Regional setting

The Torridonian succession has been a subject of study for many decades (e.g. Peach et al. 1907; Phemister 1948; Lawson 1965; Rodd & Stewart 1992; Stewart & Donnellan 1992; Nicholson 1993; Turnbull et al. 1996; Van De Kamp & Leake 1997; Young 1999; Rainbird et al. 2001; Williams 2001; Stewart 2002; Jonk et al. 2004). Stewart (2002) has compiled an exhaustive synthesis of previous work on the Torridonian rocks, including detailed descriptions of many field locations. Like many late Precambrian successions worldwide, this succession is poorly dated. However, field relationships allow a relative stratigraphy to be constructed, although only broad absolute age brackets have been established. Palaeomagnetic data for parts of the succession imply a Mesoproterozoic age for deposition (Irving & Runcorn 1957; Stewart & Irving 1974; Smith et al. 1983; Williams & Schmidt 1997; Darabi & Piper 2004) and detrital zircon studies have provided maximum depositional ages (e.g. Rainbird et al. 2001).

Of the three groups, the Stoer is considered the oldest. Exposures of the Stoer rocks are limited to a discontinuous belt from Stoer in the north to Loch Maree in the south (Fig. 1a). They rest nonconformably on Archaean–Palaeoproterozoic Lewisian basement and their upper contact is marked by an angular unconformity with the overlying Torridon Group rocks (Lawson 1965; Stewart 1966, 1969). The succession can be as much as c. 2 km thick (Stewart 2002). It is composed of fluvial–alluvial sandstones and sedimentary breccias, with minor lacustrine mudstones and rare carbonates (Stewart 1988a, 2002). A thin volcanogenic deposit (Stac Fada Member; e.g. Lawson 1972; Young 2002) makes a good marker unit for correlation. A maximum age for the Stoer Group is provided by a Rb–Sr age of 1187 ± 35 Ma on chloritized biotite from a gneissic boulder in the lowermost Stoer beds (Moorbath et al. 1967) and a Pb–Pb ‘depositional’ age of 1199 ± 70 Ma on a thin stromatolitic bed low in the group (Turnbull et al. 1996). The Stoer's detrital zircon age profile (Rogers et al. 1990; Rainbird et al. 2001) and composition (Young 1999; Stewart 2002) indicate that sediment was sourced from the underlying Lewisian basement. Thus, given existing data, the depositional age of the Stoer Group is c. 1200 Ma.

The type area of the Sleat Group is the eponymous peninsula of eastern Skye. Outcrops are limited to this island and to two narrow north–south-trending belts on the adjacent Scottish mainland (Fig. 1a). It is as much as c. 3 km in thickness and consists of fine to locally coarse feldspathic and quartzitic sandstones with varying proportions of shale. The overall vertical facies trend defines a fining-upward succession and the rocks have been interpreted as recording non-marine depositional settings (Stewart 2002). No radiometric ages have been obtained for these rocks. The Sleat and Stoer rocks do not occur in outcrop together, thus their exact stratigraphic relationship is not known. However, because the Sleat Group is assumed to be conformable beneath Torridon rocks (Applecross Formation) on Skye, its depositional age is generally regarded to be similar to that of the Torridon (see below) and thus younger than the Stoer Group (Stewart 2002).

The Torridon Group, the collective term for the Diabaig, Applecross, Aultbea and Cailleach Head Formations, has a combined thickness of 6–7 km. The basal Diabaig Formation is only preserved locally within palaeo-topographic lows developed in Lewisian basement. The Applecross and Aultbea Formations are more widespread, whereas the Cailleach Head Formation is primarily limited to a single, fault-bounded exposure at the western tip of the eponymous peninsula (Fig. 1a). These units were deposited in a variety of alluvial–fluvial–lacustrine settings with the bulk of the Group consisting of trough and planar cross-bedded, medium to pebbly, arkosic sandstones of the Applecross and Aultbea Formations (Selley 1965; Nicholson 1993; Williams 1966, 2001; Stewart 2002). Rocks from the Torridon Group have yielded detrital muscovite K–Ar and diagenetic Rb–Sr ages ranging from 1168 ± 30 to 997 ± 39 Ma (Moorbath et al. 1967; Turnbull et al. 1996) and the youngest detrital zircon grain has an age of 1060 ± 18 Ma (Rainbird et al. 2001). Diagenetic phosphate concretions in the Diabaig Formation yield a whole-rock Rb–Sr isochron age of 994 ± 48 Ma and a Pb–Pb isochron age of 951 ± 120 Ma (Turnbull et al. 1996). Hence, the depositional age of the Torridon is probably earliest Neoproterozoic. In addition, provenances other than the Scottish foreland basement rocks are required to explain the presence of clasts composed of lithologies not found in the Scottish Highlands (i.e. quartz–tourmaline, quartz–fuchsite schist, pure and iron-rich metaquartzite, fine-grained igneous rocks comparable with porphyritic rhyolite and acid tuffs, and a large variety of orthoquartzite and chert; Williams 1969) and the diverse detrital zircon age population (Rainbird et al. 2001).

