Namibia is remarkable in terms of its network of c. 300 Myr old fjords, cut by Gondwanan glaciers at high palaeolatitudes during the Late Paleozoic Ice Age (LPIA). A classic suite of structures is preserved on Proterozoic bedrock, including striations, roches moutonnées and other subglacial features within many of these palaeovalleys. Some palaeovalleys are plastered with comparatively thin diamictites (a few metres) of presumed subglacial origin, in turn capped by ice marginal delta successions (tens of metres). Close examination of deposits in the outer Orutanda Fjord palaeovalley reveals an architecture that shows departure from this trend. There, boulder-bearing diamictites pass laterally into highly contorted heterolithic successions comprising folded and faulted sandstones, siltstones and shales. Aerial imagery from uncrewed aerial vehicles in tandem with traditional field observations permits mapping of assemblages of soft-sediment deformation structures (tight to recumbent folds, deformation bands, faults, sheath folds) together with spatial mapping of the long axis of boulders in diamictite. Collectively, this assemblage of structures points towards subglacial deformation, and hence an origin as a ‘deforming bed’ beneath a glacier. Although several open questions remain regarding the precise deformation mechanisms, the concept of glacial readvances into a deeply incised palaeovalley is proposed, by direct analogy to similar stratigraphic architectures in the LPIA record of South America.

Unveiling the structure and fill of fjord systems is an important task that is equally applicable to modern-day and ancient settings. This is because fjords represent a clear connection between terrestrially sourced glaciers and the marine environment, meaning that glaciers terminating in these settings are particularly sensitive to sea-level changes (Stevens et al. 2022). Furthermore, they represent a considerable carbon sink, which, when considered at the global scale, have the potential to influence and moderate climatic warming (Smith et al. 2015). In spite of this importance, although fjord systems are well known from modern settings such as coastal Patagonia, Norway and Greenland, remarkably few convincing examples are recorded from the deep time record of glaciation on Earth, and all deep time examples appear to be restricted to the Late Paleozoic record (e.g. Kneller et al. 2004; Dykstra et al. 2006; Aquino et al. 2014; Tedesco et al. 2016; Dietrich et al. 2021; Valdez Buso et al. 2021; Rosa et al. 2023). One region that is exceptional in this regard is the northern part of Namibia, specifically the Kaokoland region, where modern-day river valleys trace the course of approximately 300 Myr old, hundreds of metres deep, glacial incisions (Martin 1981). These ancient glacial valleys cut through a series of Precambrian crystalline lithologies, notably gneiss and granite, and locally into Neoproterozoic cover rocks of the Otavi Group (Miller 2008). Some significant new work has emerged on these palaeovalleys in recent years (Dietrich et al. 2021; Fedorchuk et al. 2023; Rosa et al. 2023), yet still, owing to the remoteness of the region, much of their network remains underexplored.

In terms of previous work, Martin and Schalk (1959) described compelling evidence of glacial striations and polished Precambrian crystalline basement at several locations, extending the known range of the Late Paleozoic Ice Age (LPIA) deposits considerably further north across (then) South West Africa. Considerable work had already been undertaken in South Africa by this point, beginning with that by Sutherland (1870), who first recognized evidence for glaciation, and the synthesis of du Toit (1954). Martin (1981) published a map of deep glacial palaeovalleys in northern Namibia; remarkably, these deep gorges are occupied by modern river systems. In that paper, it was proposed that ‘the Kunene valley may therefore have been a fjord during deglaciation’ (Martin 1981). Visser (1997) similarly proposed that ‘some of the Kaokoveld valleys were typical fjords’. The transition from terrestrial, glacially cut palaeovalleys to fjords was more thoroughly developed and justified based on sedimentary facies relationships (Dietrich et al. 2021). More recently, Fedorchuk et al. (2023) revealed insights into the bedrock glacial geomorphology, describing striated and polished surfaces, roches moutonées and whalebacks from a section west of Opuwo, and closely comparable with bedrock geomorphology reported from Uruguay (Assine et al. 2018; Isbell et al. 2023). From a separate palaeovalley, Rosa et al. (2023) presented sedimentological descriptions of the northern deposits of an inner fjord, called the Orutanda Fjord, in which a variety of mass-flow deposits recording steep margin instability were documented. To exploit the excellent outcrop geomorphology, Le Heron et al. (2024) showed that a photogrammetric approach could be highly beneficial as an important complement to traditional sedimentological outcrop description (Le Heron et al. 2022a), both within the fjords and in glaciated areas of lower relief. In terms of geochronology, U–Pb analysis of zircon grains derived from tuffs interbedded with marine sequences in central and southern Namibia suggests that at least in that region glaciation probably occurred simultaneously between 300 and 298 Ma (Griffis et al. 2019, 2021). For northern Namibia, given the paucity of geochronological data, such precise dates need to be treated with caution. In an important study, Dietrich et al. (2019) described diamictites in neighbouring Botswana, concluding that some boulder-rich deposits show no evidence of glacial transport or deposition. Thus, careful analysis of the Dwyka record, both within and outside palaeovalleys, is required to properly understand dynamics of the LPIA.

