mustard et al. and Milliken et al. found that the LDM exhibited degraded or pitted textures at

89latitudes between 30 and 50° N and S; this change in texture coinciding with the change in

90surface roughness found from MOLA data . The surface age of the LDM has been estimated

91at ~0.1-5 Ma, and decreases in age with latitude . The LDM generally consists of alternating

92relatively ice- and dust-rich layers, indicating multiple generations of deposits formed under

93varying climatic conditions . There are various surface textures/draping deposits that have

94been grouped under the term “LDM”, but we will argue below that not all of these

95necessarily represent airfall dust-ice deposits. In total, LDM deposits cover 23% of the

96surface of Mars and are thought to contain 103-104 km3 of ice .

5 97

98Figure 1: Viscous flow features on Mars, north is up in all panels. (a) Lobate debris apron in

99Deuteronilus Mensae, CTX image F22_044466_2278. (b) Concentric crater fill in Utopia

100Planitia, CTX image P15_007067_2124. (c) Lineated valley fill in the Arabia Terra region,

101CTX image F06_038017_2202 (d) Glacier-like form in Protonilus Mensae, CTX image

102G03_019358_2225.

103Another significant reservoir of ice on Mars are the viscous flow features (VFF) found in the

104martian mid-latitudes . They have a total estimated volume of 4.2 x 105 km3, which is ~20%

105of the total volume of the polar caps . We will use VFF is an umbrella term for a range of

106landforms thought to be similar to debris covered glaciers on Earth. They include the

107following (illustrated in Figure 1):

108  LDA, lobate debris aprons – these are smooth aprons that extend from and

109 encompass mesas. They can extend several to tens of kilometres from their mesa

110 and orbital radar observations have confirmed that they can be up to 1 km thick and

6 111 contain >90% ice . Their ages range from ~40-500 Ma to ~1 Ga . LDA are the

112 largest reservoir of ice among the VFFs .

113  Crater interior ice-deposits, including concentric crater fill (CCF) – these deposits fill

114 and smooth-over the floors of impact craters and are ubiquitous in the mid-latitudes .

115 Levy et al. estimate that ice thicknesses can range up to 1.7 kilometres. The surface

116 of these fills often expresses lineations either concentric to the crater walls (in this

117 case they are called CCF), or can instead be oriented in one direction , and these

118 orientations vary with latitude (pole-facing 30-45° and concentric >45°). They have

119 been dated to be as young as 10 Ma and as old as 700 Ma . Crater interior ice-

120 deposits can cover part or all of the crater floor. Crater interior deposits are found in

121 craters 2-72 km in diameter , and represent the second largest reservoir of ice of the

122 VFFs.

123  LVF, lineated valley fill – these infill existing valleys, being particularly abundant near

124 the martian dichotomy boundary. Their longitudinal ridges are thought to represent

125 flow lines, or lateral compression from convergence of LDA on opposite valley walls .

126 Age estimates for LVF are similar to those for LDA; 100-500 Ma . LVF represent the

127 third largest reservoir of ice among the VFFs.

128  GLF, glacier-like-forms – these landforms are the most similar in scale and form to

129 terrestrial valley glaciers , they usually originate from large alcoves in escarpments

130 which can be on crater walls, valley walls or mesas, are laterally confined to a valley/

131 depression and can extend out onto plains. GLF are on average ~4 km long and ~1

132 km wide and the largest examples are ~35 km long . We also include in this category

133 lobate forms, such as those described in Milliken et al. , Berman et al. , and

134 Hartmann et al., which are not topographically confined in their source area. Due to

135 the lack of superposed craters these landforms are generally assumed to be younger

136 than other VFF at 10-100 Ma .

7 137Apart from occasional radar evidence for water ice under the debris cover of VFFs ,

138evidence for ice content is indirect and includes: presence of ablation textures (fractures,

139pitted surface, inverted/deformed impact craters, including “ring mold” craters) , lineated

140surface topography thought to represent flow lines , and depositional landforms (e.g.

141terminal moraines) (Figure 2). VFF are believed to behave like cold based glaciers on Earth,

142where the ice is frozen to the bed and the flow is dominated by deformation of mass above

143by gravity-driven viscous creep . Only in two locations have eskers been linked to such

144landforms, providing evidence of basal glacial melting . Rare supraglacial valleys have been

145attributed to transient supraglacial melting encouraged by focussing of solar radiation onto

146VFF surfaces from steep adjacent topography .

147The ice present in both the ground (including LDM) and in VFF is believed to be in diffusive

148equilibrium with the atmosphere, yet insufficient water vapour can be mobilised at the

149present day to explain the presence of these large ice masses. The formation of VFF and

150LDM has been attributed to increases in orbital obliquity, leading to transfer of large amounts

151of ice to the mid-latitudes . Mars’ present axial obliquity is similar to that of the Earth (25°),

152but in the past 5 Ma has oscillated between 15° and 30° (with a periodicity of ~100 ka) and

153from 5 to 10 Ma between 28° and 45° . Seasonal temperature asymmetries increase with

154obliquity: at high obliquity polar regions receive increased insolation and the north polar cap

155is believed to have been completely destabilised around 4 Ma when the obliquity was on

156average higher . The surface-atmosphere exchange in water ice and CO2 becomes more

157intense at higher obliquity and it is predicted that surface ice and ground ice can become

158stable even at equatorial latitudes.

1591.2 Gullies and their relation to ice on Mars

160The global distribution of martian gullies is spatially correlated with the distribution of water

161ice landforms described above. Gullies occur across the same range of latitudes as VFF, but

162they are less common where there are high concentrations of LDA and GLF and are only

163found in 12% of craters with interior ice deposits . They are strongly linked to the presence of

8 164LDM , although in the case of gullies the textures of this slope-side surficial deposit, into

165which they incise, are somewhat different to the LDM found in the plains; hence in this

166contribution we will refer to this unit using the term first coined by Christensen in relation to

167martian gullies: “pasted-on terrain”.

168

169Figure 2: Surface textures of Viscous Flow Features on Mars at scales of 1:15,000 (north is

170up in all panels). (a) Surface texture of crater interior ice-deposits showing surface lineations

171comprising alternating discontinuous ridges and chains of elongated pits in HiRISE image

172ESP_028651_1370. (b) Ring-mold craters on the surface of CFF in HiRISE image

173ESP_024319_1490. (c) A lobate debris apron in Deuteronilus Mensae showing typical

174contorted raised and pitted surface patterns in HiRISE image ESP_018515_2225. (d)

9 175Lineated valley fill in the Acheron Fossae Region with pitted and cracked textures and a

176deformed impact crater (the valley is oriented NW-SE) in HiRISE image ESP_016266_2165.

177Gullies are typically kilometre-scale landforms found on steep slopes in the mid- to high-

178latitudes of both hemispheres . They comprise a tributary source alcove, transport channel

179and depositional apron . They are found primarily on pole-facing slopes at latitudes between

18030° and 40° and then mostly on equator-facing slopes polewards of 40° (but they can also

181occur on pole-facing slopes in this latitude interval). The lack of gullies in regions with no

182evidence for ground ice (between 30°N and 30°S) and their trends in orientation with latitude

183has led authors to conclude that changes in orbital obliquity are required to explain their

184present-day distribution . However, the process by which gullies form is under strong debate.

185Present-day activity in gullies is thought to be brought about via slope instability caused by

186sublimation of carbon dioxide and the distribution of surface carbon dioxide ice is strongly

187controlled by the presence of (water) ground ice . However, it remains uncertain as to

188whether such a process alone can explain the formation of the whole landform. In particular,

189equator-facing slopes at the present-day and in the past are not favourable locations for

190deposition, and thus sublimation, of CO2 ice . Formation by liquid water is consistent with the

191morphology and morphometry of gullies , yet liquid water cannot explain the present-day

192activity and can only be invoked under certain specific conditions in the past . Hence,

193understanding the interaction of gullies with other landforms in time and space has the

194potential to break this impasse.

195Our knowledge on the timing of the activity and the overall age of gully-systems is relatively

196limited. We know that gullies on martian dunes have such frequent and voluminous activity

197that their full-extent can be explained on a timescale of hundreds of years . Dundas et al.

198estimate that mid-latitude gullies (not formed in sand, but in other materials including

199bedrock and mantling materials) could be formed within the last few millions of years given

200the currently observed magnitude and frequency of their activity. Non-sand gullies also have

201morphological evidence for multiple episodes of activity – in one case authors have

10 202ascertained the age of part of a gully-fan, where other parts of the fan both pre- and post-

203date the determined age of ~ 1.25 Ma . Dickson et al. found common evidence of relict

204gullies, being both exhumed from and buried by “LDM” (pasted-on terrain), showing that

205gully-activity must extend into epochs when this deposit was being formed/removed .

206De Haas et al. found that gully-alcoves had a similar size independent of the age of the host

207crater, and de Haas et al. proposed that gullies could be completely erased and their

208erosion “reset” during glacial epochs. In the youngest craters where gullies are found

209emanating from well-defined bedrock alcoves cut-back into the crater rim, de Haas et al.

210estimated headwall retreat rates of 10-4 to 10-1 mm yr-1, which correspond to the

211backweathering rates experienced on rockwalls in Arctic, Nordic and Alpine environments on

212Earth. De Haas et al. classed a crater as “glaciated” if it contained one or more of the

213following morphological features: arcuate ridges and/or spatulate depressions at the foot of

214the slope , floor-filling CCF indicated by concentric ridges and/or pitted textures within

215identifiable in-crater deposits , or crater-scale downslope lobate features which diverge

216around obstacles on the crater floor and have longitudinal or transverse ridges and/or pitted

217textures on their surfaces . Several authors have suggested that these glacial landforms

218represent different stages of ice-deposition and removal (glacial-interglacial cycles) during

219martian obliquity cycles. VFF were able to form during glacial epochs when average obliquity

220was higher than today (> 5 Ma). Since the last formation of VFF, LDM (pasted-on terrain)

221was deposited during smaller ice ages, which occurred during obliquity peaks in the present,

222lower-average obliquity regime . Many gullies that are present on Mars today have formed

223into VFF or LDM remnants , and thus their formation is strongly linked to VFF and LDM

224(pasted-on terrain).

225Both Berman et al. and Head et al. noted that the rims of craters with arcuate ridges and

226gullies on their pole-facing walls had a lower elevation than equator-facing walls of the same

227craters. The model proposed by de Haas et al. hypothesises that evidence of gully-erosion

228(alcoves and channels) is removed from crater walls by glacial erosion and/or buried by

11 229glacial deposits after sublimation/melting. Further, Levy et al. estimate that VFF can be

230responsible for 10-5-10-4 mm yr-1 of erosion. However, no study to date has attempted to

231directly assess the amount of erosion associated with the gullies and glacial landforms found

232on crater walls.

233

234Figure 3: Niquero Crater on Mars, an example with CCF, arcuate ridges, pasted-on terrain

235and gullies. North is up in all panels. (a) Overview of crater in CTX image

236P03_002383_1417. Arrows indicate the southernmost extent of the crater interior deposit.

237(b) Detail of gullies and arcuate ridges within a spatulate depression with the presence of

238pasted-on terrain labelled as “po”. HiRISE image ESP_030021_1410.

2391.3 Objectives and structure

240Here we use high resolution topographic data and dating of host craters to assess the

241amount and rate of erosion that these small-scale crater wall glaciers may have been

242responsible for and use analogy with glacial systems on Earth to assess by what

243mechanisms this erosion could have occurred. In our approach we use the consilience and

244coherence of the landforms and their spatial and topographic properties and relationships to

245go beyond analogy using the similarity of appearance of single landforms . We also compare

12 246to erosion rates in gullied craters without pasted-on terrain to assess whether martian

247glaciation is capable of erasing martian gullies.

248This paper is structured as follows. First, we explore the link between the pasted-on terrain

249(into which gullies often incise) and glaciation by undertaking a global survey and

250highlighting key relationships and observations. Second, we undertake a detailed

251topographic analysis of a small subset of these craters and of gullied craters to ascertain the

252amount and rate of erosion of their walls and its link to previous glaciation. Third, we

253consider the morphological and topographical evidence gathered on Mars in light of the

254present knowledge of glacial erosion on Earth in order to build a coherent picture of the

255degradation of crater walls on Mars. Finally, we bring together knowledge from the martian

256literature to assess the consistency of our new hypothesis within the context of previous

257findings.

2582. Global distribution and thickness of pasted-on terrain

2592.1 Approach

260We performed the majority of our morphological observations on NASA Mars

261Reconnaissance Orbiter High Resolution Imaging Science Experiment (MRO HiRISE)

262images at 25-50 cm/pix taken up to April 2017. We used the Harrison et al. gully-distribution

263database as the basis for our initial survey, down-selecting to HiRISE images that were in

264the vicinity of gullies. Within this database we noted the occurrence of arcuate ridges

265associated with gully-incisions and made observations on the nature of the pasted-on

266deposits associated with the gullies. We also searched for examples of craters without

267gullies, but with pasted-on terrain and/or arcuate ridges by examining craters < 10 km from

268the Robbins and Hynek database with (projection corrected) slopes of greater than 20°

269derived from the MOLA gridded data which had overlapping HiRISE images. We performed

270a random sample of this dataset and many craters that do not have gullies, but have pasted-

13 271on terrain and/or arcuate ridges will have been missed. Hence, this dataset was not used to

272derive any global distribution statistics, but to highlight pertinent relationships.

273We recognised arcuate ridges in HiRISE images using the following criteria: a sinuous to

274highly arcuate ridge located at the foot of the crater wall and somewhat parallel to the crater

275rim (Figure 4a-c). We identified pasted-on terrain as being a draping unit which smoothed

276the appearance of the crater wall at scales of ~1:25,000 with an upper lobate boundary

277(Figure 4 d-f). Gullies incised into this unit have a distinctive v-shape cross-section in their

278mid-section and these gullies are often completely sourced within this unit, having a simple

279single elongate incision . It is worth noting that not all gullies have such v-shaped incisions.

280We previously performed measurements on HiRISE digital terrain models (DTM) and

281determined that the slopes of these incisions were on average 20°, and ranged between 15°

282and 30° . Assuming a 20° slope, these incisions can be used as an estimate of the minimum

283depth of the pasted-on terrain simply by measuring the wall-length of the incision . We

284undertook a systematic survey of the incision wall-lengths in our HiRISE database in order to

285assess any trends in the thickness of pasted-on deposits with latitude. For craters hosting

286multiple gully systems, we measured the deepest incision, because the depth of this incision

287is closest to the total thickness of the pasted-on terrain.

14 288

289Figure 4: Recognition of arcuate ridges and pasted-on terrain in HiRISE images at 1:25,000,

290north is up in all panels, the downslope-direction is indicated by the black arrows and the

291scale for all panels is indicted in a. (a-c) Arcuate ridges in HiRISE images:

292ESP_038236_1410, ESP_023809_1415 and ESP_020051_1420. (d-f) Pasted-on terrain in

293HiRISE images: ESP_027477_2170, ESP_038236_1410 and ESP_013858_1405. Note in

294panel e the sharply incised gullies with v-shaped incisions to the left of the panel, where the

295right-hand one of the pair is completely within the pasted-on terrain.

2962.2 Pasted-on terrain

297The surface texture of the pasted-on terrain often appears smooth and uniform at scales of

298~1:25,000, yet at scales of ~1:5,000 or less we were able to identify three principal textures:

299downslope lineated (Figure 5a), smooth (Figure 5b) and polygonised (Figure 5c) . Polygonal

300textures were more often most clearly expressed within the gully incisions, as previously

301noted in the literature (Figure 5d). As noted by Conway and Balme , the lineations are cut by

302the gully-incisions (Figure 5e) and fresh incision walls have no polygonal textures and sharp

303breaks in slope, whereas incisions with polygonal textures are shallower and have a more

15 304subdued break in slope (Figure 5f). We found that the pasted-on terrain is systematically

305associated with a change in the appearance of the crater rim, compared to both the opposite

306crater rim (Figure 6f vs i) and compared to crater rims of fresh equatorial craters (Figure 6f

307vs d,g). Bedrock is often exposed in the top few hundred metres of crater walls, and is

308particularly evident in young impact craters (Figure 6d,g). It usually appears as a series of

309sub-parallel relatively massive beds, and particularly in fresh equatorial impact craters

310displays a “spur-and-gully” morphology with a series of regularly spaced alcoves. The rim

311crest of these craters remains relatively linear despite the presence of these alcoves (Figure

3126d,g). In comparison, the crater wall immediately above the upslope boundary of pasted-on

313terrain does not have a spur-and-gully morphology, but is rather planar with a distinctive

314“mottled” texture (we refer to this as “texturally altered bedrock” in the rest of this paper,

315Figure 6f). The rock exposure is rather patchy with no clear massive bedrock outcrop, and

316the surface often dominated by loose-appearing metre-scale boulders. Discontinuous rim-

317parallel lineations (Figure 6f) are sometimes apparent with no clear relationship to the

318underlying bedrock structure. Where texturally altered bedrock is present the crater rim has

319very little deviation from circular in planform (Figure 6c), and sometimes the crest is sharp

320while it can also be rounded (Figure 6j,k), particularly in older craters. This texture has been

321previously associated with erosion of crater walls (diameter < 35 km) by lobate debris aprons

322in Alba Mons by Sinha and Vijayan . This contrasts with the appearance of the rims of

323craters with gullies and no pasted-on terrain, which maintain the massive bedrock

324appearance, and the rim-trace is crenulated with the spur-and-gully morphology is greatly

325accentuated compared to unmodified crater walls as noted in de Haas (Figure 6b,e).

16 326

327Figure 5: Examples of pasted-on terrain textures, each image has the same scale as in (a)

328at 1:5,000. Black arrows indicate the downslope direction. (a) Lineated texture of the pasted-

329on terrain on the south-facing wall of Bunnik Crater, HiRISE image PSP_002514_1420. (b)

330Smooth texture of the pasted-on terrain, with rocks visible at the surface in HiRISE image

331ESP_033173_1400. (c) Polygonised texture of the pasted-on terrain in HiRISE image

332ESP_011672_1395. (d) lineated texture of the pasted-on terrain cut by gully incision in

333Bunnik Crater, where polygons are weakly visible in the sun-facing wall of the incision,

334HiRISE image PSP_002514_1420. (e) Polygons visible inside the gully incision (white

335arrows), but not on the surrounding pasted-on terrain in HiRISE image ESP_033173_1400.

336(f) Sharp gully incision whose walls are not polygonised cutting through the polygonised

337pasted-on terrain (white arrow). Polygonal patterns are again present on more gently sloping

338incisions. HiRISE image ESP_011672_1395.