New field observations

Two aspects of Torridonian geology are striking. The first is the lithological monotony of the Applecross–Aultbea Formations, essentially cross-bedded arkosic sandstone and locally pebbly sandstone preserved along 150 km of strike exposure and up to several kilometres in stratigraphic thickness. Detailed facies and palaeocurrent studies (e.g. Williams 2001) have shown diverging sediment transport paths locally, but this is only a moderate variant on the overall sedimentological theme, namely, broadly ESE-directed sediment transport recording fluvial–alluvial deposition over the entire outcrop belt. The second aspect, in contrast to above, is the lithological diversity exhibited by the other Torridonian units, particularly the Sleat Group and the Diabaig Formation, which are all marked by lateral and/or vertical facies changes (the Cailleach Head Formation is also distinctive but is not discussed here and will be the subject of a forthcoming paper). It is these first-order observations that have driven our curiosity to re-examine these rocks and the nature of the contacts between them.

The contact between the Diabaig and Applecross formations is interpreted to be transitional and thus conformable (Stewart 2002). However, some rather pronounced lithological changes occur across this contact, indicating that it may not be conformable. The contact between the Diabaig and Applecross formations is well exposed on the shore of the eponymous type area (Figs 1a and 2). There, the Diabaig Formation displays an overall coarsening- and thickening-upward trend from dark grey mudstone exhibiting rhythmically alternating and commonly desiccation-cracked laminae into interbedded mudstone and subgreywacke sandstone. Thin, discontinuous beds of coarser-grained sandstones and carbonate rocks occur locally. In places phosphate nodules are developed and these contain acritarchs (e.g. Downie 1962; Leiosphaerid forms were identified by A. Knoll, pers. comm.). The upper part of the succession is marked by an increase in the proportion of sandstone beds, which are of metre-scale thicknesses, have sharp bases, define tabular to broadly lenticular geometries (over tens of metres) and commonly display ripple-drift cross-lamination. Palaeocurrent data from the rippled linsen- and flaser-bedded units indicate that the main flow component was towards the SW (Fig. 2). In many places, the Diabaig Formation exhibits an interesting lateral sedimentological changeability, which is well expressed at the type locality (Fig. 2; see also Stewart 2002). There, the basal unit is a variably developed sedimentary breccia shed off palaeo-highs of the Lewisian basement. The breccia beds grade into tabular, red sandstone and these in turn pass laterally and upward into micaceous siltstone and fine sandstone. At Lower Diabaig, all these facies are sharply overlain by the Applecross Formation. It consists of red–grey medium to pebbly arkosic sandstone (the Allt na Bieste member of Stewart 2002) that erodes into the underlying Diabaig rocks; sandstones above this contact contain abundant, irregularly shaped (‘rip-up’) clasts derived from the Diabaig units. Interestingly, detritus (e.g. quartz, plagioclase, epidote and mica) in the Diabaig rocks can be attributed to mostly local sources whereas numerous varieties in the Applecross sandstones (e.g. quartz–tourmaline, quartz–fuchsite schist, pure and iron-rich metaquartzite, fine-grained igneous rocks comparable with porphyritic rhyolite and acid tuffs, and a large variety of orthoquartzite and chert; Williams 1969) were derived from provenances not known within the foreland area of the Scottish Highlands (Selley 1965; Rodd & Stewart 1992; Stewart 2002). Thus, this contact marks a sharp change in facies and in clast provenance compositions. Additionally, the contact also defines a difference in petrography and diagenetic histories. Rodd & Stewart (1992) and Van De Kamp & Leake (1997) documented differences in the relative abundances of plagioclase (2:1 Diabaig:Applecross), K-feldspar (1:4–1:3) and mica (8:1). Our observations (from samples collected in the vicinity of Loch Torridon) indicate very similar findings of relative abundances of plagioclase (4:1), K-feldspar (1:3) and mica (10:1). At Lower Diabaig, the pore-fill cement is composed predominantly of illite and sericite, with authigenic chlorite (developed partly by the alteration of plagioclase, as indicated by relict feldspar) and as interstitial matrix between framework grains. In contrast, the Applecross sandstone cement consists mainly of syntaxial overgrowths of quartz and K-feldspar, with minor albite. Diagenetic chlorite and limonitic oxide are present in the Diabaig units whereas these minerals are absent in the adjacent Applecross sandstone, despite it containing the correct mineral constituents for such a reaction.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

