U–Pb apatite geochronology is increasingly recognized as a valuable tool for constraining the age of mid-crustal ductile shear zones. The crustal-scale Outer Hebrides Fault Zone (OHFZ) within the Laurentian foreland of the Scottish Caledonides has long been of uncertain age and tectonic significance. Earliest deformation within the OHFZ was associated with top-to-the-NW ductile thrusting that formed a belt of greenschist facies mylonites within host Archean−Paleoproterozoic basement gneisses. Previous estimates for the timing of thrusting vary between c. 1600 Ma and c. 430 Ma. The mylonitic fabrics are defined by a recrystallized assemblage of quartz + albite/oligoclase + sericite + actinolite + epidote + apatite ± calcite, consistent with deformation temperatures of 400–500°C and within the range of reported closure temperatures for Pb diffusion in apatite. U–Pb (LA−ICP−MS) dating of two texturally distinct apatite grain types within the mylonites has yielded ages mostly in the range c. 1100–900 Ma. The OHFZ is therefore interpreted as a Grenville–Sveconorwegian structure that formed during the tripartite collision of Laurentia, Baltica, and Amazonia and the assembly of Rodinia.

Supplementary material: U–Pb isotopic data (Table S1), trace element data (Table S2) and laser ablation spot images are available at https://doi.org/10.6084/m9.figshare.c.7084925

Major ductile shear zones are initiated during regional scale tectonic events and once formed typically exist as long-lived zones of structural weakness which often contain evidence for repeated reactivation (e.g. Grocott 1977; White et al. 1986; Hatcher 2001; Holdsworth et al. 2001). Such structures are therefore of fundamental importance in understanding ancient plate movements and the long-term evolution of the lithosphere. Understanding of the timing, strain-rates and duration of mid-crustal tectonic processes has been enhanced within the last two decades by the application of U–Pb geochronology to a wide-range of accessory minerals (e.g. Parrish 1990; Corfu et al. 1994; Schoene and Bowring 2006; Storey et al. 2007; Darling et al. 2012). The calcium-phosphate mineral apatite is nearly ubiquitous in crystalline rocks of the crust and detritus derived from their denudation (Chew and Spikings 2015) and has a closure temperature to Pb diffusion that ranges between 350 and 575°C (Cherniak et al. 1991; Chew and Spikings 2021). Within low- and medium-grade metamorphic rocks, apatite can yield U concentrations of up to ∼100 ppm (O'Sullivan et al. 2020). With the caveat that efforts to understand the mechanisms for deformation-related Pb loss in apatite remain ongoing (e.g. Kirkland et al. 2018; Odlum et al. 2022), U–Pb apatite geochronology is therefore potentially a powerful tool that can be used to decipher the evolution of mid-crustal shear zones (e.g. Odlum and Stockli 2020; Ribeiro et al. 2020a, b).

The Outer Hebrides Fault Zone (OHFZ) in NW Scotland (Figs 1, 2) is a reactivated, crustal-scale fault developed within Archean–Paleoproterozoic basement gneisses of the Laurentian craton (e.g. Sibson 1977; Walker 1990; Fettes et al. 1992; Butler et al. 1995; Imber 1998; MacInnes et al. 2000; Imber et al. 2001, 2002; Osinski et al. 2001; Szulc et al. 2008). Despite the extent of the fault zone, it has attracted relatively little attention in regional tectonic syntheses as it does not appear to separate fundamentally different terranes. The complex kinematic evolution of the fault zone has been well documented (e.g. Sibson 1977; Walker 1990; Fettes et al. 1992; Butler et al. 1995; Imber 1998; Imber et al. 2001, 2002) but there is a lack of geochronological data on the timings of successive episodes of deformation. This is partly because deformation mainly occurred at temperatures of less than 500°C, and neomineralization in fault-related rocks did not produce any mineral phases that could be reliably dated using either Rb–Sr or 40Ar–39Ar systems (Sibson 1977; Walker 1990; Fettes et al. 1992; Butler et al. 1995; Imber 1998; Imber et al. 2001, 2002; Szulc et al. 2008). However, there is widespread evidence for apatite recrystallization within the mylonites formed during early top-to-the-NW ductile thrusting (Sibson 1977; Butler et al. 1995; Imber et al. 2001, 2002). Here we report the results of U–Pb apatite dating which places constraints on the timing of ductile thrusting and discuss the regional tectonic implications.