In this paper, we present a dataset consisting of new outcrop descriptions geographically distinct from those published by Rosa et al. (2023) in the southern Orutanda Fjord (Fig. 1a). Given the local north to south ice flow in this region (Rosa et al. 2023) they can be considered to belong to a distal part of the same valley. In this paper, we thus refer to the inner and outer parts of the Orutanda Fjord, reflecting the northern study area of Rosa et al. (2023) and the work contained herein, respectively. From an aerial 3D perspective, the low-lying regions of the Orutanda Fjord are characterized by outcrops of the Dwyka Group, which variably consist of shales and siltstones, sandstones, conglomerates and boulder-bearing diamictites (Fig. 1b and c). These deposits onlap against the margins of the fjord, and hence might be superficially interpreted as recording ice recession, valley abandonment and flooding consistent with observations of other fjords in Namibia (Dietrich et al. 2021). In this paper, however, we recognize a suite of evidence for abundant soft-sediment deformation in fjord-filling sediments, which we interpret as recording subglacial shearing of a soft bed beneath an LPIA glacier, and hence ice readvance in the fjord.

Fieldwork was conducted in February 2023, as part of an effort to better understand the geomorphology and infill architecture of north Namibian palaeovalleys. Sedimentological descriptions, including basic logging, striation orientation measurements and outcrop photography, were supplemented with targeted uncrewed aerial vehicle (UAV) flights with the specific aim of mapping the orientation of lineations (e.g. boulder orientations in diamictite) and bedding plane orientations in deformed sedimentary rocks. These datasets were collected in parallel to test the hypothesis that glacial readvances had occurred once the glacial palaeovalley had been cut, leaving a classic subglacially carved landscape as described elsewhere in South Africa and Namibia (Andrews et al. 2019; Le Heron et al. 2019, 2022a). Furthermore, extensive exploration of each outcrop was undertaken on foot to guarantee thorough ground truthing. This aspect was especially important for those areas where large boulders were mapped, ensuring that they were completely in situ and encased in diamictite, rather than weathered out or having been mobilized by later gravitational or slope processes. We used a DJI Mavic 3 equipped with a Hasselblad camera in manual mode. We flew double grid missions with a camera angle of approximately 60° and approximately 70% overlap between images. The photographs were imported into Agisoft Metashape Professional and using a standard workflow a dense point cloud, a 3D model, an orthomosaic and a digital elevation model (DEM) were generated for each dataset. To map boulder orientations, we used a customized Excel spreadsheet that allows rose diagrams to be generated from directional data. The orientation of the boulders was drawn in Corel Draw on the north-facing orthomosaic using two-point lines and then exported as an SVG file, which can be read out in a text editor. The x and y values obtained in this way describe exact positions that can be used to specify the orientation of the lines in a virtual NSWE space. We generated rose diagrams from clast a-axis orientations, the strike of bedding surfaces and fold axial traces. This is the same approach as that developed to map clast orientations in outcrop photogrammetry of diamictites and in thin sections (Kettler et al. 2023).

Our outcrop descriptions derive from three principal outcrops in the outer Orutanda Fjord (Fig. 1). These are presented below, in order from the northernmost to the southernmost section. The observations from each form the foundation of the interpretation of a major subglacial shear zone presented in the interpretation section and the basis of a synoptic palaeovalley transect.