 Geological map of the Diabaig area (see Fig. 1a for location). Palaeocurrent directions of the various mapped units are shown and include: (i) trough cross-stratified sandstone; (ii) nested trough cross-bedded sandstone; (iii) ripple-drift cross lamination. Stratigraphic sections show (a) the upper Diabaig–lower Applecross formational contact (modified from Rodd & Stewart 1992) and (b) a typical log through the Applecross Formation.

As well as the dissimilarity of facies and composition between the Applecross and Diabaig units, the contact between these two formations commonly displays erosive channelling and marks a sharp increase in grain size, and in many places a slight angular discordance between the two formations can be discerned. For example, at the Diabaig type locality, and at Inveralligin, the discordance is 4–7° across the contact (Fig. 2). One could argue that this may reflect a difference in the initial dip of the Diabaig and Applecross rocks and/or differential compaction. However, the fact that the basal Applecross sandstones typically rest sharply and erosively across all facies of the Diabaig Formation, in combination with the differences noted above, leads us to question the assumption of a conformable, transitional contact between these formations.

The Sleat Group is best developed on Skye (Fig. 1a), where it consists of four formations: in ascending order, the Rubha Gail, Loch na Dal, Beinn na Seamraig and Kinloch Formations (Fig. 1b; Sutton & Watson 1964; Stewart 1988b, 1991, 2002). The basal ‘Torridon’ unit on Skye is the Applecross Formation. The contact between the Sleat and Torridon rocks is nowhere well exposed but has nevertheless been assumed to be conformable (Stewart 1988b, 2002). This is questionable for a number of reasons. The Applecross and Sleat rocks display a number of facies differences (e.g. coarser v. finer grained, thicker, lenticular beds v. thinner, more tabular beds, absence v. presence of interbedded shale, respectively) and the contact between the two groups is discordant, an aspect noted decades ago in that the strike of beds of the upper Sleat rocks is ‘oblique to the general north-easterly trend of the Applecross/Kinloch boundary’ (Sutton & Watson 1964, p. 266). These features were explained as representing a lateral facies change (e.g. Stewart 2002), but such an inference is unsupportable given the genuinely different lithologies, compositions and inferred palaeoenvironmental settings between the basal Applecross and upper Sleat units. Furthermore, even though the general dip of both the Sleat and Applecross units on Skye is moderately westward, there is an overall difference in attitude between these two units (Fig. 3). Consequently, like the Diabaig–Applecross boundary, the Sleat–Applecross contact is not unambiguously conformable. In fact, taken objectively, these contacts would appear to be potential candidates for unconformities.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

 Simplified geological map of the Sleat Group, Sleat Peninsula, Skye (see Fig. 1a for location).

Detrital zircon analyses

Detrital zircon age profiling has become a widely used stratigraphic tool of choice to assess not only provenance issues but also as a means of stratigraphic fingerprinting of units to independently test correlations and depositional frameworks (Rainbird et al. 2001; Cawood et al. 2003; Kirkland et al. 2007). To apply this tool, we have collected samples from Torridonian rocks to compare and contrast their detrital zircon age distributions. Rogers et al. (1990) and Rainbird et al. (2001) have provided an ample dataset for the Stoer and Applecross–Aultbea units, thus we focused our efforts on the other units.