An early Mesozoic continental reconstruction of the North Atlantic region reveals an intersecting mosaic of orogens that developed over c. 1.5 Ga of Earth's history (Fig. 1). Together these record three supercontinent cycles, each separated by periods of rifting and ocean basin development. The youngest and most prominent orogens are the Variscan (Devonian–Carboniferous) and Caledonian (Ordovician–Devonian) which resulted from the collision of Laurentia, Baltica, and Gondwanan terranes and culminated in the formation of Pangaea (Cocks and Torsvik 2002; Torsvik and Cocks 2004). The Caledonian orogen overprints the c. 1.1–0.9 Ga Grenville–Sveconorwegian orogen of NE North America and southern Sweden which resulted from the collision of Laurentia, Baltica, and Amazonia, a culminating event in the assembly of Rodinia (Li et al. 2008; Cawood and Pisarevsky 2017; Bingen et al. 2021, see however Slagstad et al. 2017). Relicts of basement complexes that were reworked at c. 1.2–1.0 Ga are found within the Caledonides in Scotland (Walker et al. 2020; Bird et al. 2023) and Norway (e.g. Wang et al. 2021). The Archean to early Mesoproterozoic basement complexes of Laurentia and Baltica record the amalgamation of crustal blocks at c. 1.9–1.7 Ga to form Nuna, which was followed by accretionary orogenesis at c. 1.7–1.4 Ga along the southern margin of the supercontinent (e.g. Karlstrom et al. 2001). The OHFZ is developed entirely within Mesoarchean to early Paleoproterozoic basement rocks which form the foreland of the Caledonian orogen in Scotland and have been correlated with the Laurentian basement of SE Greenland (Figs 1, 2a; Friend and Kinny 2001). The initiation of the OHFZ has been variously attributed to all three supercontinent cycles, reflecting the lack of reliable field and geochronological constraints.

The OHFZ is a NNE-trending belt of mylonites, cataclasites and phyllonites that is exposed over a strike length of ∼170 km along the eastern seaboard of the Outer Hebrides island chain (Fig. 2b; Fettes et al. 1992). Deep seismic profiling shows that the fault zone dips at ∼25° to the ESE to a depth of ∼25 km (Smythe et al. 1982; Brewer and Smythe 1984), and it has been suggested (e.g. Blundell et al. 1985; Stein 1988; Lailey et al. 1989) to intersect, and possibly offset, the Moho (Fig. 3). Normal faults in the immediate hanging wall of the OHFZ (Figs 2b, 3) are interpreted as short-cut structures that resulted from Permo-Triassic reactivation of the fault zone during extension and initiation of the Minch Basin (Stein 1988, 1992). The OHFZ is hosted by the Lewisian Gneiss Complex exposed on the Outer Hebrides and the mainland of NW Scotland (Fig. 2b). The Lewisian Gneiss Complex originated as a series of tonalite-trondhjemite-granodiorite (TTG) suites of c. 3.1–2.7 Ga age that were affected by a broadly similar tectonothermal history: (1) c. 2.7 Ga granulite facies metamorphism (Badcallian event), (2) c. 2.5 Ga amphibolite facies metamorphism (Inverian event), although this has yet to be recognized in the Outer Hebrides, (3) intrusion of mafic (Scourie) dykes c. at 2.4 Ga and c. 2.0 Ga, and (4) amphibolite facies metamorphism and deformation at c. 1.9–1.65 Ga (Laxfordian) (e.g. Corfu et al. 1994; Friend and Kinny 1995, 2001; Whitehouse et al. 1996; Kinny and Friend 1997; Zhu et al. 1997; Love et al. 2004; Mason and Brewer 2004; Davies and Heaman 2014; Crowley et al. 2015). In the Outer Hebrides (Fig. 2b), the Laxfordian event imposed the dominant NW–SE trending foliation and formed the Uig Hills Complex of late- to post-tectonic granite pegmatite sheets that were emplaced at c. 1.67 Ga (Fig. 2b; Fettes et al. 1992; Friend and Kinny 2001). The c. 1.89–1.87 Ga calc-alkaline South Harris Igneous Complex (Whitehouse and Bridgwater 2001; Mason et al. 2004) and the metasedimentary rocks of the Leverburgh belt to the south (Fig. 2b) were affected by UHT granulite facies metamorphism (Baba 2004; Hollis et al. 2006) and may be allochthonous, defining a folded suture between colliding Archean blocks (Park 2005; Mason 2012, 2016). 40Ar–39Ar hornblende data indicate that the whole of the northern Outer Hebrides (Lewis and Harris; Fig. 2b) west of the OHFZ cooled through a closure temperature of ∼500°C in the interval 1.7–1.6 Ga (Cliff et al. 1998). The mainland Lewisian Gneiss Complex has yielded K–Ar ages from hornblende, biotite, and muscovite in the range c. 1700–1400 Ma (Moorbath and Park 1972; Park et al. 1994).

The OHFZ was divided by Butler (1995) into the ‘Northern Segment’, exposed on the islands of Lewis and Harris, and the ‘Southern Segment’ of the Uists and Barra (Fig. 2b). Fault rocks, kinematic indicators, and overprinting relationships show that the OHFZ records up to five deformational events (Table 1; Butler et al. 1995; Imber et al. 2001, 2002), although these vary in their development and relative importance and in some locations are absent entirely. The earliest phase of deformation involved top-to-the-NW ductile thrusting, evidence for which is preserved within a belt of upper-greenschist facies mylonites exclusive to the Northern Segment. Reactivation of the fault zone affected both segments and involved episodes of top-to-the-NW brittle thrusting, sinistral strike-slip motion, and low-angle top-to-the-SE extensional displacements (Butler et al. 1995; Imber et al. 2001, 2002). The final phase of reactivation involved the development of high-angle brittle faults with a top-to-the-SE sense of throw, and which are attributed to the formation of the offshore Minch Basin (Butler et al. 1995; Imber et al. 2001).