17 339

340Figure 6: Rock exposures in fresh equatorial impact craters (left panels), gullied impact

341craters (middle panels) and craters with pasted-on terrain (right panels). The scale in a-c is

342given in panel (a) and is 1:120,000 and the scale in d-I is given in panel d and is 1:15,000.

18 343North is up in all panels. All craters are in the southern hemisphere. Red panels show the

344pole-facing slopes and blue panels show the equator-facing slope. (a,d,g) A fresh equtorial

345crater from the database of Tornabene et al. Kenge crater in CTX image

346B07_012315_1635 (a) and HiRISE image ESP_011893_1635 (d,g). (b,e,h) Galap Crater

347with gullies eroded into the bedrock in CTX image B09_012971_1421 (b) and HiRISE image

348PSP_003939_1420 (e,h). (c,f,i) Nybyen Crater with pasted-on terrain and gullies in CTX

349image G09_021563_1427 (c) and HiRISE image PSP_006663_1425 (f,i). Discontinuous

350ridges are highlighted by white arrows. (j,k) A crater in the Newton Basin with a planed-off

351rim in HiRISE image PSP_ 002620_1410 (j) and elevation profile (k) taken from the

352publically available DTM DTEEC_002620_1410_002686_1410_A01.

19 353

354Figure 7: Configuration of pasted-on terrain within craters on Mars. North is up in all panels.

355(a) Pasted-on terrain extending to within ~20-50 m of a crater rim in HiRISE image

356ESP_034363_1380. (b) Pasted-on terrain covering only the lower 1/2-1/3 of the inner crater

357wall in HiRISE image ESP_038157_1415. (c) Pasted-on terrain as remnants between gully

358systems (labelled by white “r”) in HiRISE image ESP_011672_1395. (d) Pasted-on terrain

359on the exterior wall of an impact crater in HiRISE image ESP_023809_1415. (e,f) Pasted-on

360terrain where only some of the bedrock has been texturally altered above it (extent indicated

361by arrowed line) and the rest still has alcoves. HiRISE images ESP_014400_1525 and

362PSP_009164_2140, respectively. (g,h) Pasted-on terrain extending upwards into wide

20 363alcoves on crater walls and leaving spurs of bedrock with an unmodified texture in g. HiRISE

364images PSP_006629_1425, and ESP_033398_1420, respectively. (i) Pasted-on terrain

365which appears to emerge from beneath mantle material (marked by “m”) with degradation

366textures as described by Mustard et al. and Schon et al. , HiRISE image

367ESP_011839_1460.

368The pasted-on terrain can be found extending nearly all the way up to the top of the crater

369rim (Figure 7a), or only on the lower half of the crater wall (Figure 7b), or as remnants

370between extensive gully systems (Figure 7c). Pasted-on terrain can be found on the interior

371(Figure 7a-c) and exterior crater walls (Figure 7d) and is almost exclusively associated with

372an altered texture of the upslope bedrock. In two rare circumstances texturally altered

373bedrock is found only on part of the wall with pasted-on terrain (illustrated in Figure 7f-i). The

374first of these circumstances involves very young craters (as attested by few superposing

375impact craters ~<1 Ma Table 1), where pasted-on terrain is found in the absence of any

376other evidence of glaciation (e.g., crater interior ice-deposits, arcuate ridges, large lobate

377tongues, glacial textures) and also occurs without gullies. In the two cases that we found

378(one in the northern and one in the southern hemisphere; Figure 7e,f) alcoves in texturally

379unaltered bedrock are present above some of the pasted-on terrain, but where the pasted-

380on terrain extends beyond the foot of the crater wall the alcoves are reduced and textural

381alteration has occurred. In the second circumstance pasted-on terrain can extend upwards

382into wide (hundreds of metres) alcoves leaving texturally unaltered bedrock on the spurs

383separating the alcoves, yet texturally altered bedrock upslope (Figure 7g). In other places

384where pasted-on terrain extends into alcoves the bedrock on both the alcove spurs and the

385rim-crest are texturally altered as shown in Figure 7h. In one example we found pasted-on

386terrain coexisting with terrain hosting textures normally associated with degrading latitude

387dependent mantle, namely exposed layers and extensive pitting . As shown in Figure 7i the

388pasted-on terrain seems to be emerging from beneath this mantle.

21 389Our global incision survey shows that the thickness of the pasted-on terrain tends to

390increase with increasing latitude in both hemispheres, from a mean of ~10 m at 28-30° to

39140 m at around 60°, with increasing scatter at higher latitudes (Figure 8). The only difference

392between the hemispheres in Figure 8 is that incisions are noted at lower latitudes in the

393southern hemisphere compared to the northern hemisphere (28°S vs 32°N). Incision depths

394are greater in the Argyre Basin than elsewhere (Figure 9a). Gullies with incisions follow the

395same spatial distribution as the gully population as a whole (Figure 9). Gullies with incisions

396form a larger percentage of the overall gully-population in the northern hemisphere

397compared to the southern hemisphere (~100 compared to ~50%, respectively; Figure 9).

398The pasted-on terrain, where present with other crater floor ice-deposits, gradually

399transitions in texture and topography into these other bodies, not presenting a single clean

400delimitation between these units (Figures 10g, 10l, 4c, 7b).

22 401

402Figure 8: Estimated gully incision depth against latitude, top: northern hemisphere and

403bottom: southern hemisphere. The solid lines represent mean values for each 5° of latitude.

404Dashed lines represent standard deviations for each 5° of latitude. The variable n is the

405number of data points within the graph.

23 406

407Figure 9: Global and latitudinal distribution of gullies, gully-incisions, crater interior ice-

408deposits and arcuate ridges. (a) Map of gullies and depth of incisions measured in this

409survey overlain on a relief-shaded rendering of the Mars Orbiter Laser Altimeter elevation

410data. (b) Summed CCF area per 5° latitude bin, mean gully density per latitude from Conway

411et al. which normalises gully-density based on the frequency of steep slopes found at that

412location, number of incisions as a percentage of the number of gullies in the same latitude

413bin and number of arcuate ridges as a percent of the number of incisions per latitude bin.

414Note the percent of incisions exceeds 100% because of the nature of the two surveys:

415Harrison et al. counted sites (in which many gullies could be present) and our incision

24 416survey counted individual gullies, hence there can be latitude bins where the number of

417incisions exceeds the number of sites. (c) Bar charts giving the counts of the features in (b)

418per latitude band. Bar graphs were made from supplementary data of Levy et al. and

419Harrison et al. .

4202.3 Arcuate ridges and crater interior ice-deposits

421Figure 9c shows the latitudinal distribution of the 148 arcuate ridges coexisting with incised

422gullies. Where ridges are associated with incised gullies they tend to occur at higher

423latitudes, a trend more obvious in the southern hemisphere. We find no difference between

424with thickness of pasted-on terrain in the systems with arcuate ridges and those without,

425even when considered by latitude (Figure 9).

426The arcuate ridges are often found in association with crater interior ice-deposits , which can

427form lobes of material that extend from the pole-facing crater wall beyond the ridges to the

428crater floor (Figures 3a, 10a-c,i). In both cases the arcuate ridges are commonly outlined by

429spatulate depressions which appear to push into the VFF. We found no evidence of arcuate

430ridges occurring without any kind of existing crater interior ice-deposit. We find that pasted-

431on terrain always occurs upslope of the arcuate ridges and, by association that arcuate

432ridges always coexist with texturally altered bedrock upslope (Figure 10).

25 433

434Figure 10: Relationship between pasted-on terrain, arcuate ridges and lobate or crater

435interior ice-deposits. North is up in all panels. (a) Arcuate ridges within a lobe of crater

436interior ice-deposits where the lobate margin appears to curve around obstacles. HiRISE

437images ESP_023809_1415 and ESP_024943_1415. (b) Arcuate ridges into crater interior

438ice-deposits which extend across the whole crater floor with flowlines extending towards the

439south in the southern hemisphere. HiRISE images ESP_016227_1405 and

26 440ESP_016438_1405. (c) Arcuate ridges into crater interior ice-deposits which extend across

441the whole crater floor with no flowlines, yet pitted surface texture similar to other VFF, in the

442northern hemisphere. HiRISE images ESP_013277_2155 and ESP_022059_2150. (d)

443Crater E overview in CTX image P19_008307_2138 overlain with HiRISE image

444PSP_009164_2140 showing the position of the cross-profile in (e). (f-h) Panels using

445HiRISE image PSP_009164_2140 at 1:10,000 showing: (f) the boundary between the

446pasted-on terrain and the crater floor, (g) the transition from the lineated pasted-on terrain

447texture to glacial texture mid-way down the crater wall and (h) the boundary between the

448lineated pasted-on terrain and the texturally altered bedrock. (i) Taltal Crater overview in

449CTX image G09_021712_1402 overlain by HiRISE image ESP_21712_1400, showing the

450location of the cross-profile in (j). (k-m) Panels using HiRISE image ESP_21712_1400 at

4511:10,000 showing: (k) the boundary between the pasted-on terrain and the arcuate ridges, (l)

452the transition between the polygonised pasted-on texture and VFF surface textures mid-way

453down the crater wall and (m) the boundary between the polygonised pasted-on terrain and

454the texturally altered bedrock.

455Although we did not survey for arcuate ridges which do not coexist with gullies (this is

456ongoing), Berman et al. systematically surveyed all the gullies and arcuate ridges in Newton

457Basin using Mars Orbiter Camera (MOC) narrow angle images at 1.5 m/pix and THEMIS

458(Thermal Emission Imaging System) visible images. They found 188 craters with gullies, 118

459craters with arcuate ridges and 104 craters with both, consistent with our finding that up to

460~55% of incised gullies also have associated arcuate ridges (Figure 9). Berman et al. found

461that for craters with diameters > 20 km in Arabia terra, Newton Basin and Eastern Hellas,

462arcuate ridges were almost exclusively associated with gullies. The arcuate ridges had an

463even stronger preference for pole-facing crater walls than gullies. Berman et al. and de

464Haas et al. noted a spatial correlation between gully-alcoves and the position of the most

465arcuate sections of the ridges: inflections between successive arcuate segments have

466similar widths as the alcoves located topographically above them; our observations agree.

27 467Moreover, the arcuate ridges below the largest alcoves also protrude furthest onto the crater

468floor. The arcuate ridges mapped by Berman et al. systematically occurred with “patterned

469floors” interpreted as ice-rich crater fill (here termed crater interior ice-deposits) and mantling

470materials. These crater interior ice-deposits sloped away from the margins of the fill proximal

471to the arcuate ridges, the slope-direction aligning with the lineations expressed within the

472surface of these fills, also noted by Head et al. .

4733. Erosion of crater walls

4743.1 Approach

475In order to assess the magnitude and rate of erosion associated with crater glaciation we

476made a series of topographic measurements and dated host craters via measurement of

477crater size-frequency distributions.

478The first criteria in our site selection was the existence of either a pre-existing digital terrain

479model (DTM), or a suitable stereo pair of images for creating a DTM. From these we

480selected three classes of craters:

481 1. Pristine impact craters in the databases of Watters et al. and Tornabene et al. ,

482 preferably located in the equatorial latitudes, but otherwise without evidence for any

483 of the following features: pasted-on terrain, other mantling deposits, arcuate ridges

484 or gullies.

485 2. Craters with gullies, but no evidence of pasted-on terrain, other mantling deposits, or

486 arcuate ridges. Preferably possessing HiRISE elevation data spanning north- and

487 south-facing walls.

488 3. Craters with pasted-on terrain which can have arcuate ridges, and/or gullies.

489 Preferably having HiRISE elevation data spanning north- and south-facing walls.

28 490In all cases we favoured craters with a simple rim morphology, lacking wall terraces and wall

491collapse features. This criteria and use of HiRISE images restricted our crater diameters to <

49220 km. We avoided craters located on antecedent relief (faults, other crater rims, valleys).

493

494Figure 11: Overview of approach for measuring recession rates of crater walls. (a) Talu

495Crater with elevation swath overlain on HiRISE image ESP_011672_1395 (blue colours are

496high elevations, red colours low). (b) Elevation data against distance from the rim-crest

497extracted from swath in (a) with blue-shades representing the raw data and black points

498showing the mean elevation every 1 m. (c) The same as in b, but restricted to 300 m either

499side of the crater rim. Red points are those used to perform a linear fit to obtain the slope on

500the inner wall (20-300m from the rim) and the green points are those used to perform a

501linear fit to obtain the slope on the exterior wall (20-100 m from the rim). Linear fits are

502shown as black lines. (d) Diagram illustrating the method by which we calculated the

503headwall retreat. The black line represents the crater wall for which we want to estimate the

504retreat with the green line representing the linear fit to obtain the slope β. The dotted grey

29 505line, either a pristine crater wall or an unmodified wall in the same crater with the blue line

506representing the linear fit to obtain the slope α. The orange distance l is the exposed length

507of bedrock as observed from the orthorectified images. The recession, r, is calculated as l - l

508(tan β / tan α).).

509Where existing HiRISE elevation data did not exist we used the freely available Ames Stereo

510Pipeline (ASP) to produce additional elevation data . We followed standard procedures in

511ISIS3 to produce single unprojected seamless images from the 9-10 individual HiRISE CCD

512images. These mosaicked images were map projected in ISIS3 to improve the point

513matching in ASP. We used standard settings to run ASP, with the Bayes EM subpixel mode

514with a 15 pixel subpixel kernel to improve results on steep terrain. We used ESA Mars

515Express High Resolution Stereo Camera elevation data to co-register and correct any tilting

516in the resulting point cloud with the routine “pc_align” and in one case we used a CTX DTM

517aligned with MOLA to co-register the data (as HRSC was not available). This CTX DTM was

518created using the same ASP procedure described above for HiRISE.

519In order to estimate the headwall retreat of the crater bedrock we measured the slope of the

520bedrock materials found just below the crest of the crater rim. We automatically extracted

521these data by taking 300-m-wide swath profiles perpendicular to the rim-trace. We used

522swath profiles in order to generate data robust to noise in the elevation data and whose

523slopes would be independent of the presence or absence of alcoves. For each swath profile

524we created an average profile, by binning the data by distance every metre from the crater

525rim and taking the mean of the points within each bin. We performed a linear fit of the

526elevation data between 20 and 300 m from the crater rim as an estimation of the upper

527bedrock slope of the inner crater wall and between 20 and 100 m on the exterior crater wall.

528This procedure is illustrated in Figure 11. We did not measure any cases where pasted-on

529terrain extended up to the crater rim, because no bedrock is exposed.

530We classified the swath profiles into four types depending on their association with identified

531landforms within each crater: 1) unmodified, 2) gullies present, but no evidence of glacial

30 532modification, 3) gullies present with evidence of glacial modification and 4) evidence of

533glacial modification, but no gullies.

534In order to convert the reduction in slope to a headwall retreat rate we first need to estimate

535the amount (length) of crater wall that has experienced this reduction in slope and then

536combine this recession length with the crater age to obtain a minimum headwall retreat rate

537for our studied craters (Figure 11.). We measured the distance from the crater rim over

538which bedrock was exposed to provide a minimum estimate of the horizontal length (l in Fig.

53911) of the crater wall that had undergone slope reduction. We calculated two slope-

540reduction quantities:

541 i. the slope reduction with respect to the opposing (unmodified) wall in the same crater

542 (if available) to give an estimate of the acceleration in erosion over the background

543 rate.

544 ii. the slope reduction with respect to the maximum average wall slope value of pristine

545 equatorial craters to give the cumulative erosion over the lifetime of the crater.

546To estimate the headwall retreat rate under the more likely scenario that periods of

547enhanced erosion punctuated the background headwall retreat rate, we followed Levy et al.

548and calculated the retreat rate over a time period of 0.5 Ma based on the slope-reduction

549given by (i). Levy et al. used a time interval of 0.5 Ma based on previous estimates for the

550duration of the last glacial epoch by Fassett et al. and the amount of time required,

551according to glacial flow modelling, for a typical LDA to form . For these relatively small

552systems this duration is probably too long, but provides a conservative estimate of erosion

553rate.

554We used Crater Tools and CraterStats extensions for ArcGIS to estimate the ages of those

555ejecta blankets of our study craters which did not already have published ages. In brief: we

556outlined the continuous impact ejecta on CTX images, which was then used as our crater

557count area. We counted all impact craters that had raised rims and were not covered/infilled

31 558by the ejecta. We fitted the Hartmann and Neukum isochrons to the resulting cumulative

559size-frequency distributions to obtain an estimated age and error. The plots can be found in

560Figure 12. These ages are taken as maximum values for the age of the features found within

561the craters, because the features formed after the impact.

32 562

563Figure 12: Crater size-frequency distribution for the eight previously undated craters in our

564study (Table 1). (a) unnamed crater “Crater I” where CTX image G22_026861_1557 was

33 565used to count the craters. (b) Unnamed crater “Crater E” where CTX images

566P21_009164_2137, P15_006949_2150, P19_008307_2138 and P21_009098_2138 were

567used to count the craters. (c) Unnamed crater “Crater H” where CTX image

568G06_020539_2114 was used to count the craters. (d) Unnamed crater “Crater D” where

569CTX images P12_005798_1396, P07_003675_1391, G14_023612_1375 and

570D10_031102_1378 were used to count the craters. (e) Nybyen crater whereCTX image

571B05_011436_1427 was used to count the craters. (f) Niquero crater where CTX images

572F01_036258_1410, P03_002383_1417 and P11_005284_1419 were used to count the

573craters. (g) Unnamed crater “Crater F” where CTX image G14_023662_1960 was used to

574count the craters. (h) Unnamed crater “Crater G” where CTX image F23_044780_1635 was

575used to count the craters.

5763.2 Headwall retreat rate of glaciers and gullies

577We used a dataset comprising 2 gullied craters, 7 craters with arcuate ridges, 1 crater with

578pasted-on terrain yet no arcuate ridges, 6 fresh craters and one degraded-looking equatorial

579crater with no gullies, pasted-on terrain or arcuate ridges (Table 1, Figure 13). These craters

580have diameters ranging between 2.5 and 20 km.

581

34 582Figure 13: Location map of the DTMs used in this study (red dots) and locations of the

583additional sites used in the figures (green dots) on a MOLA elevation shaded relief base

584map. The numbers/letters in brackets identify the relevant figure(s).