Discussion

The Sleat Group rocks yield two dissimilar detrital zircon profiles. The lower Sleat profile (Rubha Guail Formation) displays distinct Archaean and late Palaeoproterozoic clusters; in many respects it is similar to the Stoer Group profile. In contrast, the upper Sleat profile (Loch na Dal to Kinloch Formations) is characterized by a dominance of late Palaeoproterozoic and early Mesoproterozoic zircons. Archaean detrital grains are either absent (the Kinloch sample) or constitute only a small percent of the final spectrum (the Loch na Dal sample). It is not surprising that the ‘lower’ Sleat profile is dominated by Lewisian-aged zircons given its ‘local’ derivation from Lewisian basement. What is interesting, however, is that the ‘upper Sleat’ zircons are strongly dominated by late Palaeoproterozoic ages with the youngest grain being of mid-Mesoproterozoic age, 1247 ± 34 Ma. This is not the case for the overlying Torridon rocks, in which grains younger than c. 1.2 Ga are present. Thus, the age distribution of detrital zircons from the Sleat Group is substantially different from that of the Applecross Formation, implying that the two units were derived from different source areas. Furthermore, there is a c. 200 Ma age difference between the youngest detrital grains in the Sleat rocks (1247 ± 34 Ma) and the Applecross Formation (1060 ± 18 Ma). Although this is not an absolute reflection in difference of depositional ages (especially considering the low number of the Sleat analyses), when combined with the discordant attitudes and facies differences between the two formations, it is consistent with our interpretation that the base of the Applecross Formation is an unconformity.

The Diabaig detrital zircons yield Archaean (2.9–2.7 Ga) and Palaeoproterozoic (1.9–1.6 Ga) clusters. The youngest grain has an age of c. 1090 Ma. Given the small sample size for the Diabaig, we are cautious in drawing strong conclusions. Nevertheless, unlike the Sleat profile, the Diabaig profile is similar to that of the Applecross Formation. The Diabaig Formation is generally inferred to be a ‘distal’ portion of the Applecross depositional system, thus it follows logically that the age profiles should be similar. However, as noted above, the base of the Applecross marks a surface of erosion that emplaces relatively coarse, braided fluvial rocks onto much finer-grained lacustrine and lacustrine-margin Diabaig strata. Thus, this contact is a prime candidate for being a major sequence boundary and suggests that the Applecross depositional system was probably genetically distinct from that of the Diabaig.

Rift basin settings for each of the Torridonian Groups have been championed since the 1950s (e.g. Pavlovsky 1958; Stewart 2002). It is noteworthy, however, that not a single basin-bounding extensional fault has been identified in outcrop; all such faults are purely hypothetical or inferred to be hidden beneath the Minch seaway between the Scottish mainland and Outer Hebrides. Seismic lines have been suggested to image potential candidates (e.g. Smythe et al. 1982; Blundell et al. 1985; Stein & Blundell 1990) but it is interesting to note that these inferred basin-bounding faults are shown cutting the entire Torridon succession; thus, for the most part they must post-date those rocks. Nevertheless, focusing on what can be observed, two aspects are noteworthy: (1) with the exception of the thin Stac Fada Member of the Stoer Group, no rift-related volcanism or significant volcaniclastic detritus is known within the Torridonian; (2) the main part of the Torridon succession exhibits a remarkable consistency of facies character along its >150 km of strike length and >4 km of stratigraphic thickness. With respect to the former, in the mid-continent Mesoproterozoic rift system exposed in the Lake Superior Region, USA, volcanic rocks are as much as c. 20 km thick (Ojakangas et al. 2001). Hence, it is justified to critically re-evaluate the evidence upon which interpretations for rifting are based.