The timing of early ductile thrusting, and thus initiation of the OHFZ, is poorly constrained by field relationships. An upper limit is provided by the manner in which the fault zone truncates regional foliation trends in its footwall and hanging wall, as well as cuts across the boundary of the Uig Hills Complex with its host gneisses (Fig. 2b). These map-scale relationships suggest that it must have formed after the main phases of Laxfordian deformation and metamorphism and intrusion of the Uig Hills Complex at c. 1.67 Ga. A lower limit of c. 0.3 Ga is provided by a 4 km thick sequence of Permo-Triassic conglomerates and sandstones, the Stornoway Formation, exposed in NE Lewis (Fig. 2b; Steel and Wilson 1975). The Stornoway Formation unconformably overlies Lewisian gneisses and intensely faulted rocks of the OHFZ and is bound on its western margin by steeply east-dipping normal faults (Steel and Wilson 1975; Butler et al. 1995). The sediments are undeformed and incorporate various fault rock clasts derived from the OHFZ. This implies that there was no significant displacement along the OHFZ following their deposition (Butler et al. 1995). These map-scale and field observations thus only restrict development of the OHFZ to between c. 1.67 and 0.3 Ga. Lailey et al. (1989) proposed that amphibolite sheets, which they correlated with the Scourie Dykes, cross-cut the mylonitic fabric in SE Scalpay. This would imply a latest Neoarchean or early Paleoproterozoic age of ductile thrusting. However, Imber et al. (2002) demonstrated that the amphibolite sheets featured in Lailey et al. (1989) and Laxfordian granite pegmatites are overprinted by the thrust-related mylonites. Fettes et al. (1992) suggested that early thrusting overlapped the emplacement of late Laxfordian granite pegmatite sheets. However, no precise details were provided of localities where this could be verified and hence the Laxfordian granite-pegmatite sheets have been subsequently viewed as entirely pre-tectonic with respect to the thrust-related mylonites (Butler 1995; Butler et al. 1995; Imber 1998; Imber et al. 2001, 2002).

Geochronological data with potential implications for the age of the OHFZ were provided by Rb–Sr biotite ages obtained from the footwall Lewisian gneisses across Harris and Lewis (Cliff and Rex 1989). The data indicate that the area north of the Langavat Shear Zone (LaSZ; Fig. 2b) was affected by a ‘general reheating’ event close to 1100 Ma, in contrast to the area to the south that preserves ages >1300 Ma. The distribution of biotite ages was taken by Cliff and Rex (1989) to imply kilometre-scale displacements across the Langavat Shear Zone at c. 1100 Ma. Imber et al. (2002) attributed the widespread isotopic resetting of biotite ages north of the Langavat Shear Zone to Grenvillian ductile thrusting along the OHFZ and associated burial of the Lewisian footwall block. Limited isotopic data are relevant to some younger events within the OHFZ. Kelley et al. (1994) used 40Ar–39Ar laser-probe techniques to indirectly date pseudotachylyte veins on Grimsay Island, North Uist at 430 ± 6 Ma. Whilst the kinematics of the dated sample remain unconstrained, it is generally accepted that the brittle thrusting event occurred during the Caledonian orogeny (e.g. Butler 1995; Butler et al. 1995; Imber 1998; MacInnes et al. 2000; Imber et al. 2001; 2002; Osinski et al. 2001; Szulc et al. 2008; Campbell et al. 2021). A series of K–Ar whole rock ages of c. 397 Ma obtained from phyllonite belts within the OHFZ suggest that they also formed during the Caledonian orogeny (D. C. Rex in Sibson 1977).

The results of early, top-to-the-NW ductile thrusting are represented by a north–south to NE–SW trending belt of mylonites exposed between NE Lewis and Scalpay (Fig. 2b) (Butler et al. 1995; Imber et al. 2001; Szulc et al. 2008). South of here, onshore exposures of mylonite are absent. Either early ductile thrusting did not affect the Southern Segment of the OHFZ or the mylonite belt lies offshore and is unexposed (e.g. Imber et al. 2002). Field analysis and sampling was undertaken at two localities within the footwall of the OHFZ, east and west of Tarbert (Fig. 4a), and three localities within the mylonite belt at Scalpay, Loch Sgiobacleit, and Aird Raerinish (Fig. 4b–d). Hand specimens were cut into polished thick-sections and mineralogically and texturally characterized using optical microscopy. The field (Fig. 5) and microscopic (Fig. 6) observations made from these three localities are described below.


The Island of Scalpay (Fig. 4b) is situated off the NE coast of south Harris and hosts the southernmost exposures of the mylonite belt. Here, the mylonite belt is believed to reach its maximum thickness of ∼600 m (Butler 1995). A gradational increase in ductile strain is evident from relatively unmodified Lewisian gneisses, resembling those near Tarbert (Fig. 5a), in the NW to pervasively foliated mylonites along the SE coast (Fig. 5b–e). Quartz ribboning along the foliation characterizes the shear fabric that strikes approximately NE–SW and dips ∼35° SE. A well-developed lineation of quartz rods and elongate quartz-feldspar aggregates lies within the foliation, plunging ∼35° ESE, indicating a dip-slip sense of shear (Butler et al. 1995; Imber et al. 2001, 2002; Szulc et al. 2008). Other kinematic indicators within the mylonites include σ-type feldspar porphyroclasts, asymmetrical shear bands, and asymmetrical, rotated amphibolite pods (Fig. 5e). All indicate a top-to-the-NW sense of shear (Butler et al. 1995; Imber et al. 2001, 2002; Szulc et al. 2008).