585We found that the upper wall slopes are ~39° for the youngest equatorial craters (the

586average for all six) and in our oldest equatorial example (Kilmia, 1.1 Ga) the upper wall slope

587has reduced to 26° (Table 2 and Figures 14, 15). In the terrestrial literature it has been found

588that a peak in slope exists at a similar value of ~39° in mountain ranges, which is thought to

589represent the strength of the bedrock . Similarly for the youngest craters in our modified

590sample (0.5-6.5 Ma Crater A, B, C, Istok, Jaisalmer, Galap) the slopes of the walls

591unmodified by either gullies or glacial processes have very similar high slope values (Figures

59214, 15). The unmodified walls of older craters (1.3-1.4 Ga, Niquero, Corozal,) have lower

593slopes ~32-34°.

35 594

595Figure 14: Boxplots displaying results from our crater wall slope analysis. Blue boxes

596indicate the interquartile range, whiskers the maximum and minimum values, black dots the

597median value and blue dots outliers (classed as those values further than 1.5 interquartile-

598ranges from the median). (a) Boxplots of slope data gathered for the inner crater wall of our

36 599studied craters. Dotted line is arbitrarily located at 38° for reference. (b) Boxplots of slope

600data gathered for the exterior crater wall of our studied craters. Dotted line is arbitrarily

601located at 15° for reference. (c) The same as in (a) for Crater E except the data are split into

602one further class – “pasted-on” refers to parts of the crater wall with pasted-on terrain yet

603little textural alteration of the bedrock, whereas “glacial” refers to areas with pasted-on

604terrain and textural alteration of the bedrock (Figure 7f).

605The reduction in interior crater wall bedrock slope in the two gullied yet unglaciated craters is

6064-5° compared to their unmodified crater walls, and 6-9° compared to slopes in fresh

607equatorial craters (Figures 14, 15; Table 2). The interior walls of craters that have evidence

608of glaciation are further reduced: up to 8° (on average 5°) lower compared to conjugate

609unmodified walls and up to 21° (on average 14°) compared to fresh equatorial craters. We

610also found that when pasted-on terrain was present on exterior crater walls, these bedrock

611slopes were reduced compared to unmodified exterior crater walls (Figures 14,15; Table 2),

612although the signal is not as strong as for the inner walls. This is probably because there is a

613larger initial variation in the slope of exterior crater walls, as shown by their variability in fresh

614equatorial craters (a range of 8° to 17° with standard deviations of 3-10°, Table 2). Finally,

615for Jaisalmer crater (the youngest glaciated crater) ANDCraterB? we noted that wall sectors

616where there was pasted-on terrain and texturally altered bedrock had lower slopes than rim

617portions with only pasted-on terrain and both were lower than the unmodified portions of the

618crater wall (Figure 14c)

619For the gullied craters the bedrock extends on average 390-530 m down the crater wall

620(Table 2); therefore, the crater wall has receded by 70-140 m at the top. For Galap this

621results in an estimated headwall retreat rate of 96 m Myr-1 and for Istok 794 m Myr-1,

622compared to 4.5 m Myr-1 and 75 m Myr-1 backweathering rates estimated by de Haas et al.

623using an independent method.

624The slopes within glaciated craters are partially covered by pasted-on terrain, hence as a

625conservative estimate we can assume that only the bedrock that is exposed has been

37 626affected by the slope reduction. The pasted-on terrain is found between 44 and 1188 m and

627on average 369 m distance from the rim in our studied craters (Table 2). For comparability

628we use the same slope-lengths to calculate the headwall retreat for the glaciated and the

629unmodified walls of a given crater. We find that headwall retreat rates range between 0.02

630and 151 m Myr-1 for unmodified walls of glaciated and gullied craters and overlap with the

631values of 0.16 to 121 m Myr-1 for completely unmodified craters. We further find that the

632headwall retreat rates of glaciated walls fall between 0.09 and 1009 m Myr-1. When these

633headwall retreat rates are plotted against crater age (Figure 15e), a decreasing trend in

634headwall retreat rate against age is observed. Similarly for the glaciated crater walls, the

635calculated headwall retreat rate has a similar decrease with crater age, but transposed to

636higher headwall retreat rates than the unmodified walls. Calculating the headwall retreat rate

637for the glaciated crater walls over a fixed time interval of 0.5 Ma (as detailed in Section 3.1)

638and using the slope reduction compared to the unmodified wall of the same crater (rather

639than compared to a fresh equatorial crater wall), we find that headwall retreat rate has no

640noticeable trend with age (Figure 15f).

641We briefly note here that our data do not show that glaciated inner crater walls have

642significantly and/or systematically lower elevation than unmodified inner walls in the same

643crater as observed previously by both Berman et al. and Head et al. . We posit that this

644could be due to the fact that we find erosion on the inner and outer crater walls, hence both

645north- and south-facing walls are being lowered by the action of glaciation.

38 646

647Figure 15: Wall slopes, reduction in wall slope and headwall retreat rate for studied craters.

648(a) The slope of exposed bedrock in the unmodified walls of craters in this study, including

649craters with neither glacial nor gully-features (unmodified), craters with gullies but no glacial

650features (gullies) and craters with glacial features (glacial features). (b) The slope difference

651between glaciated walls and unmodified walls from the same crater vs the unmodified wall

39 652slopes for outer and inner crater walls. (c-d) The slope difference against age and

653unmodified wall slope, respectively. The legend for both plots is in in panel c. (e) Estimated

654headwall retreat rate against age for our craters separated into unmodified crater walls and

655those affected by glacial erosion. Plotted for comparison are: erosion rates from modelled

656glacial landscapes in New Zealand and measured glacial erosion from Iceland. (f) The same

657plot as in panel e, but the headwall retreat rate is calculated over a fixed 500 ka interval and

658uses the slope difference compared to the unmodified wall in the same crater (rather than

659comparison to an equatorial crater wall slope). NB: The error bars for slope represent the

660standard deviation of the slope measurements and age error bars the potential age range

661for the host-craters. The uncertainty in slope difference is calculated by propagation of errors

662and the uncertainty in retreat rate is dominated by the length of exposed bedrock, hence

663uncertainties are calculated by recalculating the retreat rate using the minimum and

664maximum length of the exposed bedrock for that crater.

6654. Glacial erosion

6664.1 Mechanisms of glacier erosion on Earth

667Mechanical erosion by glaciers is dominated by abrasion, quarrying, plucking and

668glaciotectonism . Abrasion is the process by which the glacier bed and entrained clasts are

669scoured either by debris entrained within basal glacier ice or, less commonly, by the basal

670ice itself . Quarrying is the process by which bed-clast contact, or overriding of bedrock

671cavities, generates foci of pressure at the ice-bed interface and liberates fragments from the

672bed . Plucking involves freezing of meltwater within, or deformation of basal ice into, bedrock

673fractures, and prizing-off of fragments under subsequent glacier motion . Finally,

674glaciotectonism is the process by which subglacial, submarginal and/or proglacial materials

675deform under stresses induced by glacial ice .

676Glacier thermal regime exerts a fundamental control upon the efficacy of glacial erosion on

677Earth. This arises from its influence on the generation of meltwater, entrainment of erosional

40 678‘tools’ (i.e. debris), and dynamics of ice-bed interactions . The thermal regimes of glaciers

679are categorised according to their temperature relative to the pressure melting point of ice,

680and are controlled by complex interactions between climatic, environmental and glaciological

681parameters . Temperate (warm-based) glaciers are at the pressure melting point of ice

682throughout, whereas cold-based glaciers are entirely below the pressure melting point.

683Polythermal glaciers , where ‘warm’ ice at the pressure melting point coexists with ‘cold’ ice

684below the pressure melting point, represent an intermediate condition between temperate

685and cold-based glaciers.

686The efficiency of mechanical erosion by glaciers is greatly enhanced in the presence of

687meltwater. Under a wet-based thermal regime, significant basal sliding can occur at an

688unfrozen ice-bed interface, promoting efficient abrasion of the bed and liberation of rock

689fragments via plucking. Sliding over obstacles at the bed also opens lee-side cavities,

690promoting quarrying . Quarrying is particularly effective under wet-based regimes because

691meltwater, which is dynamic on shorter timescales than glacial ice, can greatly enhance

692pressure fluctuations at the bed . Liquid water is thought to play an important, possibly

693essential, role in promoting glaciotectonic deformation. Pore water reduces the yield stress

694of glacial sediments, making them more susceptible to deformation under stresses induced

695by glacial ice . Despite generally lower meltwater volumes associated with glaciers in cold-

696climate or permafrost regions, ground ice and/or frozen glacier margins may encourage

697glaciotectonic processes by preventing efficient drainage of meltwater from aquifers which

698they confine .

699Given the frozen ice-bed interface and absence of meltwater in cold-based glacial systems,

700it is commonly assumed that cold-based glaciers flow entirely by internal deformation, do not

701erode their beds, and exert little or no detectable geomorphic influence upon the underlying

702landscape. Indeed coverage by cold-based ice is frequently invoked as a protective

703mechanism to explain the preservation of features generated by previous wet-based

704glaciations . However, theoretical and field-based studies have noted that sliding rates of

41 705cold-based glaciers are, in fact, non-zero when integrated over long timescales . Field

706observations have also identified evidence for reworking of cold glacier beds , such that it is

707becoming increasingly evident that cold-based glaciers can exert significant geomorphic

708influence over timescales of glacial advance, although still substantially less than warm-

709based glaciers.

710In examining isotopic composition of dirty basal ice layers derived from marginal apron

711overriding at Meserve glacier in the Antarctic Dry Valleys, Cuffey et al. suggested that

712interstitial water films between ice and immersed solids within dirty basal ice layers of cold-

713based glaciers may permit sliding and abrasion down to temperatures of -30°C, an effect

714that could be particularly important in the presence of a highly saline substrate. Additionally,

715Lloyd Davies et al. proposed a mechanical model by which abrasion and quarrying can

716operate beneath cold-based glaciers in the absence of liquid water, based on field

717observations in the Allan Hills region of Antarctica. Ice-marginal aprons comprising collapsed

718ice blocks commonly accumulate at the foot of steep terminal ice cliffs of cold-based

719glaciers. These aprons can be incorporated into basal ice as they are overridden during

720glacier advance . Lloyd Davies et al. suggest that this incorporates a weak, low-density

721layer into the submarginal basal ice which focusses stress onto up-glacier bedrock contacts

722and promotes fracturing and quarrying of bedrock protruberences. Entrained rocks are then

723available as ‘tools’ for abrasion during subsequent glacier motion .

7244.2 Landscapes of glacial erosion and application to Mars

725We now consider the potential contributions of these erosional processes to the following

726geomorphic features that we observe on Mars: texturally altered bedrock of upper walls of

727impact craters, lowering of slopes on texturally-altered crater walls compared to fresh crater

728walls, instances of sharp crater rim crests despite lowered wall slopes, and associated

729slope-side pasted-on terrain and arcuate ridges which are commonly, but not ubiquitously

730associated with spatulate depressions within crater floor CCF materials.

42 731Many geomorphic features that are indicative of past glacial erosion, including striae,

732gouges, fractures, scrapes and chattermarks , are undetectable at the decametre- to metre-

733scale resolutions of existing orbital remote sensing datasets for either Earth or Mars.

734However, the integrated effect of these processes operating over large areas typically

735manifests as smoothed, rounded, planed-off, or ‘areally-scoured’ surfaces . Such landscapes

736bear similarities to the smoothed morphologies of the walls of the glaciated craters in our

737survey, within which the upper bedrock protruberances (‘spur and gully’ forms) have been

738planed-off, and possibly old gully alcoves strongly modified . If the textural disruption of

739these slopes is a primary morphology inherited directly from glacial erosion, it seems more

740consistent with erosion via quarrying and plucking, which can generate rough, ‘craggy’

741surfaces , than by abrasion which tends to smooth and polish bedrock surfaces (FIGURE).

742Highly brecciated and fractured bedrock associated with impact crater walls (de Haas et al.,

7432015a) may make these bedrock surfaces particularly susceptible to erosion by quarrying

744and plucking, as in tectonically-active glaciated regions on Earth . Further, the amount of

745slope lowering we observe is independent of the estimated initial slope, which is consistent

746with glacial erosion that depends on the ice-surface-slope rather than the bed-slope.

747NEW FIGURE 16?

748Another possible contributor to the textural alteration of the bedrock is enhanced weathering

749under the ice or snowpack at the top of the slope. In a terrestrial setting rock weathering in

750cold regions is controlled by moisture availability and by the existence of the thermal

751conditions required for ice segregation. These conditions are enhanced by the presence of

752snowpatches or snowfields , as ice lens growth during ice segregation is favoured by slow

753freezing rates and sustained below-freezing temperatures . Ice lens formation is the primary

754cause of physical rock breakdown or frost-shattering . These physical weathering

755mechanisms require the presence of thin films of liquid water, yet ice segregation on Mars

756has been hypothesised to occur without need for the liquid phase .

43 757The observed lowering of crater wall slopes presents a more complex challenge for the

758glacial hypothesis. Glaciers tend to steepen upper valley slopes, both in longitudinal and

759cross-sectional profile, giving rise to classic u-shaped valley and cirque landscapes. This

760seems inconsistent with our observations of crater walls, where slopes are lowered and

761there is an absence of pronounced u-shaped undulations or bedrock-confined

762overdeepenings. However, Hirano and Ania and Harbor found topographic steepening to

763be an outcome of enhancement of vertical erosion by topographic confinement of glaciers,

764and its effect on ice flow. Glaciers that are not topographically confined can preferentially

765widen rather than deepen their host valleys, resulting in broad-scale lowering of slopes . We

766therefore suggest that valley and cirque-style glacial configurations are inconsistent with our

767observations, and that unconfined, piedmont-style (LDA-like) glaciers may provide the

768optimum explanation for the observed morphologies. The location of pasted-on terrain in

769wide alcoves (which could be interpreted to resemble cirques), does not necessarily imply

770that the ice was limited to those alcoves. Weak vertical erosion could explain the presence of

771distinct rim and inter-alcove crests in some instances, despite clear evidence that these

772same rim segments have undergone significant lowering of slope.

773However, the present understanding of glacial slope-modification on Earth is largely based

774on assumptions that glaciers inherit pre-existing fluvial valleys with gentle slopes and have a

775ready supply of meltwater to their beds . These assumptions conflict significantly with both

776the present understanding of recent glacier thermal regimes on Mars (largely cold-based)

777and the impact crater wall landscapes explored here, which have steep initial slopes and

778lack mature fluvial valleys. Therefore, we emphasise that constraining the configuration of

779glacial ice at small spatial scales would be highly speculative given existing data and the

780distinctly atypical initial topography compared to terrestrial case studies upon which glacial

781theory is based.

782Given the magnitude of lowering of crater wall slopes, it is clear that significant volumes of

783material have been mobilized by crater wall erosion. Within the surveyed craters, ‘pasted-on’

44 784terrain was invariably located downslope of texturally-altered bedrock. Accordingly, we

785propose that upper crater walls and rims may have provided a viable source of debris for this

786‘pasted-on’ material, and that glaciers could have acted as important erosional and

787depositional agents in its formation. In this context, the origin for the downslope lineations

788within pasted-on terrain could be streamlined glacial bedforms such as flutes or megascale

789glacial lineations. However, the pasted-on terrain is significantly more extensive and

790voluminous than rare, patchy, and thin (<2 m-thick) till deposits identified in association with

791cold-based glaciers on Earth . If the pasted-on terrain on formerly glaciated crater walls is

792indeed glacial till, it would seem necessary to invoke at least some degree of meltwater

793activity, contrary to the idea that Amazonian glaciation was almost entirely cold-based . We

794therefore emphasise that the origin of the pasted-on terrain should be investigated further.

795Given that aeolian processes are widespread and active geomorphic agents on Mars in the

796present day, and have been shown to modify gullied-slopes , an aeolian origin for the

797downslope lineations within pasted-on terrain might be just as likely as an origin as

798streamlined glacial bedforms such as flutes or megascale glacial lineations. Therefore,

799convergence of form between morphologies generated by glacial and aeolian streamlining

800prevents conclusive distinction between these formation mechanisms given the resolution of

801HiRISE images. Thus, their potential origin as streamlined glacial bedforms should not be

802excluded considering the apparent close association between pasted-on terrain and

803glaciated slopes.

804

45 805Figure 16: Examples of sinuous arcuate ridges with and without associated upslope

806alcoves. (a) HiRISE image ESP_038236_1410 where loops in the arcuate ridges appear to

807be associated with upslope alcoves of a similar width. (b) Avire Crater in HiRISE image

808ESP_029467_1390 with a similarly sinuous arcuate ridge to (a), but without any obvious

809upslope alcoves.

810Prominent arcuate ridges provide particularly convincing evidence for past glaciation of

811crater walls. Although hypotheses that such arcuate ridges represent protalus ramparts have

812been proposed , they are widely interpreted as end moraines based on their similarity to

813moraine ridges on Earth and to arcuate ridges bounding ice-rich glacier-like forms on Mars .

814Typically, multiple arcuate ridges coexist, forming laterally extensive moraine complexes.

815Although some arcuate ridge complexes can clearly be associated with bedrock alcoves in

816the crater walls above (Figure 16a) , many arcuate ridges lack distinct association to discrete

817undulations in the crater walls (Figure 16b). Thus, we consider the ridges to be more

818consistent with unconfined piedmont-style glaciation, as proposed above, than with the

819formation of each arcuate section at the terminus of a discrete glacier tongue. The highly

820curvilinear morphologies of the ridge complexes are not inconsistent with such an origin,

821since differential flow within broad ice lobes on Earth commonly forms highly curvilinear

822moraine complexes (Figure 17b) .

823The common association between arcuate ridges and spatulate depressions within crater

824floor CCF materials indicates that ice-contact bulldozing of CCF materials contributes to

825their formation . However, several features lead us to support that glaciotectonic processes

826also provided a significant contribution to their formation, as was considered by Arfstrom and

827Hartmann . The distinction between bulldozing and glacitectonic processes is in the nature of

828ice contact; while bulldozing relates to movement of material in contact with glacial ice,

829glacitectonism can influence materials tens to hundreds of metres below glacier beds and

830into the proglacial zone . Compressive stresses induced by convergence of crater wall

831glaciers with pre-existing crater floor ice-bodies, such as CCF (e.g., Figure 10), would

46 832enhance the likelihood that glacitectonism supplemented bulldozing in moraine construction

833where these pre-existing bodies are present. Furthermore, bulldozing of crater floor

834materials alone cannot necessarily provide adequate explanation for apparent disruption

835within CCF surfaces beyond the arcuate ridges (Figure 10). Thus, complex stress regimes

836influencing submarginal and proglacial materials may also be required to explain the full

837suite of arcuate ridge and CCF-modification features that we observe.