Outcrop 1: plateau (north)

The northernmost of our outcrops comprises a flat-topped hill (Fig. 1), for which we present a map of deformation structures within the sedimentary rocks (Fig. 2), a sedimentary log through the succession (Fig. 3a) and a series of outcrop photographs (Fig. 4). At ground level a heterolithic assemblage of intercalated fine- to medium-grained sandstones and mudstones is preserved in an outcrop approximately 25 m high. Sandstone beds vary from a few centimetres to a few metres thick (Fig. 4a), and sandstones rest sharply on underlying mudstones (Fig. 4b), locally with downcutting relationships. On bedding surfaces primary current lineations and tool marks including bounce marks (Fig. 4c) are preserved. In map view (UAV orthomosaic), a complex suite of deformation structures can be mapped, including folds (both antiforms and synforms), many of which are rootless, together with evidence of fault offset. The trend of the hinge zones shows great complexity. The strike of fold limbs and fold axial traces is recorded on rose diagrams (Fig. 2). The strike of bedding (n = 255) shows little overall trend, although clusters of north–south and east–west beds are recognized. Fold axial trends show weak east–west and north–south clusters. In addition, beds are locally offset by faults measuring up to 10 m long. These are associated with fault bend deformation of the affected beds. These are modified into a spectrum of fold geometries, including variably upright to recumbent, open to isoclinal structures, as well as sheath folds (Fig. 3a). Where thicker sandstone intervals or beds occur, well-expressed deformation bands cross-cut the fold structures (Fig. 4a). Within individual beds, intrabed deformation including centimetre-scale duplexes (Fig. 4b) is observed. The sandstones are massive to stratified and show evidence of normal grading (Fig. 4d). In general, pinching and swelling of laminae at the margins of intrabed folds is commonplace.

Outcrop 2: Impala

The Impala section (Figs 1 and 3b) comprises a 13 m thick, sandstone-dominated succession. When examined from a 100 m distance at ground level, the outcrop appears to have an asymmetrical cross-sectional profile with a smooth upper surface (Fig. 5a). From UAV data, the outcrop reveals a rounded profile, slightly elongated in a north–south direction (Fig. 5b). It should be noted that these data do not reveal any evidence for consistent or convincing parallel structures such as soft-sediment striations or flutes seen on other sandstone substrates beneath Dwyka diamictite (Le Heron et al. 2019). Lithologically, the outcrop comprises interbedded medium- to fine-grained sandstone with gravelly sandstone intervals and scattered pebble-rich zones. Its well-bedded character is divisible into 1–4 m thick packages or sets between which complex sets of humpback dunes (Lang and Winsemann 2013), wavy bedding, trough cross-bedding and planar cross-bedding are recognized. At the outcrop, the set boundaries (foresets) dip at approximately 20° (parallel to the top of the outcrop), and antidunes and scour-like chute and pool deposits (Fralick 1999; Lang et al. 2021) punctuate the succession (Fig. 5c and d).

Outcrop 3: roadside (south)

The southernmost section (Figs 1 and 3b) is the most lithologically diverse of the studied outcrops. From the base upwards, we recognize a heterolithic facies association consisting of lonestone-bearing shale (Fig. 6a and b) punctuated by thin (5 cm) sandstone beds. These deposits are directly overlain by a diamictite facies association that includes 1–3 m diameter boulders (Fig. 6c) that are embedded within a sandy diamictite matrix. In detail, the massive diamictites are dominated by large (0.5 m to greater than 1 m diameter) boulders of red siltstone and red sandstone, in addition to boulders of a similar diameter consisting of grey paragneiss. In these cases, the boulders are subrounded to rounded, exhibit moderate to excellent evidence for striations, and demonstrate elongation of the a-axis with respect to the b-axis. Striations cross-cut on the surface of the boulders but predominantly trend parallel to the long (a) axis. Together, the red siltstone, sandstone and gneiss boulders account for approximately 50% of this grain size fraction, with weathered dolostones, limestones and cherts, all typically angular to subrounded in shape, interpreted to derive from the Otavi Group (Miller 2008) comprising the remaining fraction. The diamictite matrix shows evidence for extensive deformation between and beneath the boulders (Fig. 6d). Where the diamictite matrix contains fine- to medium-grained sandstone, well-developed deformation bands are preserved (Fig. 6e). In places, the diamictite matrix is intensely weathered, leaving boulder-rich scree (Fig. 7a). Directly above, 3 m of gravelly sandstone is exposed (Fig. 3c), with basal layers incorporating striated boulders (Fig. 7b). Here, the gravelly sandstone facies association comprises 0.5–1.5 m thick trough cross-strata, with foresets typically developed in poorly sorted gravel-rich to pebbly sandstones. At the top of the logged succession, these are in turn capped by a second occurrence of the diamictite facies association (Fig. 3c). However, these in situ deposits are overlain by recessive, poorly exposed and highly disaggregated carbonate beds (Fig. 3d) that contain striated clasts (Fig. 3e). The high resolution of the UAV imagery (Fig. 8) allowed the orientation of 55 boulder a-axes to be measured. Although the individual boulders are too small to be observed at the map scale, an inset on the orthomosaic image shows how the mapped boulders are embedded within the diamictite matrix as opposed to being weathered out on the desert surface (Fig. 8). This, together with extensive ground truthing, provides confidence in the dataset we present. It should be noted that lines represent the orientation of the boulder a-axes but their length is exaggerated to be visible on the map (‘Interpretation’, Fig. 8).