In thin section, the mylonites are characterized by interbanded quartz ribbons and flattened aggregates of partially sericitized albite (Fig. 6a). Quartz displays undulose extinction and has likely undergone dynamic recrystallization through sub-grain rotation, producing an entirely polycrystalline internal structure. Similarly, albite has sporadically developed new sub-grains, usually along grain boundaries or shear bands, indicating at least partial recrystallization. Fine-grained, syn-kinematic aggregates of sericite needles and epidote, coupled with retrogressive chlorite, sometimes develop c-type asymmetrical shear bands indicative of top-to-the-NW shear (Fig. 6a). Within mafic pods, semi-prismatic hornblende porphyroclasts are deformed in a brittle sense only, mainly through trans-granular fractures commonly infilled with quartz-actinolite fibres (Fig. 6b), which are usually oriented NW–SE (Imber 1998). Relict hornblende grains exhibit partial alteration to actinolite, chlorite, and biotite and are often mantled by top-to-the-NW actinolite-chlorite strain shadows (Fig. 6b).

Loch Sgiobacleit

Along the crags of the northwestern flank of Loch Sgiobacleit in central Lewis (Fig. 4c), the mylonite belt strikes roughly north–south, and comprises protomylonitic gneisses that dip ∼20–30° to the east. The protomylonites are cross-cut by top-to-the-NW brittle thrust faults and associated cataclastic gneiss and pseudotachylyte (Fig. 5g). Concordant pseudotachylyte veins also occur sporadically. These features are interpreted as the products of subsequent top-to-the-NW brittle thrusting and represent a greater magnitude of brittle strain here relative to Scalpay where only minor evidence for later brittle thrusting was detected. The protomylonitic foliation at Loch Sgiobacleit contains a lineation defined by quartz rods and quartz-feldspar aggregates that plunge 20–30° to the east. Kinematic indicators within the protomylonites predominantly consist of σ-type K-feldspar porphyroclasts (Fig. 5f) and asymmetrical shear bands, which are consistent with top-to-the-NW shear. Mafic bodies within the protomylonites usually exist as concordant sheets and do not appear to have been strongly deformed by top-to-the-NW ductile thrusting.

In thin section, the protomylonites are characterized by core-and-mantle quartz ribbons, which anastomose between abundant porphyroclasts of K-feldspar, oligoclase and hornblende (Fig. 6c, d). Feldspar grains consist of partially sericitised K-feldspar and oligoclase. Unlike the albite grains observed at Scalpay, oligoclase grains are often stubby and do not appear to have deformed plastically. However, there is evidence for localized sub-grain development of oligoclase along micro-shear zones and grain rims, hence oligoclase may still have recrystallized syn-kinematically. Rounded σ-type K-feldspar porphyroclasts are wrapped by quartz ribbons, biotitic laths, and top-to-the-NW actinolite strain shadows (Fig. 6c, d). Hornblende typically forms semi-prismatic porphyroclasts tailed by top-to-the-NW actinolite strain shadows and is commonly cross-cut by trans-granular veins infilled by NW–SE oriented quartz-actinolite aggregates (Imber 1998). Hornblende also exhibits partial alteration to actinolite and chlorite. A high density of sub-parallel micro-faults and fractures are oblique to and cross-cut the protomylonitic fabric and are associated with both normal and reverse displacements (Fig. 6c). Discontinuous bands of intensely brecciated material are also common, the interstices of which are infilled by very fine-grained quartz and feldspar.

Aird Raerinish

The Aird Raerinish peninsula (Fig. 4d) lies just east of Ranish along the NE coast of Lewis (Fig. 2b). Here, exposures of the mylonite belt strike north–south or NNE–SSW, occupying a thin strip along the eastern coast of the peninsula (≤200 m wide) (Fig. 4d). The mylonite belt at Aird Raerinish consists of fissile and steeply (up to 60°) east-dipping ultramylonites that have produced a spectacular cliff topography facing out into the Minch. Previous workers (e.g. Butler 1995) mapped discrete parallel belts of mylonites near Beinn Mhor, ∼300–400 m inland from the eastern coast of the peninsula. The present study instead suggests that the intensity of mylonitization gradually dissipates west of the coastal belt to the point where exposures are dominated by apparently unmodified Lewisian gneisses (Fig. 4d). Quartz ribboning and elongated feldspar aggregates characterize the ultramylonites at Aird Raerinish, and produce an intense, fissile shear fabric which is imprinted on felsic gneisses and dm- to m-scale Laxfordian pegmatites (Fig. 5h). An intense down-dip lineation of quartz rods and quartz-feldspar aggregates is developed along the foliation and indicates a dip-slip sense of shear. Additional kinematic indicators predominantly consist of σ-type K-feldspar porphyroclasts and boudinaged mafic pods, all of which imply a top-to-the-NW sense of shear. Cross-cutting the ultramylonites are numerous high-angle (∼60°), east-dipping mesoscopic faults and fractures. Displacement along the brittle faults is usually extensional, possibly related to Mesozoic basin development.