838Landforms associated with glaciotectonic deformation of material at or below glacier margins

839on Earth include thrust block moraine ridges , hill-hole pairs , and rafted megablocks . We

840have identified in our study possible examples of all three landform types, shown in Figure

84118. Longitudinal ridges within CCF surfaces (Figure 18), proximal to the spatulate

842depressions, could represent faulting of proglacial materials under stresses induced by

843crater wall glaciation (Figures 17a, 18a).

844

845Figure 17. Examples of arcuate glaciotectonic landforms on Earth. (a) Oblique view of

846active thrusts within proglacial sediments of a piedmont glacier, Canada. Taken from

847Evans . No scale was provided in the original publication, but we estimate the valley floor to

47 848be 1.5 km wide. (b) Oblique view of small seasonal glaciotectonic moraines at

849Brieðamerkurjökull, Iceland illustrating highly arcuate nature of some moraine complexes in

850front of a broad glacier terminus. Taken from Bennett . No scale was provided in the original

851publication, but such seasonal ridges are typically metres in width. (c) A thrust-block

852moraine comprising lacustrine sediments at the terminus of Wright Lower Glacier, Antarctic

853Dry Valleys. The ridge is ~80m wide at its widest point and North is towards the top. Digital

854Earth image from Google Earth. (d) Marine bathymetry showing an arcuate hill-hole pair on

855the Norwegian continental shelf. Taken from Rise et al. . 856

857Figure 18. Possible examples of glaciotectonic landforms beyond the foot of the glaciated

858crater wall, possibly analogous to the proglacial thrust block moraines in Figure 17a. HiRISE

859image ESP_020051_1420. (b) Example of lower crater wall ‘hole’ which truncates gully fans

860upslope of arcuate ridge (‘hill’) at CCF margins. This may be analogous to hill-hole pairs on

861Earth (e.g. Figure 17d). HiRISE image PSP_003253_1405. (c) Possible example of a rafted

862megablock at the foot of a slope within the rim of the Argyre impact basin. Note that this

863feature is not within an impact crater included in our slope measurement survey. However, it

864provides the best type example of similar features observed within impact craters and

865illustrates the same relationships between texturally altered bedrock, pasted-on terrain and

866arcuate ridges. The bright lobate plateau appears to have been mobilized downlope along

867shear planes that are visible within upslope lateral extensions of the block. Its plateau

868morphology, gently-sloping upslope margin, steep downslope margin and clear source

48 869region leads us to propose that it represents a rafted megablock mobilised by stresses

870induced below the margin of a glacier. It does not appear to have been distorted or

871reworked. Hence we favour a raft origin over a hill-hole origin. An alternate interpretation is

872that it represents an extant lobe of glacier ice. CTX image G11_022685_1402.

873Cold-based glaciers on Earth are rarely associated with prominent moraine ridges at their

874margins. However, despite sub-freezing temperatures and deep, continuous permafrost in

875the Antarctic Dry Valleys, Fitzsimons described several arcuate, sharp-crested, ice-cored

876moraine ridges at the entry points of glaciers here into saline proglacial lakes (Figure 17c).

877Salinity of pore water exerts a critical control upon the deformability of sediments and

878glacitectonic moraine formation at the cold margins of glaciers in permafrost environments

879on Earth . Thus, while we do not invoke saline proglacial lakes such as those in the Antarctic

880Dry Valleys to explain the origin of the arcuate ridges on Mars, we do suggest that salinity of

881the Martian substrate may play an essential role in generating deformable glacier beds

882despite sub-freezing temperatures. Salts, such as perchlorates and sulphates capable of

883freezing-point depression are known to be abundant on the surface in-situ investigations

884and orbital spectral observations as well as from theoretical considerations of surface

885geochemistry . There are a few observations of basal sliding of cold-based glaciers due to

886small interstitial meltwater films between ice crystals and basal impurities. So basal sliding is

887negligible on small timescales but can be significant over very long timescales.

888Although recent studies have identified compelling evidence for rare, localised occurrences

889of past basal melting of glaciers on Mars, both in relation to existing VFF and past late

890Amazonian-epoch glaciation , a majority of Amazonian-aged glacial assemblages on Mars

891provide little apparent evidence for the role of meltwater . Fassett et al. attributed rare

892valleys on VFF surfaces to localised melting due to reflection of solar insolation by adjacent

893topography, rather than melting induced by bulk VFF thermal regime or climate. Hence,

894Amazonian glaciation on Mars was likely predominantly cold-based and the closest

49 895terrestrial analogues to existing VFF are thought to be cold-based debris-covered glaciers in

896the Antarctic Dry Valleys .

897In summary, the landscape assemblage described here provides the first evidence that liquid

898water could indeed have played a role in late Amazonian glaciation. Although the quantities

899were likely relatively limited, the geographical distribution was widespread: pasted-on terrain

900and texturally altered bedrock are found on almost every sloping surface in the mid-latitudes.

901We do not support the production of large quantities of meltwater as this would have

902obliterated the arcuate ridges and even the pasted-on terrain, and produced a suite of

903landforms akin to wet-based glaciers on Earth (e.g., eskers, meltwater channels, hummocky

904moraines).

9054.3 Glaciation and erosion rates

906Comparisons of sediment fluxes from glaciated and non-glaciated basins by Hallet et al.

907found that, on Earth, glaciation is ‘unsurpassed’ in its ability to erode and mobilize sediment

908compared to other erosional mechanisms operating over similar timescales. They attribute

909comparable sediment yields from lowland rivers draining marine clays in non-glaciated

910basins, to high-susceptibility of these clays to mass-wasting and, in particular, gullying .

911Based on the discussion above, it is clear that even cold-based glaciers can erode their

912beds, but at rates that are orders of magnitude lower than erosion by wet-based glaciation.

913Cold-based polar glaciers typically erode their beds at rates of ~101 m Myr-1 , compared to

914rates of ~103 m Myr-1 for small temperate valley glaciers in the Swiss Alps, and 104-105 m

915Myr-1 beneath large, fast-flowing temperate glaciers in southeast Alaska . However, when

916integrated over long timescales, the cumulative geomorphic influence of cold-based

917glaciation may be detectable; Cuffey et al., estimate that 10 to 30 m of erosion may have

918occurred in the u-shaped valley occupied by Meserve glacier on Earth, over a period of ~10

919Myr of cold-based glaciation. Rates of glacial erosion are greatly enhanced in the presence

920of weak (e.g. sedimentary or highly-fractured) bedrock, particularly in tectonically-active

921regions such as southeast Alaska and the Himalaya .

50 922The fastest retreat rates found in this study of ~103 m Myr-1 (Table 2; Figure 15) exceed the

923expected erosion rate on Earth for cold-based glaciers and are instead equivalent to erosion

924rates for temperate valley glaciers where melt is abundant. Similar erosion rates for icecaps

925and glaciers are predicted over 0.1 Ma timescales on Earth for newly constructed volcanic

926provinces or actively uplifting areas . Our highest erosion rate is above that estimated by

927Levy et al. for VFF - namely 10-2-101 m Myr-1 of erosion. In our study headwall retreat rates

928<101 m Myr-1 are found for craters with ages >10 Ma and these craters have similar headwall

929retreat rates to those estimated by Levy et al. . For comparison crater erosion rates by

930aeolian processes in equatorial regions of Mars are estimated to be on the order of 100-101

931m Myr-1 over 0.5-2 Ma timescales , decreasing with increasing temporal baseline (Figure 15).

932The decrease in erosion rate over time reflects the diffusive nature of the dry denudation

933processes, and the diffusivity is 102-103 times lower than terrestrial values and similar to that

934on the Moon . The decrease in erosion with time predicted by the diffusion model presented

935by Golombek et al. starts at a lower erosion value, probably because their model was tuned

936for smaller craters (~100 m) with a starting wall slope of ~30°, whereas our craters are

937kilometres in diameter and bedrock exposures can reach ~40°. However, it is interesting to

938note that both the glaciated craters and unmodified craters follow a similar trend with

939headwall retreat rate over time, yet displaced to a higher magnitude.

940One way in which to explain the parallel trends is that there has only been a single (recent)

941event of intense glacial erosion related to the presence of the pasted-on terrain. This does

942not preclude the possibility that previous erosion events have occurred and are either: too

943ancient to be distinguishable from the diffusive decline in crater bedrock outcrop slopes, or

944occurred earlier than our studied time-frame and been overprinted by this later event – a

945notion supported by the slightly elevated erosion rates expressed by craters older than 1 Ga

946in Figure 15f. As shown in Figure 15b, the erosion intensity seems independent of slope –

947which is not the case for a diffusion-like processes which tends to diminish in intensity with

948decreasing slope – hence it seems improbable that the glaciated craters would follow this

51 949trend unless it was imposed by the ongoing background crater degradation. If punctuated

950accelerated erosion by glaciers was happening throughout a crater’s history we would

951expect the glaciated crater walls and unmodified glacier walls to diverge with crater age on

952Figure 15e and the glaciated crater walls to have increasing erosion rates with time in Figure

95315f, which is not the case. If we assume a single 500 kyr event of glaciation we see that the

954glacial retreat rates become independent of age at ~102 m Myr-1 (Figure 15f). This single

955event is supported by the morphology of the arcuate ridges – they are never found as

956multiple superposed sets. Because our youngest crater with pasted on terrain dates to 0.5

957(0.4-50) Ma, this one event is likely to be shorter, hence the retreat rates are probably an

958order of magnitude higher. Further it is likely that the bedrock slope lowering extends

959underneath the pasted-on terrain that we see today, another factor leading to a potential

960under-estimation of the headwall retreat rate. Given the large uncertainty on the age of this

961crater and that our other craters with pasted-on terrain date to >10 Ma , these dates point to

962the “glacial event” having occurred between 5 and 10 Ma. It seems logical to link it to the

963shift in mean obliquity that happened ~5 Ma, but further dating on young craters with pasted-

964on terrain and texturally altered bedrock would be needed to substantiate this claim.

965Previous work has linked this decrease in mean obliquity as the transition point between

966“glacial” and “interglacial” Mars with recent high obliquity excursions representing mini-ice-

967ages . This shift marks a transition from a period with lower average surface temperature to

968higher average surface temperatures, particularly in the mid-latitudes .

969 This general shift in climate conditions is supported by observational evidence, as

970follows. Berman et al. reported that GLFs (termed lobate forms by these authors) tend to be

971found in smaller craters (< 70 km in diameter) with no particular trend observed for other

972crater interior ice deposits. This crater diameter dependency was further explored in Fassett

973et al. who found that synglacial craters tend to be of smaller diameter than pre-glacial

974craters. They concluded that this observation is a function of the age and duration of the

975“glacial” epoch superposed on the crater production function. In agreement, recent work by

52 976Hepburn et al. reported that smaller glacial forms are significantly younger than LDA, LVF or

977crater interior ice deposits (including CCF), as detailed in Section 1.1. It is broadly

978acknowledged that the present epoch is one of glacial retreat . Our results agree with the

979broad overall picture from this previous work; that there has been a long glacial epoch which

980comprised several phases of glacial growth and that glacial retreat dominates present

981conditions. Our results also show that that crater interior ice deposits predate the pasted-on

982terrain and that the presence or absence of the crater interior ice deposits do not seem to

983have any relation to the amount of erosion recorded by these craters (Figure 20).

984 However, the question still remains as to how the terminal stages of an ice age under

985cold and dry Amazonian conditions could generate the conditions for production of

986meltwater, even if only in small amounts. Other authors examining glacial melt on Mars

987have reviewed and proposed the following scenarios: 1) orbital-driven increases in surface

988temperature with current atmospheric composition/pressure, 2) increased surface snow/ice

989thickness reducing heat loss to the atmosphere, 3) geothermal heat increase, 4)

990atmospheric change inducing heating caused by impact or volcanic events (greenhouse

991effect), 5) direct heating caused by impact events.

992 We can immediately rule out direct heating caused by impact events and geothermal

993heat increase, because both of these would be expected to be expressed locally, rather than

994at a global scale as our results indicate. Insulation by thick snow and ice are unlikely,

995because the ice-thicknesses in our proposed glaciers is certainly less than the > 1km

996thickness estimated for this effect to become important ,because their moraines are < 100 m

997in height. Atmospheric changes caused by impacts require a large enough impact to

998substantially change the atmosphere and such impacts are estimated to produce craters

999hundreds of kilometres in size . According to the Hartmann and Neukum production function

1000should happen on Mars every ~1Ga and Lyot crater at 215-km in diameter is generally

1001thought to be the youngest example, whose age is generally agreed to be older than 1 Ga .

1002This leaves the two remaining scenarios of atmospheric heating through secular changes in

53 1003orbital parameters, and/or injection of greenhouse gases from volcanic events. Alone

1004changes in orbital parameters is not anticipated to cause substantial surface warming with

1005the current atmospheric density and composition . Volcanism is thought to be ongoing in the

1006Amazonian with estimates as recently as the last few Ma for Elysium Mons and low shields

1007in Tharsis , which could tie with the timing of the glacial erosion event. However, further work

1008would be needed on the dating and duration of the glacial erosion event to assess if recent

1009volcanic activity would be of sufficient intensity and duration to cause the required

1010temperature increase.

1011Summary figure 20.

10125. Gully-glacier interactions on Mars

1013Our headwall retreat rate calculations reveal that, unlike on Earth, glacial and gully headwall

1014retreat rates appear to be broadly equivalent on Mars. This means that even if gullies were

1015present on the crater walls prior to the glacial erosion event, then it is unlikely that much

1016evidence remains (Figure 20). Hence, we hypothesise that all gullies that are visible today

1017were created after this glacial erosion event (~5 Ma), consistent with other dating studies.

1018Craters that postdate the glacial erosion event possess gullies that erode the bedrock and

1019craters that pre-date the glacial erosion event possess gullies that primarily erode the

1020pasted-on terrain (Figure 20). Gullies in both contexts have similar sized-alcoves . Our

1021estimates of headwall retreat by gullies is an order of magnitude higher than those

1022calculated by de Haas et al. . However, these authors only considered the backweathering

1023required to form the alcoves, whereas our estimation includes the retreat of the whole

1024bedrock outcrop (including the alcoves). Our new estimates bring the retreat rates attributed

1025to martian gullies directly in line with estimates of terrestrial rockwall retreat rates in Arctic,

1026Nordic and Alpine environments of 101-104 m Myr-1 , strengthening the case made by these

1027authors for water as a catalyst for backweathering in these craters.

54 1028We suggest that pasted-on terrain on formerly glaciated crater walls could partly represent

1029subglacial deposits, which would explain the frequent observation of downslope lineations

1030present on the surface of this unit. We also note that the pasted-on terrain also often

1031expresses a polygonised surface texture and that it blends gradually into more glacier-like

1032textures lower on the crater wall (topographically and morphologically), suggesting that it

1033could also represent part of a glacier-like body in and of itself. This is supported by

1034measurements made by Conway and Balme who found that the pasted-on terrain contains

1035between 46% and 95% ice by volume. Also Dickson et al. found that the pasted-on terrain

1036often expressed contouring fractures, which cross-cut and are superposed by gully-fans

1037(located mid- to lower-crater wall) hinting that this unit could still be in motion. The

1038topographic relaxation of gully-incisions as they become polygonised supports this notion.

1039The fact that our work shows that gully-incisions tend to deepen towards higher latitudes in

1040both hemispheres could therefore be interpreted as evidence that the pasted-on terrain

1041thickens and has higher ice content at higher latitudes. Our work reveals that glacial erosion

1042must have liberated a substantial quantity of rock debris and logically this should be found

1043downslope within the pasted-on terrain. Further the surface textures expressed by the

1044pasted-on terrain do not support debris covered ice, as suggested for the martian VFF (lack

1045of pitting, expanded craters, holds cracks and polygons). Hence, we suggest these are ice-

1046rich sediments, rather than sediments covering ice. De Haas et al. found that incisions into

1047gully-fans downslope of pasted-on terrain were much less likely to expose metre-scale

1048boulders than incisions whose catchments included exposed bedrock, implying that the

1049sediments within the pasted-on terrain tend to be less massive than those liberated directly

1050from rockwalls.

1051Our hypothesis for a single “glacial event” having happened ~5-10 Ma fits with the

1052observation made by Dickson et al. that there seems to be only one episode of pasted-on

1053terrain emplacement recorded in the stratigraphic relationships between gullies and pasted-

1054on terrain (they call it “LDM”). Conway and Balme also noted that there was no evidence for

55 1055layering within the incisions into the pasted-on terrain. Dickson et al. noted that inverted

1056gully channels and fans mainly occur on equator-facing slopes between 40°S and 50°S,

1057where our study shows that the pasted-on terrain is at its thickest. We hypothesise that

1058these inverted topographies are produced by the loss of ice: gullies could have deposited

1059debris on top of glacial ice and when the surrounding ice has been lost, the gully deposits

1060(channels and fans) preserve some of the ice and protrude from the surrounding surface.

1061These inverted gullies appear beheaded, not because their upper half has been buried , but

1062because it has been eroded away. If too much ice is removed then the inverted gullies

1063disappear (explaining their paucity where the pasted-on terrain is thin).

1064The observation by Dickson et al. that some gullies appear to be buried by the pasted-on

1065terrain seems to contradict our hypothesis that this unit represents a glacial till under which a

1066significant erosive event has occurred – such deposits should have been removed.

1067However, Dickson et al. only present two examples of gully deposits being revealed from

1068under the pasted-on terrain and both of these could be interpreted as being degraded gully

1069deposits atop pasted-on terrain. Another possibility to explain these observations is that

1070there is another mantling deposit which superposes the pasted-on terrain, a possibility

1071supported by occasional observations of pasted-on terrain emerging from under a light-toned

1072mantle (e.g., Figure 20): a phenomenon also observed by Soare et al .Many authors have

1073noticed that alcoves appear from under the pasted-on terrain , yet the link between these

1074alcoves and pre-existing gullies is uncertain and only circumstantial. Hence, we believe that

1075the population of gullies we see today postdates the glacial event that laid down the pasted-

1076on terrain.