In the following, we proceed on a location-by-location interpretation of the sedimentary rocks before attempting to integrate these in terms of a synoptic model of fjord evolution. The justification for this approach is that the isolated nature of the exposures means that any correlation (Fig. 3) is based on several interpretational caveats. At Outcrop 1, the heterolithic facies association, although intensely deformed, is interpreted to record an ice-marginal depositional environment that witnessed ice-rafting, gravitational flow processes and local bottom current reworking within the fjord. Subsequently, a major phase of soft-sediment deformation overprinted these deposits, which we elaborate upon below. The presence of lonestones, warping the laminae of recessive shales, allows these to be interpreted as dropstones and thus points to ice-rafting. The presence of graded beds, in conjunction with bounce marks on the sole of beds and primary current lineations testifies to turbidite deposition. These interpretations are comparable with those of Rosa et al. (2023), where based on sections 5–10 km north of Outcrop 1, they recognized graded beds, sole marks, amalgamated sandstone beds and diamictites, which they interpreted as a complex mix of turbidites and channelized debrites.

In the northern Orutanda fjord, Rosa et al. (2023) interpreted fjord-side and fjord-head deltas (depending on the position of individual sections with respect to the palaeovalley sides), which also contained a suite of soft-sediment deformation structures, but which were interpreted as gravitational collapse structures. The sense of displacement along reverse faults (n = 3) together with fold axes (n = 5) were used to argue for a SE-oriented slope for one of their measured sections (Rosa et al. 2023, fig. 4, their location 5). Although we have 255 bedding strike measurements (Fig. 2) it should be noted that these are 2D data, unlike the 3D data of Rosa et al. (2023). The orientation of the beds and fold axes is highly complex, and apart from a weak north–south clustering trend, there is little evidence of simple shear involved in the deformation process. The development of minor throughgoing faults, one of which has a sinistral sense of displacement, may explain the secondary east–west and NW–SE strike orientations on the data through a fault-bend process. The wide scatter of fold axes (Fig. 2) also underscores the absence of pure shear involved in the deformation process. In mass-transport complexes (MTCs) produced through delta-slope failure or other subaqueous slope, deposits are complex and often associated with cogenetic debrites and turbidites. Despite this, fold orientation and vergence is often consistent (Sobiesiak et al. 2016, fig. 16), because simple shear is dominant in the downslope movement of materials. This tends to apply in both the proximal (upslope) and distal (downslope) sectors of MTCs (Assis et al. 2024, figs 5 and 13).

If an MTC origin is put to one side, the issue of subglacial deformation is complex, and to illustrate this point it is worth drawing on Late Ordovician analogues from South Africa. This is because over an outcrop belt stretching over 200 km of the West Cape Fold Belt, the glaciogenic origin of the so-called ‘Fold Zone’ is not contested per se, but the mode of deformation in the subglacial environment has been vigorously debated. Earlier models viewed this as a subglacial shear zone produced by the flow of Hirnantian ice sheets, which applied simple shear to a soft, deformable bed (Blignault 1970), but Backeberg and Rowe (2009) challenged this on account of the lack of asymmetry of the fold structures, together with complete enclosure of diamictites of the Sneeukop Member of the Pakhuis Formation as ellipsoidal synforms enclosed within the pre-glacial, paralic Peninsula Formation. Consequently, an origin purely through gravitational instability (load casts) was proposed, an interpretation countered the following year by Blignault and Theron (2010), who further promulgated the thesis of Blignault (1970) with new maps, integrated with the addition of flute and striation data. For similar reasons to the MTC interpretation, we cannot support a subglacial simple shear model similar to that of Blignault and Theron (2010).