In thin section, the ultramylonites are characterized by mm-wide bands of ultrafine-grained epidote and discontinuous, polycrystalline quartz ribbons. Rounded σ- and δ-type K-feldspar and albite porphyroclasts are abundant and are often wrapped by recrystallized quartz ribbons, biotitic laths, and top-to-the-NW actinolite strain shadows (Fig. 6e, f). Feldspars are partially sericitised, whilst albite has locally developed a sub-grained internal structure indicative of at least partial recrystallization. High-angle micro-faults and fractures of both normal and reverse shear sense frequently cross-cut the ultramylonitic fabric (Fig. 6e) and are typically infilled by very fine-grained feldspar, acicular chlorite, and fabric-parallel quartz fibres. Across all samples, accessory apatite, titanite, zircon and allanite typically exist either as inclusions or interstitial grains. Euhedral magnetite sporadically overprints the mylonitic fabric and typically retains a pyritic core.

Metamorphic conditions

The mylonitic fabrics are defined by a recrystallized assemblage of quartz + albite/oligoclase + sericite + actinolite + epidote + apatite ± calcite. The interpretation of some authors (e.g. Fettes and Mendum 1987; Walker 1990; Imber 1998; Imber et al. 2002) that hornblende and biotite belong to the recrystallized, syn-kinematic assemblage has led to suggestions that metamorphic conditions during top-to-the-NW ductile thrusting reached amphibolite facies, exceeding temperatures of ∼500°C. The present study, however, interprets hornblende, biotite, and K-feldspar as relict, pre-kinematic phases from older (Laxfordian?) events (Table 2) as they do not show any evidence for recrystallization and have predominantly deformed in a brittle manner. Because of this, coupled with the dynamic recrystallization of quartz through sub-grain rotation (Stipp et al. 2002), the localized recrystallization of albite (Scalpay & Aird Raerinish) and oligoclase (Loch Sgiobacleit) (Passchier and Trouw 2005), and the syn-kinematic growth of sericite, epidote, and actinolite, it is suggested here that metamorphic conditions were within the broad temperature range of 400–500°C, equivalent to mid- to upper greenschist facies. This is consistent with the temperature of 500 ± 30°C proposed by Szulc et al. (2008) on the basis of fluid inclusion studies of sheared vein quartz within mylonites near Eilean Glas on Scalpay (NG 2470 9480). The sericitization of feldspar and the growth of epidote is more pervasive within the mylonite belt than the surrounding unmodified Lewisian gneisses and is particularly advanced at Aird Raerinish, suggesting that metamorphism during early ductile thrusting was probably accompanied by a significant fluid phase which varied locally along strike of the fault zone.

Eight samples, two from each principal locality, were selected for detailed microstructural, trace element and geochronological analysis and are described below (Table 3).

RS-20-01 & RS-20-03 (Tarbert)

RS-20-01 (NB 1259 0328) and RS-20-03 (NB 1696 0023) are high-grade, coarse-grained quartzo-feldspathic gneisses obtained from road-cutting exposures of the Lewisian Gneiss Complex near Tarbert (Fig. 4a). RS-20-01 was collected from Aird Asaig on the western coast of north Harris and was the most distal sample to the OHFZ. This sample demonstrates no evidence for top-to-the-NW ductile strain and has instead preserved a Laxfordian fabric and texture (Fig. 5a). RS-20-03 was collected eastward of RS-20-01, just NE of Tarbert, hence situated closer to the OHFZ. RS-20-03 also preserves a Laxfordian fabric and texture, although exhibits a subtle top-to-the-NW ductile overprint through rare, sigmoidal, top-to-the-NW sheared quartz ribbons.

RS-20-07 & RS-20-09 (Scalpay)

RS-20-07 (NG 2474 9479) and RS-20-09 (NG 2467 9493) are fine-grained, quartzo-feldspathic mylonites, sampled adjacent to the Eilean Glas lighthouse on the SE coast of Scalpay and were collected ∼150 m apart (Fig. 4b). Both samples comprised part of the NE–SW trending belt of thrust-related mylonites, within which increasing ductile strain can be demonstrated towards the SE of the island (Fig. 4b).

RS-20-27 & RS-20-29 (Loch Sgiobacleit)

RS-20-27 (NB 2998 1658) and RS-20-29 (NB 2992 1648) are fine-medium grained quartzo-feldspathic protomylonites, sampled from the crags on the northwestern flank of Loch Sgiobacleit in central Lewis (Fig. 4c). The two samples were collected ∼115 m apart within the northern continuation of the belt of thrust-related mylonites observed on Scalpay. At Loch Sgiobacleit they have been displaced by a series of north–south striking, east-dipping brittle thrust faults.

RS-20-39 & RS-20-40 (Aird Raerinish)

RS-20-39 and RS-20-40 (NB 4201 2435) are very fine-grained quartzo-feldspathic ultramylonites, obtained from the southeastern cliffs of the Aird Raerinish peninsula in NE Lewis (Fig. 4d). RS-20-39 & RS-20-40 were collected directly adjacent to one another from a ∼150–200 m wide coastal strip of ultramylonites, interpreted as the north to northeastern continuation of the mylonite belt observed on Scalpay and at Loch Sgiobacleit.