1077Our hypothesis that the pasted-on terrain is predominantly an ice-rich glacial till is supported

1078by the way in which gullies erode into it. Gullies always have a v-shaped incision, or chute

1079whenever they encounter the pasted-on terrain. Such incisions are also common in similar-

1080scaled systems on Earth when the sediment they incise into is granular and loose (Figure

108119). The examples we present here are from volcanoclastic and glacial terrains where the

56 1082loose surface deposits are quickly stripped by flowing liquid water, often involving debris

1083flows, producing sharp v-shaped incisions whose floors are located upon more resistant

1084materials . The v-shaped incisions into glacial moraine on Earth occur during the paraglacial

1085phase of glacial retreat and can penetrate through ice-rich moraine deposits . Where

1086stripping of the pasted-on terrain is advanced, rills can be seen converging over the bedrock

1087to meet at the incisions and a similar phenomenon happens on Earth (compare Figures 19c

1088and g). Isolated gullies within the pasted-on terrain could originate by flow of water at the

1089lower-boundary of the pasted-on terrain, initiating collapse and upslope propagation of the

1090incision (see Figure 19b for a terrestrial example and Figures 5d,7g for martian ones). Our

1091work shows that gullies can substantially erode the rims of impact craters without pasted-on

1092terrain. However, in the presence of pasted-on terrain it is rare that substantial incision into

1093the bedrock by gullies is observed . One of the reasons could be that gully-forming

1094processes are enhanced on steep slopes, hence the lowering of the crater wall slope

1095associated with the pasted-on terrain reduces the erosive capacity of the gullies. Secondly,

1096much of the erosional potential of the gullies could be taken up in stripping the pasted-on

1097terrain.

57 1098

1099Figure 19: Gullies incising into unconsolidated glacial and volcanic sediments on Earth. (a)

1100Debris flows dissecting the unconsolidated volcanoclastic deposits on the side of La Fossa

1101Cone on Vulcano Island, Italy. Image from Ferrucci et al. . (b) Photo from 1979 of gullies

1102incised into the west side of Mt St Helens from the USGS photo library, id: mhob0056. (c)

1103Google Earth image of the same suite of gullies as shown in (a). (d) Google Earth image of

58 1104gullies cut into moraine from glaciers in west Greenland (70.381N, 52.255W). (e-f) Slopes in

1105Fåbergstølsdalen, where gullies cut into the drift deposits and debris flow deposits are seen

1106in the fans in the foreground. Images from Ballantyne and Benn , figures 4 and 5. (g) Gullies

1107stripping pasted-on terrain on Mars, HiRISE image PSP_002066_1425.

1108Our previous work has shown that dense concentrations of extant VFF are anti-correlated

1109with dense gully-populations , yet gullies are intimately associated with pasted-on terrain and

1110arcuate ridges . In this work we have presented evidence in favour of small quantities of

1111meltwater being involved in lowering crater wall slopes where texturally altered bedrock and

1112pasted-on terrain are present. Does this suggest a role of water in the formation of martian

1113gullies? Unfortunately, the findings reported in this paper do not provide any resolution to the

1114ongoing debate between CO2 sublimation and meltwater-generation for forming martian

1115gullies. Given a meltwater hypothesis for this recent glacial erosion event on Mars, it would

1116seem consistent that gullies can be produced via a similar, yet less widespread mechanism,

1117which has been occurring periodically since this glacial erosion event. However, both CO2

1118and H2O ices are often found together and without more detailed knowledge on how CO2

1119sublimation might engender sediment transport it is difficult to assess whether the

1120morphological evidence fits with this transport mechanism .

1121Finally, texturally altered bedrock and pasted-on terrain are more spatially widespread than

1122gullies and occur at low slope angles where gullies are not found. Hence, even if gullies do

1123not represent episodes of flowing liquid water at the surface of Mars, the wet-based glacial

1124erosion event that we report on here does present evidence for a significant occurrence of

1125widespread, recent melt on Mars.

11266. Conclusions

1127We have found evidence for crater wall erosion by very recent (probably 5-10 Ma) small-

1128scale glaciers on Mars which we estimate achieved headwall retreat rates up to ~102 m Myr-

59 11291. This erosion rate is of the same order of magnitude as the landscape change brought

1130about by the youngest gullies (<1 Ma) incised into bedrock on Mars.

1131We posit the erosion of these crater walls is driven by small amounts of subglacial melt,

1132possibly favoured by brines, using the following lines of evidence based on Earth-analogy:

1133 - Recent headwall retreat rates are equivalent to wet-based rather than cold-based

1134 glaciers on Earth;

1135 - Formation of arcuate ridges and associated deformational features at the base of

1136 crater walls involved a component of glacial tectonism, which requires pore-water;

1137 - The presence of texturally altered bedrock which is indicative of ice-segregation and

1138 frost-shattering on Earth; and

1139 - Pasted-on terrain, found topographically below the texturally altered bedrock, may

1140 represent glacial deposits. We suggest the downslope lineations often found on this

1141 deposit could be glacial in origin. The till interpretation is supported by the nature of

1142 the gully-erosion. This provides an alternate explanation to that the pasted-on terrain

1143 is simply a thicker version of the latitude dependant mantle - an airfall deposit of ice

1144 nucleated on dust .

1145Our results suggest that the accelerated crater wall slope-reduction was brought about in a

1146pulse of erosion, which may have occurred sometime 5-10 Ma coincident with a significant

1147climate-shift brought about by a change in Mars’ mean orbital obliquity. Gullies seem to

1148postdate this event and have caused significant reworking of the pasted-on terrain on the

1149steepest slopes. The magnitude of the glacial erosion is the same as that brought about by

1150the youngest modern gullies and hence evidence of gullies predating the glacial erosion is

1151likely difficult to find, and ambiguous at best. Although our results cannot advocate directly

1152that gullies are produced by meltwater runoff, the nature of this “wet” glacial event does

1153provide strong evidence for widespread meltwater generation in Mars’ recent history.

60 11546. Acknowledgements

1155We thank the three reviewers, Goro Komatsu, Daniel Berman and Vic Baker, for their

1156thoughtful comments, and whose suggested modifications greatly improved the manuscript.

1157SJC is supported for her HiRISE work by the French Space Agency, CNES. FEGB is

1158supported by STFC grant ST/N50421X/1. TdH is funded by the Netherlands Organization for

1159Scientific Research (NWO) via Rubicon grant 019.153LW.002.

11607. References cited

1161 1162Aston, A.H., Conway, S.J., Balme, M.R., 2011. Identifying Martian gully evolution, in: Balme, 1163 M., Bargery, A.S., Gallagher, C., Gupta, S. (Eds.), Martian Geomorphology. The 1164 Geological Society of London, pp. 151–169. 1165Atkins, C.B., Barrett, P.J., Hicock, S.R., 2002. Cold glaciers erode and deposit: Evidence 1166 from Allan Hills, Antarctica. Geology 30, 659–662. https://doi.org/10.1130/0091- 1167 7613(2002)030<0659:CGEADE>2.0.CO;2 1168Baker, D.M.H., Head, J.W., 2015. Extensive Middle Amazonian mantling of debris aprons 1169 and plains in Deuteronilus Mensae, Mars: Implications for the record of mid-latitude 1170 glaciation. Icarus 260, 269–288. https://doi.org/10.1016/j.icarus.2015.06.036 1171Baker, D.M.H., Head, J.W., Marchant, D.R., 2010. Flow patterns of lobate debris aprons and 1172 lineated valley fill north of Ismeniae Fossae, Mars: Evidence for extensive mid- 1173 latitude glaciation in the Late Amazonian. Icarus 207, 186–209. 1174Baker, V.R., 2017. Debates-Hypothesis testing in hydrology: Pursuing certainty versus 1175 pursuing uberty: PURSUING CERTAINTY, PURSUING UBERTY. Water Resour. 1176 Res. 53, 1770–1778. https://doi.org/10.1002/2016WR020078 1177Baker, V.R., 2014. Terrestrial analogs, planetary geology, and the nature of geological 1178 reasoning. Planet. Geol. Field Symp. Kitakyushu Jpn. 2011 Planet. Geol. Terr. 1179 Analogs 95, 5–10. https://doi.org/10.1016/j.pss.2012.10.008 1180Ballantyne, C.K., Benn, D.I., 1994. Paraglacial Slope Adjustment and Resedimentation 1181 Following Recent Glacier Retreat, Fabergstolsdalen, Norway. Arct Alp Res 26, 255– 1182 269. https://doi.org/10.2307/1551938 1183Bamber, J.L., Griggs, J.A., Hurkmans, R.T.W.L., Dowdeswell, J.A., Gogineni, S.P., Howat, I., 1184 Mouginot, J., Paden, J., Palmer, S., Rignot, E., Steinhage, D., 2013. A new bed 1185 elevation dataset for Greenland. The Cryosphere 7, 499–510. https://doi.org/10.5194/ 1186 tc-7-499-2013 1187Benn, D.I., Evans, D.J.A., 2010. Glaciers & Glaciation, Second Edition. ed. Hodder 1188 Education, London. 1189Bennett, M.R., 2001. The morphology, structural evolution and significance of push 1190 moraines. Earth-Sci. Rev. 53, 197–236. https://doi.org/10.1016/S0012- 1191 8252(00)00039-8 1192Bennett, M.R., Huddart, D., Glasser, N.F., 1999. Large-Scale Bedrock Displacement by 1193 Cirque Glaciers. Arct. Antarct. Alp. Res. 31, 99–107. https://doi.org/10.2307/1552627 1194Bennett, M.R., Huddart, D., Glasser, N.F., Hambrey, M.J., 2000. Resedimentation of debris 1195 on an ice-cored lateral moraine in the high-Arctic (Kongsvegen, Svalbard). 1196 Geomorphology 35, 21–40. https://doi.org/10.1016/S0169-555X(00)00017-9 1197Berman, D.C., Crown, D.A., Bleamaster Iii, L.F., 2009. Degradation of mid-latitude craters on 1198 Mars. Icarus 200, 77–95.

61 1199Berman, D.C., Crown, D.A., Joseph, E.C.S., 2015. Formation and mantling ages of lobate 1200 debris aprons on Mars: Insights from categorized crater counts. Planet. Space Sci. 1201 111, 83–99. https://doi.org/10.1016/j.pss.2015.03.013 1202Berman, D.C., Hartmann, W.K., Crown, D.A., Baker, V.R., 2005. The role of arcuate ridges 1203 and gullies in the degradation of craters in the Newton Basin region of Mars. Icarus 1204 178, 465–486. 1205Björnsson, H., Gjessing, Y., Hamran, S.-E., Hagen, J.O., LiestøL, O., Pálsson, F., 1206 Erlingsson, B., 1996. The thermal regime of sub-polar glaciers mapped by multi- 1207 frequency radio-echo sounding. J. Glaciol. 42, 23–32. 1208 https://doi.org/10.3189/S0022143000030495 1209Boulton, G.S., 1979. Processes of Glacier Erosion on Different Substrata. J. Glaciol. 23, 15– 1210 38. https://doi.org/10.1017/S0022143000029713 1211Boulton, G.S., van der Meer, J.J.M., Beets, D.J., Hart, J.K., Ruegg, G.H.J., 1999. The 1212 sedimentary and structural evolution of a recent push moraine complex: 1213 Holmstrømbreen, Spitsbergen. Quat. Sci. Rev. 18, 339–371. https://doi.org/10.1016/ 1214 S0277-3791(98)00068-7 1215Boynton, W.V., Feldman, W.C., Squyres, S.W., Prettyman, T.H., Brückner, J., Evans, L.G., 1216 Reedy, R.C., Starr, R., Arnold, J.R., Drake, D.M., Englert, P.A.J., Metzger, A.E., 1217 Mitrofanov, I., Trombka, J.I., d’Uston, C., Wänke, H., Gasnault, O., Hamara, D.K., 1218 Janes, D.M., Marcialis, R.L., Maurice, S., Mikheeva, I., Taylor, G.J., Tokar, R., 1219 Shinohara, C., 2002. Distribution of Hydrogen in the Near Surface of Mars: Evidence 1220 for Subsurface Ice Deposits. Science 297, 81–85. 1221 https://doi.org/10.1126/science.1073722 1222Brough, S., Hubbard, B., Hubbard, A., 2016a. Former extent of glacier-like forms on Mars. 1223 Icarus 274, 37–49. https://doi.org/10.1016/j.icarus.2016.03.006 1224Brough, S., Hubbard, B., Souness, C., Grindrod, P.M., Davis, J., 2016b. Landscapes of 1225 polyphase glaciation: eastern Hellas Planitia, Mars. J. Maps 12, 530–542. 1226 https://doi.org/10.1080/17445647.2015.1047907 1227Burbank, D.W., Leland, J., Fielding, E., Anderson, R.S., Brozovic, N., Reid, M.R., Duncan, 1228 C., 1996. Bedrock incision, rock uplift and threshold hillslopes in the northwestern 1229 Himalayas. Nature 379, 505. 1230Butcher, F.E.G., Balme, M.R., Gallagher, C., Arnold, N.S., Conway, S.J., Hagermann, A., 1231 Lewis, S.R., 2017. Recent Basal Melting of a Mid-Latitude Glacier on Mars. J. 1232 Geophys. Res. Planets n/a-n/a. https://doi.org/10.1002/2017JE005434 1233Byrne, S., Dundas, C.M., Kennedy, M.R., Mellon, M.T., McEwen, A.S., Cull, S.C., Daubar, 1234 I.J., Shean, D.E., Seelos, K.D., Murchie, S.L., Cantor, B.A., Arvidson, R.E., Edgett, 1235 K.S., Reufer, A., Thomas, N., Harrison, T.N., Posiolova, L.V., Seelos, F.P., 2009. 1236 Distribution of mid-latitude ground ice on mars from new impact craters. Science 325, 1237 1674–1676. https://doi.org/10.1126/science.1175307 1238Carr, M.H., 2001. Mars Global Surveyor observations of Martian fretted terrain. J. Geophys. 1239 Res. Planets 106, 23571–23593. https://doi.org/10.1029/2000JE001316 1240Carr, M.H., Head, J.W., 2003. Basal melting of snow on early Mars: A possible origin of 1241 some valley networks: BASAL MELTING OF SNOW ON EARLY MARS. Geophys. 1242 Res. Lett. 30. https://doi.org/10.1029/2003GL018575 1243Chevrier, V., Altheide, T.S., 2008. Low temperature aqueous ferric sulfate solutions on the 1244 surface of Mars. Geophys. Res. Lett. 35, L22101. 1245 https://doi.org/doi:10.1029/2008GL035489 1246Chevrier, V.F., Hanley, J., Altheide, T.S., 2009. Stability of perchlorate hydrates and their 1247 liquid solutions at the Phoenix landing site, Mars. Geophys. Res. Lett. 36, 10202. 1248Christensen, P.R., 2003. Formation of recent martian gullies through melting of extensive 1249 water-rich snow deposits. Nature 422, 45–48. 1250Cohen, D., Hooyer, T.S., Iverson, N.R., Thomason, J.F., Jackson, M., 2006. Role of 1251 transient water pressure in quarrying: A subglacial experiment using acoustic 1252 emissions. J. Geophys. Res. Earth Surf. 111, F03006. 1253 https://doi.org/10.1029/2005JF000439

62 1254Conway, S.J., Balme, M.R., 2016. A novel topographic parameterization scheme indicates 1255 that martian gullies display the signature of liquid water. Earth Planet. Sci. Lett. 454, 1256 36–45. https://doi.org/10.1016/j.epsl.2016.08.031 1257Conway, S.J., Balme, M.R., 2014. Decametre-thick remnant glacial ice deposits on Mars. 1258 Geophys. Res. Lett. 41, 5402–5409. https://doi.org/10.1002/2014GL060314 1259Conway, S.J., Balme, M.R., Soare, R.J., 2015. Using Gullies to Estimate the Thickness of 1260 the Latitude Dependent Mantle on Mars, in: Lunar and Planetary Science 1261 Conference, Lunar and Planetary Science Conference. p. 2964. 1262Conway, S.J., de Haas, T., Harrison, T.N., 2018a. Martian gullies: a comprehensive review 1263 of observations, mechanisms and the insights from Earth analogues. Geol. Soc. 1264 Lond. Spec. Publ. 467. 1265Conway, S.J., de Haas, T., Harrison, T.N., 2018b. Martian gullies: a comprehensive review 1266 of observations, mechanisms and the insights from Earth analogues. Geol. Soc. 1267 Lond. Spec. Publ. 467. 1268Conway, S.J., Harrison, T.N., Lewis, S.R., 2018c. Chapter 3: Martian gullies and their 1269 connection with the martian climate, in: Soare, R.J., Conway, S.J., Clifford, S.M. 1270 (Eds.), Dynamic Mars: Recent and Current Landscape Evolution of the Red Planet. 1271 Elsevier. 1272Conway, S.J., Harrison, T.N., Soare, R.J., Britton, A., Steele, L., 2017. New Slope- 1273 Normalised Global Gully Density and Orientation Maps for Mars. Geol. Soc. Lond. 1274 Spec. Publ. 467, in press. https://doi.org/10.1144/SP467.3 1275Costard, F., Forget, F., Mangold, N., Peulvast, J.P., 2002. Formation of recent Martian 1276 debris flows by melting of near-surface ground ice at high obliquity. Science 295, 1277 110–113. https://doi.org/10.1126/science.1066698 1278Cuffey, K.M., Conway, H., Gades, A.M., Hallet, B., Lorrain, R., Severinghaus, J.P., Steig, 1279 E.J., Vaughn, B., White, J.W.C., 2000. Entrainment at cold glacier beds. Geology 28, 1280 351–354. https://doi.org/10.1130/0091-7613(2000)28<351:EACGB>2.0.CO;2 1281de Haas, T., Conway, S.J., Butcher, F.E.G., Levy, J.S., Grindrod, P.M., Balme, M.R., 1282 Goudge, T.A., 2017. Time will tell: temporal evolution of Martian gullies and 1283 paleoclimatic implications. Geol. Soc. Lond. Spec. Publ. accepted. 1284de Haas, T., Conway, S.J., Krautblatter, M., 2015a. Recent (Late Amazonian) enhanced 1285 backweathering rates on Mars: Paracratering evidence from gully alcoves: LATE 1286 AMAZONIAN BACKWEATHERING RATES. J. Geophys. Res. Planets 120, 2169– 1287 2189. https://doi.org/10.1002/2015JE004915 1288de Haas, T., Hauber, E., Conway, S.J., van Steijn, H., Johnsson, A., Kleinhans, M.G., 1289 2015b. Earth-like aqueous debris-flow activity on Mars at high orbital obliquity in the 1290 last million years. Nat. Commun. 6. https://doi.org/10.1038/ncomms8543 1291de Haas, T., Hauber, E., Kleinhans, M.G., 2013. Local late Amazonian boulder breakdown 1292 and denudation rate on Mars. Geophys. Res. Lett. n/a-n/a. 1293 https://doi.org/10.1002/grl.50726 1294de Haas, T., Ventra, D., Hauber, E., Conway, S.J., Kleinhans, M.G., 2015c. 1295 Sedimentological analyses of martian gullies: The subsurface as the key to the 1296 surface. Icarus 258, 92–108. https://doi.org/10.1016/j.icarus.2015.06.017 1297Dickson, J.L., Head, J.W., Fassett, C.I., 2012. Patterns of accumulation and flow of ice in the 1298 mid-latitudes of Mars during the Amazonian. Icarus 219, 723–732. 1299 https://doi.org/10.1016/j.icarus.2012.03.010 1300Dickson, J.L., Head, J.W., Goudge, T.A., Barbieri, L., 2015. Recent climate cycles on Mars: 1301 Stratigraphic relationships between multiple generations of gullies and the latitude 1302 dependent mantle. Icarus 252, 83–94. https://doi.org/10.1016/j.icarus.2014.12.035 1303Diniega, S., Byrne, S., Bridges, N.T., Dundas, C.M., McEwen, A.S., 2010. Seasonality of 1304 present-day Martian dune-gully activity. Geology 38, 1047–1050. 1305 https://doi.org/10.1130/G31287.1 1306Dundas, C.M., Bramson, A.M., Ojha, L., Wray, J.J., Mellon, M.T., Byrne, S., McEwen, A.S., 1307 Putzig, N.E., Viola, D., Sutton, S., Clark, E., Holt, J.W., 2018. Exposed subsurface