Collectively, we propose a gravity spreading mechanism (Owen 2003) caused by the weight of overlying ice in a comparable model to that of Backeberg and Rowe (2009) for Late Ordovician rocks. In this model, with σ1 vertical, complex and non-vergent fold structures are produced. Other evidence for a subglacial origin for the deformation structures is provided by the deformation bands. These result from sediment compaction and are recognized in a range of settings (Fossen et al. 2007), but in the Paleozoic record are dominantly found in association with soft-sediment striated surfaces and subglacial shear zones (Le Heron et al. 2020). Furthermore, Melvin (2019) described ‘tightly folded and stretched sandstones’ as part of a major assemblage of subglacial deformation structures interpreted to have originated through a combination of simple shear and loading and that closely resemble the structures shown in Figure 4a. Comparable folds to those described herein, also overprinted with deformation bands, were described from Mariental, southern Namibia, where a subglacial assemblage was also interpreted (Le Heron et al. 2022b).

At Outcrop 2 (Impala), the stacked gravelly sandstones are interpreted as the deposits of a glaciofluvial delta that was built from sustained, somewhat energetic flows. The presence of humpback dunes testifies to supercritical flow conditions and the development of antidunes, with the chute and pool structures indicative of a hydraulic jump within the system (Lang and Winsemann 2013). Such deposits testify to sustained delivery of sandy material to the basin, possibly in response to a major flood event such as a glacial lake outburst flood (Ghienne et al. 2010). Similar facies are recognized throughout the glacial sedimentary record, from the Ordovician of northern Africa to the Quaternary of northern Germany (Lang et al. 2021), and from the LPIA deglacial record of the SE Paraná Basin (Brazil) where a similar assemblage has recently been recognized (de Souza et al. 2023). In this context, the separation of bedsets by gravelly intervals may suggest somewhat flashy, pulsed or repeated meltwater delivery events indicating oscillating flow velocities. In this context, the present-day northern slope of the outcrop (Fig. 5c) is tentatively interpreted as a relict delta-slope surface. The intrabed folding is interpreted as localized dewatering in response to rapid sedimentation. The presence of trough cross-beds and planar cross-beds within the sequence records the establishment of 2D and 3D dunes under relatively more quiescent conditions.

The southernmost of our sections, Outcrop 3, is interpreted to record a cycle of ice advance into the palaeovalley, deposition of subglacial diamictite, an ice-marginal fluvial deposit and readvance of the glacier over the sandur to deposit a second subglacial diamictite. This interpretation is based on the stratigraphic appearance of the facies associations (Fig. 3c). Based on Quaternary comparisons, readvance of glaciers on top of delta deposits may result in the delta deposits remaining fully undisturbed, even if a drumlinized till is deposited directly on top (Le Heron et al. 2023). Dropstones (Fig. 6b) in shales indicate the initial existence of a body of standing water into which the glacier advanced to deposit the subglacial tillite (Fig. 6c). The subglacial interpretation is well supported by abundant striated clasts, an assemblage of soft-sediment deformation structures in the matrix including deformation bands (see above) and a strongly preferred a-axis orientation of the boulders at the outcrop scale. This last observation testifies to a linear fabric within the diamictite, best explained through the development of a deforming bed. In contrast to the example of the gravelly sandstone facies association exposed at Impala towards the north, that at Outcrop 3 reveals typical subaqueous dune forms of the type that are associated with steady-state discharge on a modern sandur (Lang et al. 2021). The presence of a second occurrence of diamictite on top of these gravelly sandstones implies readvance of the glacier to deposit a second interval of subglacial tillite. It is noteworthy that the aerial photographs of the Impala outcrop reveal the onlapping of the diamictite facies association onto the gravelly sandstone facies association, which may indicate a similar stratigraphic context, and which we explore below. It should also be noted that the carbonate beds on the top of Outcrop 3 are considered to be unrelated to LPIA deposition: it is more parsimonious to interpret these as caliche deposits.