Energy-dispersive X-ray spectroscopy (EDS) and backscatter electron (BSE) microscopy

The eight samples described above were chosen for further compositional and textural characterization and for the identification of apatite grains using scanning electron microscopy (SEM) techniques. Analyses were performed by a Zeiss EVO MA 10 LaB6 SEM at the University of Portsmouth. Prior to imaging, samples were carbon-coated to minimize charging effects. An Oxford Instruments X-max 80 mm2 EDS detector was used to produce whole-section elemental maps from which apatite grains were located. EDS analyses were operated at a working distance of 14.5 mm. High-resolution BSE imaging was subsequently undertaken to investigate the micro-textures and phase relationships of apatite, using an accelerating voltage of 20 kV and a probe current of 400 or 500 pA.

Laser ablation inductively coupled plasma mass spectrometry (LA−ICP−MS)

In-situ U–Pb and trace element isotopic analyses were performed in unison, using an ASI RESOlution 193 nm ArF excimer laser coupled to an Analytik Jena PlasmaQuant Elite ICP-MS at the University of Portsmouth for six samples (RS-20-07, RS-20-09, RS-20-27, RS-20-29, RS-20-39, & RS-20-40). The remaining two samples (RS-20-01 & RS-20-03) were analysed using an Agilent 8900 ICP-MS Triple Quad attached to the same laser. The laser was operated at a fluence of 2.5 J cm−2, a frequency of 2 Hz and a spot size of 24 μm throughout all analytical runs. Helium was utilized as the carrier gas at a flow rate of c. 300 ml/min. In addition to measuring 202Hg, 204Pb, 206Pb, 207Pb, 208Pb, 232Th, 235U, and 238U for U–Th–Pb geochronology, 88Sr, 89Y, 139La, 140Ce, 146Nd, and 172Yb were also determined. The REEs were selected in order to monitor trace element mobility during deformation, whilst Sr enrichment and diffusion in metasomatic and metamorphic apatite is well documented (e.g. Kirkland et al. 2018; Odlum and Stockli 2020; Ribeiro et al. 2020a, b; Chew and Spikings 2021; Odlum et al. 2022).

For U–Th−Pb isotope ratio measurements, MAD apatite (weighted average 206Pb/238U age of 474.2 ± 0.4 Ma, Thomson et al. 2012) was used as the primary reference material to correct for mass bias, instrument drift, and laser-induced elemental fractionation. McClure Mountain (weighted mean 207Pb/235U age of 523.51 ± 1.47 Ma, Schoene and Bowring 2006) and Durango apatite (40Ar/39Ar age of 31.44 ± 0.18 Ma, McDowell et al. 2005) were used as secondary reference materials for six samples (RS-20-07, RS-20-09, RS-20-27, RS-20-29, RS-20-39, & RS-20-40). Due to the unavailability of McClure Mountain for the subsequent analyses of RS-20-01 and RS-20-03, Xuxa apatite (573 ± 0.8 Ma, Schuch 2018) was used instead for these samples. All discordia lower-intercept ages on Tera-Wasserburg plots obtained for the three secondary reference materials are within uncertainty of their respective reference ages. The Analytik-Jena PlasmaQuant Elite ICP-MS session yielded ages of 517 ± 11 Ma (2σ, n = 14, MSWD = 1.4) and 34.1 ± 4.3 Ma (2σ, n = 11, MSWD = 0.79) for McClure Mountain and Durango apatite, respectively. The Agilent 8900 ICP-MS session yielded ages of 568.3 ± 4.7 Ma (2σ, n = 17, MSWD = 0.72) and 33.2 ± 1.9 Ma (2σ, n = 15, MSWD = 0.86) for Xuxa and Durango apatite, respectively. U–Pb data reduction was performed using the ‘VisualAge_UcomPbine’ data reduction scheme in Iolite (Paton et al. 2011; Chew et al. 2014), applying a curve fit to correct for downhole U–Pb fractionation, and a 207Pb based common-Pb correction scheme to the primary reference material (MAD apatite). The 238U/206Pb and 207Pb/206Pb isotopic ratios for each analysis are presented uncorrected for common-Pb in Tera-Wasserburg concordia diagrams using Isoplot 4.15 (Ludwig 2008).

For trace element geochemistry, 43Ca was used as the internal elemental standard, assuming a stoichiometric Ca content in apatite of 40 wt%. NIST SRM 612 glass was used as the primary standard to correct for laser-induced elemental fractionation, mass bias, and instrument drift. Durango apatite was used as a secondary reference material across all analytical runs and produced trace element concentrations which agree within ∼2–5% uncertainty of those from published values (Marks et al. 2012; Yang et al. 2014; Chew et al. 2016). Standard errors typically ranged between 1–3%. Trace element data reduction was performed using the ‘X_Trace_Elements_IS’ data reduction scheme in Iolite (Paton et al. 2011). Analytical runs consisted of a block of four to eight unknowns interspersed between blocks of four to eight reference materials/standards, comprising one or two analyses of MAD, McClure or Xuxa, Durango, and NIST SRM 612.

Micro-textural analysis of apatite

Apatite grains can be divided into two distinctive types on the basis of their grain geometry and internal structure. Both vary in abundance across the eight samples imaged by electron microscopy, and their estimated relative proportions are listed in Table 3.