63 1308 ice sheets in the Martian mid-latitudes. Science 359, 199–201. 1309 https://doi.org/10.1126/science.aao1619 1310Dundas, C.M., Byrne, S., McEwen, A.S., Mellon, M.T., Kennedy, M.R., Daubar, I.J., Saper, 1311 L., 2014. HiRISE observations of new impact craters exposing Martian ground ice. J. 1312 Geophys. Res. Planets 119, 2013JE004482. https://doi.org/10.1002/2013JE004482 1313Dundas, C.M., Diniega, S., Hansen, C.J., Byrne, S., McEwen, A.S., 2012. Seasonal activity 1314 and morphological changes in martian gullies. Icarus 220, 124–143. 1315 https://doi.org/10.1016/j.icarus.2012.04.005 1316Dundas, C.M., Diniega, S., McEwen, A.S., 2015. Long-Term Monitoring of Martian Gully 1317 Formation and Evolution with MRO/HiRISE. Icarus 251, 244–263. 1318 https://doi.org/10.1016/j.icarus.2014.05.013 1319Dundas, C.M., McEwen, A.S., Diniega, S., Byrne, S., Martinez-Alonso, S., 2010. New and 1320 recent gully activity on Mars as seen by HiRISE. Geophys. Res. Lett. 37, 1321 doi:10.1029/2009gl041351. https://doi.org/10.1029/2009gl041351 1322Dundas, C.M., McEwen, A.S., Diniega, S., Hansen, C.J., Byrne, S., McElwaine, J.N., 2017. 1323 The Formation of Gullies on Mars Today. Geol. Soc. Lond. Spec. Publ. Martian 1324 Gullies and their Earth Analogues. https://doi.org/10.1144/SP467.5 1325Dyke, A.S., 1993. Landscapes of cold-centred Late Wisconsinan ice caps, Arctic Canada. 1326 Prog. Phys. Geogr. 17, 223–247. https://doi.org/10.1177/030913339301700208 1327Echelmeyer, K., Wang, Z., 1987. Direct Observation of Basal Sliding and Deformation of 1328 Basal Drift at Sub-Freezing Temperatures. J. Glaciol. 33, 83–98. 1329 https://doi.org/10.3189/S0022143000005396 1330Egholm, D.L., Pedersen, V.K., Knudsen, M.F., Larsen, N.K., 2012. Coupling the flow of ice, 1331 water, and sediment in a glacial landscape evolution model. Geomorphology 141– 1332 142, 47–66. https://doi.org/10.1016/j.geomorph.2011.12.019 1333Etzelmüller, B., Hagen, J.O., Vatne, G., Ødegård, R.S., Sollid, J.L., 1996. Glacier debris 1334 accumulation and sediment deformation influenced by permafrost: examples from 1335 Svalbard. Ann. Glaciol. 22, 53–62. https://doi.org/10.3189/1996AoG22-1-53-62 1336Evans, D.J.A., 2007. GLACIAL LANDFORMS | Glacitectonic Structures and Landforms, in: 1337 Elias, S.A. (Ed.), Encyclopedia of Quaternary Science. Elsevier, Oxford, pp. 831– 1338 838. https://doi.org/10.1016/B0-44-452747-8/00084-3 1339Ewertowski, M.W., Tomczyk, A.M., 2015. Quantification of the ice-cored moraines’ short- 1340 term dynamics in the high-Arctic glaciers Ebbabreen and Ragnarbreen, 1341 Petuniabukta, Svalbard. Geomorphology 234, 211–227. 1342 https://doi.org/10.1016/j.geomorph.2015.01.023 1343Fassett, C.I., Dickson, J.L., Head, J.W., Levy, J.S., Marchant, D.R., 2010. Supraglacial and 1344 proglacial valleys on Amazonian Mars. Icarus 208, 86–100. 1345 https://doi.org/10.1016/j.icarus.2010.02.021 1346Fassett, C.I., Levy, J.S., Dickson, J.L., Head, J.W., 2014. An extended period of episodic 1347 northern mid-latitude glaciation on Mars during the Middle to Late Amazonian: 1348 Implications for long-term obliquity history. Geology 42, 763–766. 1349 https://doi.org/10.1130/G35798.1 1350Fassett, C.I., Thomson, B.J., 2014. Crater degradation on the lunar maria: Topographic 1351 diffusion and the rate of erosion on the Moon. J. Geophys. Res. Planets 119, 1352 2014JE004698. https://doi.org/10.1002/2014JE004698 1353Fastook, J.L., Head, J.W., Marchant, D.R., 2014. Formation of lobate debris aprons on Mars: 1354 Assessment of regional ice sheet collapse and debris-cover armoring. Icarus 228, 1355 54–63. https://doi.org/10.1016/j.icarus.2013.09.025 1356Feldman, W.C., Prettyman, T.H., Maurice, S., Plaut, J.J., Bish, D.L., Vaniman, D.T., Mellon, 1357 M.T., Metzger, A.E., Squyres, S.W., Karunatillake, S., Boynton, W.V., Elphic, R.C., 1358 Funsten, H.O., Lawrence, D.J., Tokar, R.L., 2004. Global distribution of near-surface 1359 hydrogen on Mars. J Geophys Res 109, doi:10.1029/2003JE002160. 1360 https://doi.org/10.1029/2003je002160 1361Ferrucci, M., Pertusati, S., Sulpizio, R., Zanchetta, G., Pareschi, M.T., Santacroce, R., 2005. 1362 Volcaniclastic debris flows at La Fossa Volcano (Vulcano Island, southern Italy):

64 1363 Insights for erosion behaviour of loose pyroclastic material on steep slopes. J. 1364 Volcanol. Geotherm. Res. 145, 173–191. 1365Fisher, D.A., 2005. A process to make massive ice in the martian regolith using long-term 1366 diffusion and thermal cracking. Icarus 179, 387–397. 1367 https://doi.org/10.1016/j.icarus.2005.07.024 1368Fitzsimons, S.J., 1996. Formation of thrust-block moraines at the margins of dry-based 1369 glaciers, south Victoria Land, Antarctica. Ann. Glaciol. 22, 68–74. 1370 https://doi.org/10.3189/1996AoG22-1-68-74 1371Gallagher, C., Balme, M., 2015. Eskers in a complete, wet-based glacial system in the 1372 Phlegra Montes region, Mars. Earth Planet. Sci. Lett. 431, 96–109. 1373 https://doi.org/10.1016/j.epsl.2015.09.023 1374Geirsdóttir, Á., Miller, G.H., Andrews, J.T., 2007. Glaciation, erosion, and landscape 1375 evolution of Iceland. J. Geodyn. 43, 170–186. 1376 https://doi.org/10.1016/j.jog.2006.09.017 1377Golombek, M., Bloom, C., Wigton, N., Warner, N., 2014. Constraints on the age of Corinto 1378 Crater from mapping secondaries in Elysium Planitia on Mars, in: Lunar and 1379 Planetary Science Conference. p. 1470. 1380Golombek, M.P., Grant, J.A., Crumpler, L.S., Greeley, R., Arvidson, R.E., Bell, J.F., Weitz, 1381 C.M., Sullivan, R., Christensen, P.R., Soderblom, L.A., Squyres, S.W., 2006. Erosion 1382 rates at the Mars Exploration Rover landing sites and long-term climate change on 1383 Mars: CLIMATE CHANGE FROM THE MARS ROVERS. J. Geophys. Res. Planets 1384 111, n/a-n/a. https://doi.org/10.1029/2006JE002754 1385Golombek, M.P., Warner, N.H., Ganti, V., Lamb, M.P., Parker, T.J., Fergason, R.L., Sullivan, 1386 R., 2014. Small crater modification on Meridiani Planum and implications for erosion 1387 rates and climate change on Mars. J. Geophys. Res. Planets 119, 2014JE004658. 1388 https://doi.org/10.1002/2014JE004658 1389Goodsell, B., Hambrey, M.J., Glasser, N.F., 2005. Debris transport in a temperate valley 1390 glacier: Haut Glacier d’Arolla, Valais, Switzerland. J. Glaciol. 51, 139–146. 1391 https://doi.org/10.3189/172756505781829647 1392Hales, T.C., Roering, J.J., 2007. Climatic controls on frost cracking and implications for the 1393 evolution of bedrock landscapes. J. Geophys. Res. 112. 1394 https://doi.org/10.1029/2006JF000616 1395Hallet, B., 1996. Glacial quarrying: a simple theoretical model. Ann. Glaciol. 22, 1–8. 1396 https://doi.org/10.3189/1996AoG22-1-1-8 1397Hallet, B., 1979. A Theoretical Model of Glacial Abrasion. J. Glaciol. 23, 39–50. 1398 https://doi.org/10.1017/S0022143000029725 1399Hallet, B., Hunter, L., Bogen, J., 1996. Rates of erosion and sediment evacuation by 1400 glaciers: A review of field data and their implications. Glob. Planet. Change 12, 213– 1401 235. https://doi.org/10.1016/0921-8181(95)00021-6 1402Hambrey, M.. J., Fitzsimons, S.J., 2010. Development of sediment–landform associations at 1403 cold glacier margins, Dry Valleys, Antarctica. Sedimentology 57, 857–882. 1404 https://doi.org/10.1111/j.1365-3091.2009.01123.x 1405Hambrey, M.J., Glasser, N.F., 2012. Discriminating glacier thermal and dynamic regimes in 1406 the sedimentary record. Sediment. Geol. 251–252, 1–33. 1407 https://doi.org/10.1016/j.sedgeo.2012.01.008 1408Harbor, J.M., 1992. Numerical modeling of the development of U-shaped valleys by glacial 1409 erosion. GSA Bull. 104, 1364–1375. https://doi.org/10.1130/0016- 1410 7606(1992)104<1364:NMOTDO>2.3.CO;2 1411Harrison, T.N., Malin, M.C., Edgett, K.S., Shean, D.E., Kennedy, M.R., Lipkaman, L.J., 1412 Cantor, B.A., Posiolova, L.V., 2010. Impact-induced overland fluid flow and 1413 channelized erosion at Lyot Crater, Mars. Geophys. Res. Lett. 37, 1414 doi:10.1029/2010gl045074. https://doi.org/10.1029/2010gl045074 1415Harrison, T.N., Osinski, G.R., Tornabene, L.L., Jones, E., 2015. Global Documentation of 1416 Gullies with the Mars Reconnaissance Orbiter Context Camera and Implications for 1417 Their Formation. Icarus 252, 236–254. https://doi.org/10.1016/j.icarus.2015.01.022

65 1418Hart, J.K., Boulton, G.S., 1991. The interrelation of glaciotectonic and glaciodepositional 1419 processes within the glacial environment. Quat. Sci. Rev. 10, 335–350. 1420 https://doi.org/10.1016/0277-3791(91)90035-S 1421Hartmann, W.K., Ansan, V., Berman, D.C., Mangold, N., Forget, F., 2014. Comprehensive 1422 analysis of glaciated martian crater Greg. Icarus 228, 96–120. 1423 https://doi.org/10.1016/j.icarus.2013.09.016 1424Hartmann, W.K., Neukum, G., 2001. Cratering Chronology and the Evolution of Mars. Space 1425 Sci. Rev. 96, 165–194. https://doi.org/10.1023/A:1011945222010 1426Hartmann, W.K., Quantin, C., Werner, S.C., Popova, O., 2010. Do young martian ray craters 1427 have ages consistent with the crater count system? Icarus 208, 621–635. 1428 https://doi.org/10.1016/j.icarus.2010.03.030 1429Hauber, E., Brož, P., Jagert, F., Jodłowski, P., Platz, T., 2011a. Very recent and wide-spread 1430 basaltic volcanism on Mars: RECENT WIDE-SPREAD VOLCANISM ON MARS. 1431 Geophys. Res. Lett. 38, n/a-n/a. https://doi.org/10.1029/2011GL047310 1432Hauber, E., Reiss, D., Ulrich, M., Preusker, F., Trauthan, F., Zanetti, M., Hiesinger, H., 1433 Jaumann, R., Johansson, L., Johnsson, A., Van Gasselt, S., Olvmo, M., 2011b. 1434 Landscape evolution in Martian mid-latitude regions: insights from analogous 1435 periglacial landforms in Svalbard. Geol. Soc. Lond. Spec. Publ. 356, 111–131. 1436 https://doi.org/10.1144/SP356.7 1437Head, J.W., Marchant, D.R., 2003. Cold-based mountain glaciers on Mars: Western Arsia 1438 Mons. Geology 31, 641–644. 1439Head, J.W., Marchant, D.R., Dickson, J.L., Kress, A.M., Baker, D.M., 2010. Northern mid- 1440 latitude glaciation in the Late Amazonian period of Mars: Criteria for the recognition 1441 of debris-covered glacier and valley glacier landsystem deposits. Earth Planet. Sci. 1442 Lett. 294, 306–320. https://doi.org/DOI: 10.1016/j.epsl.2009.06.041 1443Head, J.W., Marchant, D.R., Kreslavsky, M.A., 2008. Formation of gullies on Mars: Link to 1444 recent climate history and insolation microenvironments implicate surface water flow 1445 origin. Proc. Natl. Acad. Sci. U. S. A. 105, 13258–13263. 1446Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant, D.R., 2003. Recent 1447 ice ages on Mars. Nature 426, 797–802. https://doi.org/10.1038/nature02114 1448Hecht, M.H., Kounaves, S.P., Quinn, R.C., West, S.J., Young, S.M.M., Ming, D.W., Catling, 1449 D.C., Clark, B.C., Boynton, W.V., Hoffman, J., DeFlores, L.P., Gospodinova, K., 1450 Kapit, J., Smith, P.H., 2009. Detection of Perchlorate and the Soluble Chemistry of 1451 Martian Soil at the Phoenix Lander Site. Science 325, 64. 1452 https://doi.org/10.1126/science.1172466 1453Hepburn, A., Ng, F., Livingstone, S.J., Hubbard, B., 2018. Polyphase mid-latitude glaciation 1454 on Mars evidenced by dating of superimposed lobate debris aprons. Presented at the 1455 European Geosciences Union General Assembly, Vienna, Austria, pp. EGU2018- 1456 1087. 1457Hirano, M., Aniya, M., 1988. A rational explanation of cross-profile morphology for glacial 1458 valleys and of glacial valley development. Earth Surf. Process. Landf. 13, 707–716. 1459Hobley, D.E.J., Howard, A.D., Moore, J.M., 2014. Fresh shallow valleys in the Martian 1460 midlatitudes as features formed by meltwater flow beneath ice. J. Geophys. Res. 1461 Planets 119, 2013JE004396. https://doi.org/10.1002/2013JE004396 1462Holdsworth, G., Bull, C., 1970. The flow law of cold ice: investigations on Meserve Glacier, 1463 Antarctica. Int. Assoc. Hydrol. Sci. Publ. 86, 204–216. 1464Holt, J.W., Safaeinili, A., Plaut, J.J., Head, J.W., Phillips, R.J., Seu, R., Kempf, S.D., 1465 Choudhary, P., Young, D.A., Putzig, N.E., Biccari, D., Gim, Y., 2008. Radar Sounding 1466 Evidence for Buried Glaciers in the Southern Mid-Latitudes of Mars. Science 322, 1467 1235–1238. 1468Hubbard, B., Milliken, R.E., Kargel, J.S., Limaye, A., Souness, C., 2011. Geomorphological 1469 characterisation and interpretation of a mid-latitude glacier-like form: Hellas Planitia, 1470 Mars. Icarus 211, 330–346. https://doi.org/10.1016/j.icarus.2010.10.021 1471Hubbard, B., Souness, C., Brough, S., 2014. Glacier-like forms on Mars. The Cryosphere 8, 1472 2047–2061. https://doi.org/10.5194/tc-8-2047-2014