Key questions emerge from our investigation of the outer Orutanda Fjord, including (1) implications of multiple diamictite intervals for the models of a simple retreat sequence inside the northern Namibian fjords (see Dietrich et al. 2021), (2) stratigraphic correlation of lithologically diverse and stratigraphically complex deposits, (3) the significance of soft-sediment deformation and (4) wider implications for Paleozoic fjord fills in general. In the following, we tackle each of these points in turn.

Dietrich et al. (2021) were the first to interpret the Namibian palaeovalleys as fjords. Building on the glacial valley interpretations of Martin (1981), they were able to show a relatively simple succession in the Gomatum palaeovalley. This consisted of a striated and polished bedrock, a basal diamictite, an ice contact fan (sandstones and conglomerates akin to the fluvial sandstones and gravelly sandstone facies association described herein), passing upward through intertidal to offshore deposits (shale). Based on our Outcrop 3, a second phase of ice advance would be required to explain the second diamictite in the succession. This is noteworthy, because it implies that the recession of glaciers from the Orutanda fjords was punctuated by either a stillstand or minor readvances. A similar conclusion was reached by Dietrich et al. (2021) in the Gomatum palaeovalley, whereby perched morainal banks and ice contact fans at the valley sides were also recognized. Thus, it could be proposed that the pattern of recession in Orutanda followed a comparable, perhaps synchronous evolution and that these outlet glaciers were in phase with one another. This is significant, because our study area represents an outer fjord region: study of the inner fjord region by Rosa et al. (2023) did not recognize minor readvances punctuating the overall recession. Nevertheless, to explain the stratigraphy of the succession at Outcrop 3, we argue that such a readvance must have occurred as the ice flowed basinward down the palaeovalley through the inner fjord to the outer fjord. This reasoning also opens up the possibility that significant subglacial deformation occurred, after the bedrock palaeovalley had been excavated subglacially, within the valley-filling sediment. We have already shown that this was the case to explain the first diamictite in the succession at Outcrop 3.

To address the issue of stratigraphy, we propose a possible correlation (Fig. 9). In this scheme, the dropstone-bearing heterolithic facies association is considered laterally equivalent at both Outcrop 1 and 3. Therefore, the stratigraphically lowest diamictite is recorded at the base of the succession at Outcrop 1, and the remaining two diamictites at Outcrop 3 would represent two successively higher stratigraphic occurrences. We interpret the diamictites to represent subglacial deposits (tillite). Thus, at least three glacial readvances, of unknown regional significance, are recorded within the Orutanda fjord (Fig. 9). Nevertheless, caution is advised here, because significant debate is recorded elsewhere in the literature on glacial palaeovalley fills as to whether such sequences stratigraphically repeat, are deformed or are miscorrelated (Ghienne et al. 2003). Nevertheless, in a process-based model, a shallowing upward succession, or at least a progradational motif, can be inferred at the top of the heterolithic facies association at both Outcrop 1 and 3 owing to the overall coarsening upward character, allied with thicker and more frequent sandstone beds toward the top of these deposits. Given this, it is logical to correlate the gravelly sandstone of both Outcrop 2 and 3, and to place this stratigraphically above the deposits of Outcrop 1 as they represent a coarser-grained, shallow-water to emergent deposit. Based on this argument, we propose that the cross-bedded facies of the gravelly sandstone facies association (Outcrop 3) pass laterally into the supercritical flow deposits (humpback dunes, chute and pool structures) at Impala and Outcrop 2 (Fig. 9). Furthermore, given this transition, we identify the cross-bedded deposits of Outcrop 3 as probable glaciofluvial delta topsets, with the foresets of the delta recorded at Impala. Thus, a process model linking the separate sections rather than a lithostratigraphic model is proposed.