Type-1 apatite is typically characterized by euhedral-hexagonal or stubby grains measuring up to ∼500 μm in diameter (Fig. 7a–f). Internally, Type-1 grains are frequently fractured and predominantly lack evidence of plastic deformation or recrystallization. However, some grains do appear to exhibit domains with a high density of aligned micro-vesicles (Fig. 7c–e) and evidence of partial recrystallization, particularly along grain rims (Fig. 7e). Very fine-grained, interstitial ilmenite sometimes infills fractures or forms thin, partial rims (Fig. 7d). Type-1 grains are either hosted as inclusions within quartz or feldspar or as interstitial grains. Type-1 grains are most abundant in samples collected from the unmodified footwall Lewisian gneisses east of Tarbert (RS-20-01 & RS-20-03), comprising at least 80% of all apatite grains within these samples (Table 3). By contrast, Type-1 apatite abundance is significantly reduced within the samples collected within the mylonite belt (∼5–50%).

Type-2 apatite demonstrates an elongated or asymmetrical grain geometry and a shape-preferred orientation parallel to the top-to-the-NW shear fabric (Fig. 8a–f). These grains often either display a sigmoidal geometry or quartz/actinolite strain shadows, both consistent with a top-to-the-NW sense of shear. Finer-grained apatite (∼5–50 μm) often forms discontinuous trails of very fine stubby grains which are aligned and concordant with the shear fabric (Fig. 8e). In some cases, these trails form an emanating ‘tail’ to coarser grains (Fig. 8f) and such grains are particularly abundant within samples from Loch Sgiobacleit and Aird Raerinish. Coarser Type-2 apatite grains are characterized by a deformed internal structure. Internal deformation is discernible as networks of triple junctions (Fig. 8c, d), which in some cases have advanced to form ultra-fine-grained (∼5–50 μm) ‘neoblasts’ or ‘sub-grains’ indicative of dynamic recrystallization (Fig. 8a, b and f). All variations of Type-2 apatite are most abundant within the samples collected from the mylonite belt at Scalpay and Aird Raerinish, accounting for ∼75–95% of apatite within these samples but they are less abundant within mylonites at Loch Sgiobacleit (∼50%). Significantly fewer Type-2 grains are present within the Lewisian gneisses obtained from Tarbert, accounting for ∼10–20%.

Apatite trace element geochemistry

Apatite trace element concentration (REEs + Sr) is plotted normalized to chondrites (Thompson 1982) and is investigated both by sample (Fig. 9a–h) and grain type (Fig. 9i, j). The former shows that apatite grains from the footwall Lewisian gneisses near Tarbert (RS-20-01 & RS-20-03; Fig. 9a, b) and those from the thrust-related mylonites at Loch Sgiobacleit (RS-20-27 & RS-20-29; Fig. 9e, f) exhibit the broadest range in REE enrichment and generally low Sr. By contrast, the more intensely deformed mylonites from Scalpay (RS-20-07 & RS-20-09; Fig. 9c, d) and Aird Raerinish (RS-20-39 & RS-20-40; Fig. 9g, h) host apatite with generally lower REE enrichment and higher Sr, with RS-20-07 a possible exception to this. This trend correlates well with increasing Type-2 apatite abundance within the most intensely deformed samples as outlined above. Because of this, the results of apatite trace element geochemistry are presented in the context of the apatite grain types and are outlined in further detail below.


Ninety-five analyses of Type-1 apatite were performed spanning all eight samples. All REEs (89Y, 139La, 140Ce, 146Nd, and 172Yb) analysed demonstrate a broad range of enrichment in Type-1 apatite (Fig. 9i), most notably La which is enriched between 4–2000 times chondritic values within Type-1 grains from across all four localities. On average, REE enrichment ranges from ∼90–290 times chondritic values although the top end of this range has been elevated significantly by RS-20-01, which demonstrates REE enrichments of up to 2000 times chondrites (Fig. 9a). Sr concentrations span a narrower range of ∼20–60 times chondrites and an average of ∼27. This means that Type-1 analyses dominantly exhibit an enrichment profile with relatively high REE and lower Sr concentrations, however a minority of analyses display the inverse (lower REE and higher Sr) (Fig. 9i). The former are most influenced by analyses from RS-20-01, RS-20-07 and RS-20-29 whilst the latter are most prominent within RS-20-03, RS-20-09, RS-20-27 and RS-20-39.


One-hundred and fourteen analyses of Type-2 apatite were performed spanning all eight samples. Four analyses demonstrated exceptionally enriched values and have thus been omitted. Type-2 grains largely resemble Type-1 grains by displaying a highly enriched and an exceptionally broad concentration across all the trace elements analysed (Fig. 9j), in particular La which is concentrated between ∼2–3000 times chondritic values. As with Type-1 grains, this trend continues for all the REEs analysed. An average of all REE concentrations produces a range between ∼60–200 times chondritic values. Sr produces a relatively more limited concentration spectra of between ∼20–90 times chondrites and average of ∼40. This means that Type-2 analyses also form an enrichment profile with relatively high REE and lower Sr but this is less pronounced in comparison to Type-1, which is driven by significantly more analyses with lower REE and higher Sr (Fig. 9j). These latter analyses are most influenced by samples collected from the thrust-related mylonites (notably RS-20-09, RS-20-29 and RS-20-40) but are also a significant component of RS-20-03, collected from the footwall Lewisian gneisses.