The Stoer Group is interpreted as recording sedimentation in a rift graben with intermittently active margin faults (Stewart 1988a; Van De Kamp & Leake 1997). Beacom et al. (1999) proposed a variant to this by suggesting that rifting (and the subsequent deposition of the Stoer) occurred under broadly north–south-oriented dextral transtension. Whichever is more correct, what is immediately obvious is that the facies characteristics of the Stoer rocks support an interpretation for sedimentation during extensional tectonism; laterally and vertically variable coarse- to fine-grained facies, clasts mostly locally derived, palaeocurrent indicators displaying reversals and local variability, interfingering between alluvial–fluvial and lacustrine units and, albeit thin, the presence of a mafic volcanogenic unit. These features are similar to those documented in many sedimentary basins elsewhere that formed in continental extensional settings (e.g. Leeder & Gawthorpe 1987; Schlische 1991; Friedmann & Burbank 1995). In that the Stoer basin(s) was developed on Lewisian basement, the detrital zircon data yield a profile reflecting characteristic Archaean and Palaeoproterozoic clusters. These data are consistent with previously obtained geochemical data indicating that sediment was derived from local provenances (Stewart 2002, and references therein).

The Sleat Group rocks similarly have facies characteristics typical of sedimentary successions formed in continental extensional basins; namely, rather rapid lateral and vertical changes between coarser- and finer-grained lithologies representing various components of alluvial–fluvial–lacustrine systems (e.g. Stewart 2002). The detrital zircon profile of the lower Sleat rocks fits a Lewisian derivation but the late Palaeoproteorozoic–early Mesoproterozoic ages that dominate the upper Sleat profile show that source areas outside the Scottish foreland were contributing detritus during the later stages of basinal evolution. This supports the inference of previous workers, who suggested that basinal overstepping occurred following an initial phase of Sleat ‘rifting’ (e.g. Nicholson 1993).

In striking contrast is the Torridon Group, in particular the Applecross–Aultbea rocks. These rocks do not display features typical of deposits in active rift basins developed in continental crust, a fact pointed out over a decade ago by Nicholson (1993). Their stratigraphic monotony of kilometre after kilometre of cross-bedded arkoses and pebbly arkoses with relatively low palaeocurrent variance for thousands of metres of vertical section is not a facies feature readily attributable to active continental rifting and half-graben basins. Thus, the puzzling observation is that, even though all of the Torridonian Groups were purportedly deposited in similar rift basin settings (e.g. Stewart 1982, 1988a, 1988b, 2002), the Torridon Group is unlike the Sleat and Stoer Groups. Jonk et al. (2004) in a study of Lewisian- and Torridonian-hosted clastic dykes at Gairloch, speculated that the dykes formed under ESE–WNW-directed extension and that they developed during deposition of the upper Torridon Group. They used this evidence to promote the idea of north–south- to NE–SW-trending faults in the area, from the onset of Torridonian sedimentation (Soper & England 1995; Beacom et al. 1999) and throughout Torridonian times. However, no evidence of significant extensional faulting is recorded in the Applecross Formations. Thus, our conclusion, like that of Nicholson (1993) and Rainbird et al. (2001), is that there is no objective evidence to support the Applecross–Aultbea sequence being a rift basin fill. Instead, the facies characteristics of these rocks are better interpreted as a fluvial–alluvial apron whose genesis can be attributed to the erosional denudation of the Grenville orogenic highlands, i.e. a non-marine molasse. Such an interpretation is also consistent with the detrital zircon profile.

Advocates of a rift setting for the Torridonian Group are rather dismissive of a molasse interpretation (e.g. Stewart 2002). However, as discussed above, many of the arguments for a solely rift-related genesis are equivocal and criticisms against a molasse-style setting are not convincing. For example, the most common criticism is that, given that the Grenville orogenic belt lay SW of the Torridonian outcrop, the SE-directed palaeocurrent trends of the Applecross–Aultbea rocks are incompatible as a molasse (Stewart 2002, and references therein). This is readily dismissed. Many fluvial systems (modern and ancient) flow initially orthogonal to the orogenic front and then veer towards basin axial (or orogenic-front-parallel) directions. For the Torridon example, the SE-directed Applecross–Aultbea system would represent the axial component. At this point, it is worth noting that the preserved width of the Torridon outcrop belt (a few tens of kilometres) is miniscule compared with the scale of the depositional systems that the Applecross–Aultbea rocks are inferred to represent, thus palaeogeographical reconstructions are inescapably speculative. The inferred timing of deposition also casts doubt on the interpretation of a rift-dominated tectonic setting for those rocks. Although a direct depositional age is lacking, given the bracketing age constraints provided by detrital zircon and diagenetic ages, the Torridon was probably deposited c. 1000 Ma. Global tectonic reconstructions show that this was a time near the zenith to waning phases of supercontinental amalgamation (Grenville Orogeny) and that break-up did not begin until later (e.g. Dalziel 1997; Cawood 2005). Consequently, the interpretation that most objectively fits the available evidence is that the Applecross–Aultbea rocks are a remnant of what would have been a much larger, non-marine molasse shed off the rising eroding Grenville highlands.