Our study of the southern part of the Orutanda fjord sheds further light on the complexity of the fill in LPIA fjords across Gondwana. Prior to the last few years, investigation of LPIA fjords had largely focused on a long tradition of study in South America. From the perspective of stratigraphic architecture, Kneller et al. (2004) focused on the role of ‘catastrophic’ sedimentation (i.e. slope collapse and mass flows) in a fjord within the San Juan Province of Argentina. Dykstra et al. (2006) developed this theme with detailed stratigraphic analysis of the glacial to postglacial sequence. Tedesco et al. (2016) investigated the geomorphology as well as the sedimentary fill of a fjord example in the southernmost Paraná Basin (Brazil). Dropstone-bearing shales and diamictites comparable with those of the Orutanda outer fjord were also reported from the base of a 400 m deep palaeovalley, although the surrounding area was estimated to have experienced up to 4 km of erosion based on apatite fission-track analysis. From their stratigraphic and sedimentological analysis of the Talacasto Palaeofjord in the Paganzo Basin (west Argentina), Aquino et al. (2014) demonstrated that this palaeovalley contains two high-frequency glacial cycles. Extensive soft-sediment deformation structures (Aquino et al. 2014, fig. 21), highly comparable with those exposed at our Outcrop 1, were interpreted to have resulted from the readvance of the Talacasto glacier over the glacial sediments. For the Vichigasta palaeovalley in the same basin, Valdez Buso et al. (2021) provided a terrain analysis based on a 12.5 m resolution DEM, and then described a comparatively simple three-part fill for the palaeovalley. Again, comparable dropstone-bearing lacustrine shales and diamictites at the base, with local mass-transport complexes and gravitational reworking, pass upward to outwash deltas and sandar representing stillstand deposits (comparable with those described at our outcrops 2 and 3) and finally to transgressive shales. This simple comparison with other fjord systems around Gondwana demonstrates that each palaeovalley may reveal significant complexity, but the recurring theme of glacial readvance into the deeply carved palaeovalleys, deforming previously deposited sediments, remains. An additional issue that requires future research may be preservation bias with respect to precise position of a given section within a fjord. In the outer part of the Orutanda Fjord, for example, we have revealed a somewhat more complex stratigraphy than that revealed within the inner fjord system. Factors that might potentially come into play are enhanced accommodation space in the outer fjord compared with the inner fjord.

Investigation of the Orutanda Fjord of northern Namibia reveals a complex, but convincing, record of Late Paleozoic glaciation and together with recent previous studies (Dietrich et al. 2021; Rosa et al. 2023) adds significantly to our understanding of the evolution of deep time fjords, with the prior focus having mostly been on palaeovalleys in South America. Complex stratigraphy within the infill of the valley may be understood by a process-based (rather than lithostratigraphic) approach. Although not conclusive, we tentatively propose that complex soft-sediment deformation probably records a subglacial origin, rather than originating through a slope collapse mechanism in the outer part of the Orutanda Fjord. It should be noted that this does not militate against the interpretation of slope collapse processes operating at the margins of the inner Orutanda Fjord as interpreted by Rosa et al. (2023), but is consistent with the recognition of similar suites of structures in other LPIA fjords where glacial readvance and subglacial deformation is interpreted (e.g. Aquino et al. 2014). In terms of the sedimentology, multiple levels of diamictites are interpreted as the product of minor readvances or stillstands, comparable with deformed morainal banks interpreted by Dietrich et al. (2021) from the Gomatum palaeovalley to the south. Although not all LPIA palaeovalleys may strictly be of glacial origin (see Fedorchuk et al. 2019), at its heart, this acts as a reminder of the complexity of the ancient glacial record and the interpretational challenges it brings, even with the advent of modern digital techniques. Moreover, it underscores how the remaining open questions (stratigraphic correlation, origins of soft-sediment deformation structures, synchroneity or otherwise of glacial recession) remain a core research theme in studies of deep time glaciation.

We are grateful for the positive and constructive comments of N. Fedorchuk, M. Busfield and an anonymous reviewer. The suggestions were very helpful and thought provoking, and helped improve the paper. We are also grateful for the editorial work of S. Boulton.

DPLH: conceptualization (lead), formal analysis (lead), funding acquisition (lead), methodology (lead), resources (lead), writing – original draft (lead), writing – review & editing (lead); RW: investigation (supporting), software (equal), visualization (equal); PMO: investigation (supporting); CK: investigation (supporting), methodology (supporting); AN: investigation (supporting), project administration (equal).

This work was financed by the University of Vienna.

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

The data used to write this paper were collected during fieldwork in February 2023. The drone data can be made available by written request to the authors.