66 1473Iverson, N.R., 1991. Potential effects of subglacial water-pressure fluctuations on quarrying. 1474 J. Glaciol. 37, 27–36. https://doi.org/10.3189/S0022143000042763 1475Jakosky, B.M., Mellon, M.T., Varnes, E.S., Feldman, W.C., Boynton, W.V., Haberle, R.M., 1476 2005. Mars low-latitude neutron distribution: Possible remnant near-surface water ice 1477 and a mechanism for its recent emplacement. Icarus 175, 58–67. 1478Jawin, E.R., Head, J.W., Marchant, D.R., 2018. Transient post-glacial processes on Mars: 1479 Geomorphologic evidence for a paraglacial period. Icarus 309, 187–206. 1480 https://doi.org/10.1016/j.icarus.2018.01.026 1481Johnsson, A., Reiss, D., Hauber, E., Hiesinger, H., Zanetti, M., 2014. Evidence for very 1482 recent melt-water and debris flow activity in gullies in a young mid-latitude crater on 1483 Mars. Icarus 235, 37–54. https://doi.org/10.1016/j.icarus.2014.03.005 1484Karlsson, N.B., Schmidt, L.S., Hvidberg, C.S., 2015. Volume of Martian midlatitude glaciers 1485 from radar observations and ice flow modeling. Geophys. Res. Lett. 42, 2627–2633. 1486 https://doi.org/10.1002/2015GL063219 1487Katsube, K., Oguchi, T., 1999. Altitudinal Changes in Slope Angle and Profile Curvature in 1488 the Japan Alps: A Hypothesis Regarding a Characteristic Slope Angle. Geogr. Rev. 1489 Jpn. Ser. B 72, 63–72. https://doi.org/10.4157/grj1984b.72.63 1490Kneissl, T., van Gasselt, S., Neukum, G., 2011. Map-projection-independent crater size- 1491 frequency determination in GIS environments—New software tool for ArcGIS. Planet. 1492 Space Sci. 59, 1243–1254. https://doi.org/10.1016/j.pss.2010.03.015 1493Korup, O., 2008. Rock type leaves topographic signature in landslide-dominated mountain 1494 ranges. Geophys. Res. Lett. 35. https://doi.org/10.1029/2008GL034157 1495Kostama, V.-P., Kreslavsky, M.A., Head, J.W., 2006. Recent high-latitude icy mantle in the 1496 northern plains of Mars: Characteristics and ages of emplacement. Geophys Res Lett 1497 33, L11201. https://doi.org/10.1029/2006GL025946 1498Kreslavsky, M.A., Head, J.W., 2002. Mars: Nature and evolution of young latitude-dependent 1499 water-ice-rich mantle. Geophys. Res. Lett. 29, 14–1. 1500 https://doi.org/10.1029/2002GL015392 1501Kreslavsky, M.A., Head, J.W., 2000. Kilometer-scale roughness of Mars: Results from MOLA 1502 data analysis. J. Geophys. Res. 105, 26695–26712. 1503 https://doi.org/10.1029/2000JE001259 1504Kreslavsky, M.A., Head, J.W., Marchant, D.R., 2008. Periods of active permafrost layer 1505 formation during the geological history of Mars: Implications for circum-polar and mid- 1506 latitude surface processes. Planet. Space Sci. 56, 289–302. 1507 https://doi.org/10.1016/j.pss.2006.02.010 1508Kress, A.M., Head, J.W., 2008. Ring-mold craters in lineated valley fill and lobate debris 1509 aprons on Mars: Evidence for subsurface glacial ice. Geophys. Res. Lett. 35, 23206. 1510Lacelle, D., Davila, A.F., Fisher, D., Pollard, W.H., DeWitt, R., Heldmann, J., Marinova, 1511 M.M., McKay, C.P., 2013. Excess ground ice of condensation–diffusion origin in 1512 University Valley, Dry Valleys of Antarctica: Evidence from isotope geochemistry and 1513 numerical modeling. Geochim. Cosmochim. Acta 120, 280–297. 1514 https://doi.org/10.1016/j.gca.2013.06.032 1515Langevin, Y., Poulet, F., Bibring, J.P., Gondet, B., 2005. Sulfates in the north polar region of 1516 Mars detected by OMEGA/Mars express. Science 307, 1584–1586. 1517Laskar, J., Correia, A.C.M., Gastineau, M., Joutel, F., Levrard, B., Robutel, P., 2004. Long 1518 term evolution and chaotic diffusion of the insolation quantities of Mars. Icarus 170, 1519 343–364. https://doi.org/10.1016/j.icarus.2004.04.005 1520Laskar, J., Robutel, P., 1993. The chaotic obliquity of the planets. Nature 361, 608–612. 1521Levrard, B., Forget, F., Montmessin, F., Laskar, J., 2007. Recent formation and evolution of 1522 northern Martian polar layered deposits as inferred from a Global Climate Model. J. 1523 Geophys. Res. Planets 112, doi: 10.1029/2006JE002772. 1524 https://doi.org/10.1029/2006JE002772 1525Levrard, B., Forget, F., Montmessin, F., Laskar, J., 2004. Recent ice-rich deposits formed at 1526 high latitudes on Mars by sublimation of unstable equatorial ice during low obliquity. 1527 Nature 431, 1072–1075. https://doi.org/10.1038/nature03055

67 1528Levy, J., Head, J.W., Marchant, D.R., 2010. Concentric crater fill in the northern mid- 1529 latitudes of Mars: Formation processes and relationships to similar landforms of 1530 glacial origin. Icarus 209, 390–404. https://doi.org/10.1016/j.icarus.2010.03.036 1531Levy, J.S., Fassett, C.I., Head, J.W., 2016. Enhanced erosion rates on Mars during 1532 Amazonian glaciation. Icarus 264, 213–219. 1533 https://doi.org/10.1016/j.icarus.2015.09.037 1534Levy, J.S., Fassett, C.I., Head, J.W., Schwartz, C., Watters, J.L., 2014. Sequestered glacial 1535 ice contribution to the global Martian water budget: Geometric constraints on the 1536 volume of remnant, midlatitude debris-covered glaciers. J. Geophys. Res. Planets 1537 119, 2014JE004685. https://doi.org/10.1002/2014JE004685 1538Levy, J.S., Head, J., Marchant, D., 2009a. Thermal contraction crack polygons on Mars: 1539 Classification, distribution, and climate implications from HiRISE observations. J. 1540 Geophys. Res. Planets 114, 01007. 1541Levy, J.S., Head, J.W., Marchant, D.R., 2011. Gullies, polygons and mantles in Martian 1542 permafrost environments: cold desert landforms and sedimentary processes during 1543 recent Martian geological history. Geol. Soc. Lond. Spec. Publ. 354, 167–182. https:// 1544 doi.org/10.1144/SP354.10 1545Levy, J.S., Head, J.W., Marchant, D.R., 2009b. Concentric crater fill in Utopia Planitia: 1546 History and interaction between glacial “brain terrain” and periglacial mantle 1547 processes. Icarus 202, 462–476. https://doi.org/10.1016/j.icarus.2009.02.018 1548Levy, J.S., Head, J.W., Marchant, D.R., 2007. Lineated valley fill and lobate debris apron 1549 stratigraphy in Nilosyrtis Mensae, Mars: Evidence for phases of glacial modification 1550 of the dichotomy boundary. J. Geophys. Res. 112. 1551 https://doi.org/10.1029/2006JE002852 1552Levy, J.S., Head, J.W., Marchant, D.R., Dickson, J.L., Morgan, G.A., 2009c. Geologically 1553 recent gully-polygon relationships on Mars: Insights from the Antarctic dry valleys on 1554 the roles of permafrost, microclimates, and water sources for surface flow. Icarus 1555 201, 113–126. 1556Li, H., Robinson, M.S., Jurdy, D.M., 2005. Origin of martian northern hemisphere mid- 1557 latitude lobate debris aprons. Icarus 176, 382–394. 1558 https://doi.org/10.1016/j.icarus.2005.02.011 1559Lin, Z., Oguchi, T., Chen, Y.-G., Saito, K., 2009. Constant-slope alluvial fans and source 1560 basins in Taiwan. Geology 37, 787–790. https://doi.org/10.1130/G25675A.1 1561Lloyd Davies, M.T., Atkins, C.B., van der Meer, J.J.M., Barrett, P.J., Hicock, S.R., 2009. 1562 Evidence for cold-based glacial activity in the Allan Hills, Antarctica. Quat. Sci. Rev. 1563 28, 3124–3137. https://doi.org/10.1016/j.quascirev.2009.08.002 1564MacGregor, K.R., Anderson, R.S., Anderson, S.P., Waddington, E.D., 2000. Numerical 1565 simulations of glacial-valley longitudinal profile evolution. Geology 28, 1031–1034. 1566 https://doi.org/10.1130/0091-7613(2000)28<1031:NSOGLP>2.0.CO;2 1567Madeleine, J.B., Forget, F., Head, J.W., Levrard, B., Montmessin, F., Millour, E., 2009. 1568 Amazonian northern mid-latitude glaciation on Mars: A proposed climate scenario. 1569 Icarus 203, 390–405. 1570Madeleine, J.-B., Head, J.W., Forget, F., Navarro, T., Millour, E., Spiga, A., Colaïtis, A., 1571 Määttänen, A., Montmessin, F., Dickson, J.L., 2014. Recent Ice Ages on Mars: The 1572 role of radiatively active clouds and cloud microphysics. Geophys. Res. Lett. n/a-n/a. 1573 https://doi.org/10.1002/2014GL059861 1574Malin, M.C., Edgett, K.S., 2000. Evidence for recent groundwater seepage and surface 1575 runoff on Mars. Science 288, 2330–2335. 1576 https://doi.org/10.1126/science.288.5475.2330 1577Mangold, N., 2005. High latitude patterned grounds on Mars: Classification, distribution and 1578 climatic control. Mars Polar Sci. III 174, 336–359. 1579 https://doi.org/10.1016/j.icarus.2004.07.030 1580Mangold, N., 2003. Geomorphic analysis of lobate debris aprons on Mars at Mars Orbiter 1581 Camera scale: Evidence for ice sublimation initiated by fractures. J Geophys Res 1582 108, 8021. https://doi.org/10.1029/2002JE001885

68 1583Mangold, N., Allemand, P., 2001. Topographic analysis of features related to ice on Mars. 1584 Geophys. Res. Lett. 28, 407–410. https://doi.org/10.1029/2000GL008491 1585Marchant, D.R., Head, J.W., 2007. Antarctic dry valleys: Microclimate zonation, variable 1586 geomorphic processes, and implications for assessing climate change on Mars. 1587 Icarus 192, 187–222. https://doi.org/10.1016/j.icarus.2007.06.018 1588Massé, M., Bourgeois, O., Le Mouélic, S., Verpoorter, C., Le Deit, L., Bibring, J.P., 2010. 1589 Martian polar and circum-polar sulfate-bearing deposits: Sublimation tills derived 1590 from the North Polar Cap. Icarus 209, 434–451. 1591Matsuoka, N., Murton, J., 2008. Frost weathering: Recent advances and future directions. 1592 Permafr. Periglac. Process. 19, 195–210. 1593Mellon, M.T., 1997. Small-scale polygonal features on Mars: Seasonal thermal contraction 1594 cracks in permafrost. J. Geophys. Res. Planets 102, 25617–25628. 1595 https://doi.org/10.1029/97JE02582 1596Mellon, M.T., Arvidson, R.E., Sizemore, H.G., Searls, M.L., Blaney, D.L., Cull, S., Hecht, 1597 M.H., Heet, T.L., Keller, H.U., Lemmon, M.T., Markiewicz, W.J., Ming, D.W., Morris, 1598 R.V., Pike, W.T., Zent, A.P., 2009. Ground ice at the Phoenix Landing Site: Stability 1599 state and origin. J. Geophys. Res. Planets 114, E00E07. 1600 https://doi.org/10.1029/2009JE003417 1601Mellon, M.T., Jakosky, B.M., 1995. The distribution and behavior of Martian ground ice 1602 during past and present epochs. J. Geophys. Res. 100, 3367. 1603 https://doi.org/10.1029/95JE01027 1604Mellon, M.T., Jakosky, B.M., 1993. Geographic variations in the thermal and diffusive 1605 stability of ground ice on Mars. J. Geophys. Res. 98, 3345. 1606 https://doi.org/10.1029/92JE02355 1607Michael, G.G., Neukum, G., 2010. Planetary surface dating from crater size-frequency 1608 distribution measurements: Partial resurfacing events and statistical age uncertainty. 1609 Earth Planet. Sci. Lett. 294, 223–229. https://doi.org/10.1016/j.epsl.2009.12.041 1610Milliken, R.E., Mustard, J.F., Goldsby, D.L., 2003. Viscous flow features on the surface of 1611 Mars: Observations from high-resolution Mars Orbiter Camera (MOC) images. J 1612 Geophys Res 108, doi:10.1029/2002JE002005. 1613Moran, S.R., Clayton, L., Hooke, R.L., Fenton, M.M., Andriashek, L.D., 1979. Glacier-Bed 1614 Landforms of the Prairie Region of North America. J. Glaciol. 23, 423–424. 1615 https://doi.org/10.1017/S0022143000030161 1616Moratto, Z.M., Broxton, M.J., Beyer, R.A., Lundy, M., Husmann, K., 2010. Ames Stereo 1617 Pipeline, NASA’s Open Source Automated Stereogrammetry Software. Presented at 1618 the 41st Lunar and Planetary Science Conference, LPI, The Woodlands, Texas, p. 1619 #1533. 1620Morgan, G.A., Head III, J.W., Marchant, D.R., 2009. Lineated valley fill (LVF) and lobate 1621 debris aprons (LDA) in the Deuteronilus Mensae northern dichotomy boundary 1622 region, Mars: Constraints on the extent, age and episodicity of Amazonian glacial 1623 events. Icarus 202, 22–38. https://doi.org/10.1016/j.icarus.2009.02.017 1624Murton, J.B., Peterson, R., Ozouf, J.-C., 2006. Bedrock Fracture by Ice Segregation in Cold 1625 Regions. Science 314, 1127–1129. https://doi.org/10.1126/science.1132127 1626Mustard, J.F., Cooper, C.D., Rifkin, M.K., 2001. Evidence for recent climate change on Mars 1627 from the identification of youthful near-surface ground ice. Nature 412, 411–414. 1628 https://doi.org/10.1038/35086515 1629Neukum, G., Hoffmann, H., Hauber, E., Head, J.W., Basilevsky, A.T., Ivanov, B.A., Werner, 1630 S.C., van Gasselt, S., Murray, J.B., McCord, T., The HRSC Co-Investigator Team, 1631 2004. Recent and episodic volcanic and glacial activity on Mars revealed by the High 1632 Resolution Stereo Camera. Nature 432, 971–979. 1633 https://doi.org/10.1038/nature03231 1634Pasquon, K., Gargani, J., Massé, M., Conway, S.J., 2016. Present-day formation and 1635 seasonal evolution of linear dune gullies on Mars. Icarus 274, 195–210. 1636 https://doi.org/10.1016/j.icarus.2016.03.024

69 1637Pasquon, K., Gargani, J., Nachon, M., Conway, S.J., Massé, M., Jouannic, G., Balme, M.R., 1638 Costard, F., Vincendon, M., 2017. Are the different gully morphologies due to 1639 different formation processes on the Kaiser dune field? Geol. Soc. Lond. Spec. Publ. 1640 Martian Gullies and their Earth Analogues. 1641Pierce, T.L., Crown, D.A., 2003. Morphologic and topographic analyses of debris aprons in 1642 the eastern Hellas region, Mars. Icarus 163, 46–65. 1643Plaut, J.J., Picardi, G., Safaeinili, A., Ivanov, A.B., Milkovich, S.M., Cicchetti, A., Kofman, W., 1644 Mouginot, J., Farrell, W.M., Phillips, R.J., Clifford, S.M., Frigeri, A., Orosei, R., 1645 Federico, C., Williams, I.P., Gurnett, D.A., Nielsen, E., Hagfors, T., Heggy, E., Stofan, 1646 E.R., Plettemeier, D., Watters, T.R., Leuschen, C.J., Edenhofer, P., 2007. 1647 Subsurface Radar Sounding of the South Polar Layered Deposits of Mars. Science 1648 316, 92. https://doi.org/10.1126/science.1139672 1649Plaut, J.J., Safaeinili, A., Holt, J.W., Phillips, R.J., Head, J.W., Seu, R., Putzig, N.E., Frigeri, 1650 A., 2009. Radar evidence for ice in lobate debris aprons in the mid-northern latitudes 1651 of Mars. Geophys. Res. Lett. 36, 02203. 1652Putzig, N.E., Phillips, R.J., Campbell, B.A., Holt, J.W., Plaut, J.J., Carter, L.M., Egan, A.F., 1653 Bernardini, F., Safaeinili, A., Seu, R., 2009. Subsurface structure of Planum Boreum 1654 from Mars Reconnaissance Orbiter Shallow Radar soundings. Icarus 204, 443–457. 1655Raack, J., Reiss, D., Appéré, T., Vincendon, M., Ruesch, O., Hiesinger, H., 2015. Present- 1656 Day Seasonal Gully Activity in a South Polar Pit (Sisyphi Cavi) on Mars. Icarus 251, 1657 226–243. https://doi.org/j.icarus.2014.03.040 1658Richardson, M.I., Mischna, M.A., 2005. Long-term evolution of transient liquid water on 1659 Mars. J. Geophys. Res. Planets 110, 03003. 1660Richardson, M.I., Wilson, R.J., 2002. A topographically forced asymmetry in the martian 1661 circulation and climate. Nature 416, 298. 1662Rise, L., Bellec, V.K., Ottesen, D., Bøe, R., Thorsnes, T., 2016. Hill–hole pairs on the 1663 Norwegian continental shelf. Geol. Soc. Lond. Mem. 46, 203–204. 1664 https://doi.org/10.1144/M46.42 1665Robbins, S.J., Hynek, B.M., 2012. A new global database of Mars impact craters ≥1 km: 1. 1666 Database creation, properties, and parameters. J. Geophys. Res. - Planets 117, 1667 doi:10.1029/2011JE003966. https://doi.org/10.1029/2011JE003966 1668Roberts, D.H., Long, A.J., 2005. Streamlined bedrock terrain and fast ice flow, Jakobshavns 1669 Isbrae, West Greenland: implications for ice stream and ice sheet dynamics. Boreas 1670 34, 25–42. https://doi.org/10.1111/j.1502-3885.2005.tb01002.x 1671Rӧthlisberger, H., Iken, A., 1981. Plucking as an Effect of Water-Pressure Variations at the thlisberger, H.,Iken,A.,1981.PluckingasanEffectofWater-PressureVariationsatthe 1672 Glacier Bed. Ann. Glaciol. 2, 57–62. https://doi.org/10.3189/172756481794352144 1673Sass, O., 2005. Rock moisture measurements: techniques, results, and implications for 1674 weathering. Earth Surf. Process. Landf. 30, 359–374. 1675 https://doi.org/10.1002/esp.1214 1676Scanlon, K.E., Head, J.W., Marchant, D.R., 2015. Volcanism-induced, local wet-based 1677 glacial conditions recorded in the Late Amazonian Arsia Mons tropical mountain 1678 glacier deposits. Icarus 250, 18–31. https://doi.org/10.1016/j.icarus.2014.11.016 1679Scanlon, K.E., Head, J.W., Wilson, L., Marchant, D.R., 2014. Volcano–ice interactions in the 1680 Arsia Mons tropical mountain glacier deposits. Icarus 237, 315–339. 1681 https://doi.org/10.1016/j.icarus.2014.04.024 1682Schon, S.C., Head, J.W., 2012. Gasa impact crater, Mars: Very young gullies formed from 1683 impact into latitude-dependent mantle and debris-covered glacier deposits? Icarus 1684 218, 459–477. https://doi.org/10.1016/j.icarus.2012.01.002 1685Schon, S.C., Head, J.W., 2011. Keys to gully formation processes on Mars: Relation to 1686 climate cycles and sources of meltwater. Icarus 213, 428–432. 1687 https://doi.org/10.1016/j.icarus.2011.02.020 1688Schon, S.C., Head, J.W., Fassett, C.I., 2012. Recent high-latitude resurfacing by a climate- 1689 related latitude-dependent mantle: Constraining age of emplacement from counts of 1690 small craters. Planet. Space Sci. 69, 49–61. 1691 https://doi.org/10.1016/j.pss.2012.03.015