Like better-dated and constrained successions of largely non-marine strata elsewhere (both Phanerozoic and Proterozoic examples), the Torridonian successions can now be examined in light of realistic depositional frameworks in which hiatal surfaces are a common component punctuating sequences thousands of metres thick. Our data show that the base of the Applecross succession is best viewed as a compound surface that defines a major unconformity where it rests on the Lewisian, Stoer and Sleat rocks and a less significant hiatal surface (sequence boundary) where it rests on the Diabaig Formation. This indicates that, rather than being a conformable succession, the Torridonian Group is actually a composite sequence of temporally unrelated and genetically distinct units.

Thus, we need to re-evaluate how the units fit within the framework of the late Mesoproterozoic Grenville orogenesis, which is known to have affected the length of eastern Rodinia (Gower 1996; Dalziel 1997; Rivers & Corrigan 2000; Strachan & Holdsworth 2000; Dalziel & Soper 2001; Kirkland et al. 2006, 2007). The Stoer, Sleat and Diabaig rocks record deposition that is genetically unrelated to the Applecross–Aultbea succession. Considering our data in the context of known global tectonic events, the Stoer and Sleat rocks are reasonably interpreted as recording a phase of crustal extension perhaps driven by far-field stresses associated with the c. 1.23–1.18 Ga Elzevirian Orogeny (Gower 1996; Rivers & Corrigan 2000; Gower & Krogh 2002). As argued above, the Applecross–Aultbea rocks are best interpreted as the deposits of a late to post-Grenvillian foreland trunk river system that flowed axially with respect to the orogenic belt (Rainbird et al. 2001). Young (1999) suggested that the deposition of the Applecross–Aultbea rocks might be associated with the collapse of the Grenville orogen. Syn- to post-orogenic extension is inferred to have followed the Grenville orogeny through a combination of orogenic collapse and/or mantle delamination (Streepy et al. 2000), but it remains to be established if the depositional age of the preserved Torridon units is coeval with orogenic collapse.

Conclusion

Our data have shown that the existing stratigraphic framework and the inferred basinal evolution of the Torridonian succession needs to be revised to account for the presence of previously unrecognized and/or under-appreciated unconformities within what have been generally considered to be conformable successions. The most significant of these is the base of the Applecross–Aultbea succession, which is everywhere an unconformity. We propose that a more correct and insightful late Mesoproterozoic to early Neoproterozoic tectonostratigraphic framework of the Highlands is: (1) deposition of the Stoer and Sleat Groups in extension-related sedimentary basins: existing age constraints and the detrital zircon age spectra of these rocks imply derivation from Archaean to early Mesoproterozoic sources with no input from late Mesoproterozoic (‘Grenvillian’) components and it is likely that both groups are pre-1200 Ma in age; (2) deposition of the Diabaig rocks during the latest Mesoproterozoic and/or earliest Neoproterozoic, in lacustrine and lacustrine-margin settings after c. 1090 Ma; (3) deposition of the Applecross–Aultbea succession as a late to post-Grenville Orogeny non-marine molasse that accumulated after c. 1060 Ma. Consequently, each of the major Proterozoic sedimentary successions in the foreland of the Scottish Highlands, the Stoer, Sleat and Torridon, can now be viewed as being bounded by unconformities and temporally and genetically unrelated to each other. In that the detrital zircon profiles do not provide depositional ages it is difficult to estimate the temporal magnitude, but it is plausible that the hiatus along the Sleat–Applecross contact may be of the order of many million years or more, and that of the Diabaig–Applecross as much as a couple of million years. Consequently, the view that the Torridonian rocks record deposition in a suite of long-lived rifts, whereas the rest of the consanguineous Laurentian margin experienced collisional and orogenic episodes (e.g. Rivers & Corrigan 2000), becomes equivocal and in need of reassessment, if not outright abandonment.