70 1692Schon, S.C., Head, J.W., Fassett, C.I., 2009a. Unique chronostratigraphic marker in 1693 depositional fan stratigraphy on Mars: Evidence for ca. 1.25 Ma gully activity and 1694 surficial meltwater origin. Geology 37, 207–210. https://doi.org/10.1130/g25398a.1 1695Schon, S.C., Head, J.W., Milliken, R.E., 2009b. A recent ice age on Mars: Evidence for 1696 climate oscillations from regional layering in mid-latitude mantling deposits. Geophys 1697 Res Lett 36, doi:10.1029/2009GL038554. https://doi.org/10.1029/2009GL038554 1698Segura, T.L., Toon, O.B., Colaprete, A., 2008. Modeling the environmental effects of 1699 moderate-sized impacts on Mars. J. Geophys. Res. 113. 1700 https://doi.org/10.1029/2008JE003147 1701Shean, D.E., Alexandrov, O., Moratto, Z.M., Smith, B.E., Joughin, I.R., Porter, C., Morin, P., 1702 2016. An automated, open-source pipeline for mass production of digital elevation 1703 models (DEMs) from very-high-resolution commercial stereo satellite imagery. ISPRS 1704 J. Photogramm. Remote Sens. 116, 101–117. 1705 https://doi.org/10.1016/j.isprsjprs.2016.03.012 1706Shreve, R.L., 1984. Glacier sliding at subfreezing temperatures. J. Glaciol. 30, 341–347. 1707Sinha, R.K., Vijayan, S., 2017. Geomorphic investigation of craters in Alba Mons, Mars: 1708 Implications for Late Amazonian glacial activity in the region. Planet. Space Sci. 144, 1709 32–48. https://doi.org/10.1016/j.pss.2017.05.014 1710Sizemore, H.G., Zent, A.P., Rempel, A.W., 2015. Initiation and Growth of Martian Ice 1711 Lenses. Icarus 251, 191–210. https://doi.org/10.1016/j.icarus.2014.04.013 1712Smith, C.A., Lowell, T.V., Caffee, M.W., 2009. Lateglacial and Holocene cosmogenic surface 1713 exposure age glacial chronology and geomorphological evidence for the presence of 1714 cold-based glaciers at Nevado Sajama, Bolivia. J. Quat. Sci. 24, 360–372. 1715 https://doi.org/10.1002/jqs.1239 1716Soare, R.J., Conway, S.J., Dohm, J.M., 2014. Possible ice-wedge polygons and recent 1717 landscape modification by “wet” periglacial processes in and around the Argyre 1718 impact basin, Mars. Icarus 233, 214–228. 1719 https://doi.org/10.1016/j.icarus.2014.01.034 1720Soare, R.J., Conway, S.J., Gallagher, C., Dohm, J.M., 2017. Ice-rich (periglacial) vs icy 1721 (glacial) depressions in the Argyre region, Mars: a proposed cold-climate dichotomy 1722 of landforms. Icarus 282, 70–83. https://doi.org/10.1016/j.icarus.2016.09.009 1723Souness, C., Hubbard, B., 2012. Mid-latitude glaciation on Mars. Prog. Phys. Geogr. 36, 1724 238–261. https://doi.org/10.1177/0309133312436570 1725Souness, C., Hubbard, B., Milliken, R.E., Quincey, D., 2012. An inventory and population- 1726 scale analysis of Martian glacier-like forms. Icarus 217, 243–255. 1727 https://doi.org/10.1016/j.icarus.2011.10.020 1728Squyres, S.W., 1979. The distribution of lobate debris aprons and similar flows on Mars. J. 1729 Geophys. Res. Solid Earth 84, 8087–8096. https://doi.org/10.1029/JB084iB14p08087 1730Squyres, S.W., 1978. Martian fretted terrain: Flow of erosional debris. Icarus 34, 600–613. 1731 https://doi.org/10.1016/0019-1035(78)90048-9 1732Svitek, T., Murray, B., 1990. Winter frost at Viking Lander 2 site. J. Geophys. Res. 95, 1495. 1733 https://doi.org/10.1029/JB095iB02p01495 1734Thomas, P.C., Malin, M.C., Edgett, K.S., Carr, M.H., Hartmann, W.K., Ingersoll, A.P., James, 1735 P.B., Soderblom, L.A., Veverka, J., Sullivan, R., 2000. North–south geological 1736 differences between the residual polar caps on Mars. Nature 404, 161. 1737Thorn, C.E., Hall, K., 1980. Nivation: An Arctic-Alpine Comparison and Reappraisal. J. 1738 Glaciol. 25, 109–124. https://doi.org/10.3189/S0022143000010339 1739Toner, J.D., Catling, D.C., Light, B., 2014. The formation of supercooled brines, viscous 1740 liquids, and low-temperature perchlorate glasses in aqueous solutions relevant to 1741 Mars. Icarus 233, 36–47. https://doi.org/10.1016/j.icarus.2014.01.018 1742Tornabene, L.L., Watters, W.A., Osinski, G.R., Boyce, J.M., Harrison, T.N., Ling, V., 1743 McEwen, A.S., 2018. A depth versus diameter scaling relationship for the best- 1744 preserved melt-bearing complex craters on Mars. Icarus 299, 68–83. 1745 https://doi.org/10.1016/j.icarus.2017.07.003

71 1746Vaucher, J., Baratoux, D., Mangold, N., Pinet, P., Kurita, K., Grégoire, M., 2009. The 1747 volcanic history of central Elysium Planitia: Implications for martian magmatism. 1748 Icarus 204, 418–442. https://doi.org/10.1016/j.icarus.2009.06.032 1749Vincendon, M., 2015. Identification of Mars gully activity types associated with ice 1750 composition. J. Geophys. Res. Planets 120, 1859–1879. 1751 https://doi.org/10.1002/2015JE004909 1752Vincendon, M., Mustard, J., Forget, F., Kreslavsky, M., Spiga, A., Murchie, S., Bibring, J.-P., 1753 2010. Near-tropical subsurface ice on Mars. Geophys. Res. Lett. 37, 1754 doi:10.1029/2009GL041426. https://doi.org/10.1029/2009gl041426 1755Waller, R.I., 2001. The influence of basal processes on the dynamic behaviour of cold-based 1756 glaciers. Quat. Int. 86, 117–128. https://doi.org/10.1016/S1040-6182(01)00054-4 1757Watters, W.A., Geiger, L.M., Fendrock, M., Gibson, R., 2015. Morphometry of small recent 1758 impact craters on Mars: Size and terrain dependence, short-term modification. J. 1759 Geophys. Res. Planets 2014JE004630. https://doi.org/10.1002/2014JE004630 1760Werner, S.C., 2009. The global martian volcanic evolutionary history. Icarus 201, 44–68. 1761 https://doi.org/10.1016/j.icarus.2008.12.019 1762Whalley, W.B., Azizi, F., 2003. Rock glaciers and protalus landforms: Analogous forms and 1763 ice sources on Earth and Mars. J Geophys Res-Planets 108, 1764 doi:10.1029/2002JE001864. 1765Williams, K.E., Toon, O.B., Heldmann, J.L., McKay, C., Mellon, M.T., 2008. Stability of mid- 1766 latitude snowpacks on Mars. Icarus 196, 565–577. 1767Williams, K.E., Toon, O.B., Heldmann, J.L., Mellon, M.T., 2009. Ancient melting of mid- 1768 latitude snowpacks on Mars as a water source for gullies. Icarus 200, 418–425. 1769Willmes, M., Reiss, D., Hiesinger, H., Zanetti, M., 2012. Surface age of the ice–dust mantle 1770 deposit in Malea Planum, Mars. Planet. Space Sci. 60, 199–206. 1771 https://doi.org/10.1016/j.pss.2011.08.006

72 1772Table 1: Summary of the craters included in this study, the elevation data and their ages.a

crater details Digital Elevation Model (DTM) information dating name latitude longitude diameter features method credit image1 resolutio image2 resolutio convergence vertical Source Age (km) n (m/pix) n (m/pix) angle error (m)

gullies, ridges, ESP_031259_140 Taltal (East)39.5°S 234.4°E 9.8 facing slope SS LPG 0 0.5 ESP_037074_1400 0.5 5.9 0.97 de Haas et al. 220 (100–400) Ma Taltal gullies, ridges, ESP_021989_140 (West) 39.51°S 234.17°E 9.8 facing slope ASP LPG 0 0.263 ESP_021712_1400 0.253 25.5 0.11 " " gullies, ridges, Open ESP_011672_139 Talu 40.3°S 20.1°E 9.1 facing slope SS Uni. 5 0.26 ESP_011817_1395 0.26 15.7 0.18 de Haas et al. 13 (10–22) Ma PSP_004085_142 Nqutu 37.9°S 169.6°E 20 gullies, no ridges SS Birk. 0 0.27 PSP_004019_1420 0.25 20.4 0.15 de Haas et al. 1.8 (1–3.5) Ga gullies, ridges, ESP_011436_142 Nybyen 37.03°S 343.36°E 6 facing slope SS UoA 5 0.262 PSP_006663_1425 0.252 14.5 0.20 Figure 12e 110 (5-300) Ma ESP_023546_137 Crater D 42.24°S 202.19°E 20 gullies, ridges SS UoA 5 0.555 ESP_023612_1375 0.509 19.4 0.32 Figure 12d 305 (80-600) Ma gullies, ridges, PSP_006261_141 Corozal 38.7°S 159.4°E 8 facing slope SS UoA 0 0.25 ESP_014093_1410 0.29 28.7 0.11 de Haas et al. 1.4 (0.8–2.5) Ga gullies, ridges, ESP_030443_141 Niquero 38.77°S 193.99°E 10 facing slope ASP LPG 0 0.534 ESP_030021_1410 0.508 23.5 0.25 Figure 12f 1.3 (0.5-2.5) Ga no ridges, facing PSP_009098_214 Crater E 33.47°N 84.15°E 14 slope ASP LPG 0 0.294 PSP_009164_2140 0.315 19 0.18 Figure 12b 0.5 (0.4-50) Ma Open PSP_006837_134 45.11°S 274.2°E Istok 4.7 bedrock gullies SS Uni. 5 0.25 PSP_007127_1345 0.258 20.1 0.14 Johnsson et al. 0.19 (0.1–1.0) Ma Open PSP_003939_142 37.66°S 192.93°E Galap 5.6 bedrock gullies SS Uni. 0 0.256 PSP_003939_1420 0.291 21.7 0.15 de Haas et al. 6.5 (5–9) Ma Open PSP_002118_151 Hartmann et al. , 28.65°S 226.90°E Zumba 2.8 fresh crater SS Uni. 0 0.255 PSP_003608_1510 0.278 18.1 0.17 Schon et al. (0.1-0.8) Ma ESP_023662_196 Crater F 15.82°N 267.8°E 8 fresh crater SS LPG 0 0.608 ESP_023372_1960 0.549 25.5 0.25 Figure 12g 15 (8-100) Ma ESP_011893_163 Crater G 16.37°S 102.96°E 5.8 fresh crater ASP LPG 5 0.259 ESP_012315_1635 0.268 13.1 0.23 Figure 12h 12.6 (5-30) Ma ESP_020539_211 Crater H 31.11°N 212.39°E 8.1 fresh crater ASP LPG 5 0.612 ESP_020117_2115 0.584 16.7 0.41 Figure 12c 20 (5-100) Ma PSP_003611_197 16.95°N 141.70°E 0.28 PSP_004244_1970 0.3 Corinto 13.9 fresh crater SS UoA 0 18.1 0.18 Golombek et al. (0.1-3.0) Ma PSP_005837_196 16.41°N 209.7°E 2.5 0.285 PSP_005837_1965 0.319 20.1 Crater C fresh crater SS UoA 5 0.17 de Haas et al. 5.3 (4-8) Ma old equatorial ESP_026861_155 Crater I 24.09°S 59.50°E 6.8 crater SS UoA 5 0.258 ESP_027349_1555 0.268 24.4 0.12 Figure 12a 1.1 (0.6-3) Ga 1773 1774a Abbreviations: “SS” = SocetSet, “ASP” = Ames Stereo Pipeline, “LPG” = Laboratoire de Planétologie et Géodynamique, “Birk.” = Birkbeck 1775University of London, “UoA” = University of Arizona.

73 1776

74 1777Table 2: Summary of results for each site, including number of classes, slopes, exposed rock length, retreat distance and retreat rate.a 1778Table 2(i)

name mean outer slope number of classes mean inner slope (°) (°) exposed rock length (m) inner wall outer glacial gully unmodifi glacial unmodifi glacial unmodified wall ed ed

gu(0),gl(13),u( gl(12),u(1 29.4±1. 31.9±1.9 15.2±3. 18.3±4.3 Taltal (East) 15) 5) 72 1 91 5 224 (+368/-172) 475 (+245/-280) gu(0),gl(9),u(1 gl(9),u(10 26.6±1. 33.5±2.0 12.0±6. 18.3±5.7 Taltal (West) 0) ) 33 9 16 0 138 (+149/-94) 385 (+260/-276) gu(0),gl(21),u( gl(11),u(2 30.4±1. 35.3±1.3 12.8±3. 18.5±9.8 Talu 11) 0) 43 9 64 9 435 (+247/-241) 429 (+125/-159) gu(0),gl(17),u( gl(0),u(17 20.5±2. 13.1±7.1 Nqutu 0) ) 66 1 606 (+138/-212) gu(6),gl(16),u( gl(17),u(2 32.4±1. 37.2±1.8 18.3±4. 20.7±5.8 Nybyen 15) 0) 37 5 85 4 255 (+98/-155) 325 (+161/-133) gu(18),gl(44), gl(48),u(1 27.5±2. 30.0±2.0 3.8±6.7 Crater D u(8) 5) 61 9 5 2.3±3.60 437 (+751/-385) 517 (+219/-191) gu(0),gl(17),u( gl(24),u(1 25.0±1. 31.5±3.0 12.1±3. 15.2±5.9 Corozal 22) 5) 83 2 03 7 703 (+287/-218) 683 (+326/-305) gu(0),gl(7),u(1 gl(15),u(7 26.0±3. 12.5±3. 25.9±2.5 Niquero 5) ) 18 34.3±.68 06 6 154 (+236/-96) 331 (+190/-173) gu(0),gl(8),u(1 gl(8),u(21 33.1±0. 35.5±1.8 3.6±7.9 19.1±10. Crater E 0) ) 92 3 4 76 411 (+244/-275) 327 (+276/-170) gu(20),gl(0),u( gl(0),u(41 36.2±1. 40.4±3.4 13.4±4.9 Istok 21) ) 64 0 0 434 (+225/-211) gu(21),gl(0),u( gl(0),u(43 33.1±1. 37.8±1.4 14.7±8.5 Galap 22) ) 56 6 5 434 (+225/-211) gu(0),gl(0),u(3 gl(0),u(30 Zumba 0) ) 37.8±.50 8.5±5.37 434 (+225/-211) gu(0),gl(0),u(3 gl(0),u(33 38.0±1.4 19.4±3.3 Crater F 3) ) 0 0 434 (+225/-211) gu(0),gl(0),u(3 gl(0),u(30 14.5±5.3 Crater G 0) ) 37.1±.52 0 434 (+225/-211) gu(0),gl(0),u(3 gl(0),u(32 39.5±1.5 19.1±4.0 Crater H 3) ) 0 9 434 (+225/-211) gu(0),gl(0),u(1 gl(0),u(18 41.9±3.9 23.0±10. Corinto 8) ) 6 04 434 (+225/-211) gu(0),gl(0),u(2 gl(0),u(25 Crater C 5) ) 36.0±.73 7.7±2.76 434 (+225/-211) gu(0),gl(0),u(3 gl(0),u(32 25.9±2.3 10.4±4.7 Crater I 2) ) 3 0 434 (+225/-211) 1779

75 1780

76 1781Table 2(ii)

name retreat distance (m) retreat rate (m/Myr) glacial gullied glacial gullied unmodified glacial gullied glacial gullied unmodifie compared compared compared compared compared to compared to compared to compared to compared to d to to to same to same equatorial equatorial equatorial same crater same crater compared equatorial equatorial crater crater to equatorial Taltal (East) 68.7 83.5 21.3 0.31 (+0.51/-0.24) 0.7 (+1.14/-0.53) 43 (+70/-33) Taltal (West) 36.2 61.0 33.6 0.16 (+0.18/-0.11) 0.51 (+0.55/-0.35) 67 (+73/-46) Talu 91.7 150.5 74.5 7.05 (+4/-3.9) 50 (+28/-28) 149 (+85/-83) Nqutu 353.3 0.44 (+0.1/-0.15) Nybyen 39.3 74.7 41.8 0.36 (+0.14/-0.22) 0.71 (+0.27/-0.43) 84 (+32/-51) Crater D 155.8 183.5 43.0 0.51 (+0.88/-0.45) 0.82 (+1.4/-0.72) 86 (+148/-76) Corozal 222.9 337.7 168.1 0.16 (+0.07/-0.05) 0.56 (+0.23/-0.17) 336 (+137/-104) Niquero 37.0 70.4 44.0 0.03 (+0.04/-0.02) 0.09 (+0.13/-0.05) 88 (+134/-54) Crater E 84.2 112.3 35.3 168 (+100/-113) 1123 (+667/-751) 71 (+42/-47) Istok 20.0 71.4 54.3 105 (+25/-46) 794 (+187/-344) 109 (+26/-47) Galap 71.1 143.6 83.8 10.9 (+3.5/-2.8) 96 (+31/-25) 168 (+54/-43) Zumba 58.8 131 (+68/-64) Crater F 56.1 3.74 (+1.94/-1.82) Crater G 68.2 5.41 (+2.81/-2.63) Crater H 35.3 1.76 (+0.92/-0.86) Corinto 0.0 Crater C 82.6 15.6 (+8.1/-7.6) Crater I 199.1 0.18 (+0.09/-0.09) 1782a Abbreviations: “gu” = gullied, “gl” = glaciated, “un” = un modified. For slopes standard deviations are provided as the uncertainty values and for 1783rock length the variance is given. This variance is propagated into the retreat distance and retreat rate calculations. Note: the exposed bedrock 1784length for the non-glacial craters is assumed to be the mean value of the unmodified exposed rock length from the glaciated craters. 1785 1786

77