A principal component analysis (PCA) (Fig. 10) of all analyses was undertaken in order to investigate systematic variations across the selected trace elements and U–Th–Pb isotopes and cross reference these with the texturally distinct apatite grain types described above. The PCA indicates that PC1 is associated with U, Nd, radiogenic- and common-Pb. Despite some overlap, there is a clear progression of increasing deformation where Type-1 analyses load positively on PC1 and Type-2 predominantly load negatively. Light REE (La, Ce) and Th load positively on PC2, Yb and Y load negatively. Predominantly undeformed Type-1 grains disperse widely on PC2, their deformed counterparts rather less so. Th has a strong positive influence on PC2 but cannot be conclusively linked to any data points in particular, although may be linked with Type-2 analyses within RS-20-40. Sr and U show only minor variation relative to the other analytes.

U–Pb apatite geochronology

Tarbert (RS-20-01 and RS-20-03)

Spot analyses performed on RS-20-01 (n = 37) and RS-20-03 (n = 40) were targeted on both Type-1 and Type-2 apatite. One analysis performed on a Type-1 grain in RS-20-01 failed to yield any isotopic compositional values and has thus been omitted. RS-20-01 produces a discordia lower-intercept age of 1119 ± 40 Ma (2σ, n = 32, MSWD = 1.8; Fig. 11a) and an unconstrained 207Pb/206Pb isotopic ratio of 1.06 ± 0.04. RS-20-01 may also yield a second age of 1288 ± 140 Ma comprising five analyses which appear to define a separate isochron, possibly related to incomplete thermal resetting (Fig. 11a). RS-20-03 demonstrates a discordia lower-intercept age of 1072 ± 27 Ma (2σ, n = 39, MSWD = 1.6; Fig. 11b) and an unconstrained 207Pb/206Pb isotopic ratio of 1.05 ± 0.03 with one analysis omitted. A total of six analyses were omitted across both samples and together they appear to define a separate isochron regression or ‘older population’ of apatite which yields an age of 1316 ± 110 Ma (2σ, n = 6, MSWD = 4.0) and an unconstrained 207Pb/206Pb isotopic ratio of 0.94 ± 0.14 (Fig. 11c). Whilst there is an overlap within uncertainty and therefore no statistically significant age difference, Type-1 and Type-2 analyses across both samples may also define separate populations. Type-1 produces an age of 1096 ± 29 Ma (2σ, n = 42, MSWD = 1.8) and an unconstrained 207Pb/206Pb isotopic ratio of 1.05 ± 0.04, whereas Type-2 produces an age of 1034 ± 35 Ma (2σ, n = 29, MSWD = 1.4) and an unconstrained 207Pb/206Pb isotopic ratio of 1.04 ± 0.02. Both ages and Pb isotope ratios remain within the 95% confidence interval of one another, however.

Scalpay (RS-20-07 and RS-20-09)

Spot analyses performed on RS-20-07 (n = 23) and RS-20-09 (n = 23) were targeted on both Type-1 and Type-2 apatite. One spot analysis on a Type-2 apatite in RS-20-07 yielded no U–Pb compositional values and has thus been omitted. RS-20-07 produces a discordia lower-intercept age of 1021 ± 34 Ma (2σ, n = 22, MSWD = 2.1; Fig. 11d) and an unconstrained 207Pb/206Pb isotopic ratio of 1.01 ± 0.01. RS-20-09 yields a discordia lower-intercept age of 952 ± 37 Ma (2σ, n = 23, MSWD = 2.1; Fig. 11e) and an unconstrained 207Pb/206Pb isotopic ratio of 1.05 ± 0.03. The data from both samples was then amalgamated to investigate any difference in age between texturally preserved (Type-1) and deformed (Type-2) grains (Fig. 11f). Type-1 analyses produce an age of 986 ± 43 Ma (2σ, n = 16, MSWD = 4.9) and an unconstrained 207Pb/206Pb isotopic ratio of 1.00 ± 0.03, whilst Type-2 analyses yield an age of 889 ± 95 Ma (2σ, n = 29, MSWD = 2.4) and an unconstrained 207Pb/206Pb isotopic ratio of 1.00 ± 0.01. Both ages and Pb isotope ratios thus remain within the 95% confidence interval of one another.

Loch Sgiobacleit (RS-20-27 and RS-20-29)

Spot analyses performed on RS-20-27 (n = 24) and RS-20-29 (n = 31) were targeted on both Type-1 and Type-2 apatite. Five analyses on a Type-2 apatite in RS-20-27 and one analysis on a Type-1 apatite in RS-20-29 produced widely discordant U–Pb ratios and are presented as greyed ellipses in Tera-Wasserburg space (Fig. 11g). RS-20-27 produces a discordia lower-intercept age of 1024 ± 23 Ma (2σ, n = 19, MSWD = 1.8; Fig. 11g) and an unconstrained 207Pb/206Pb isotopic ratio of 1.05 ± 0.01. RS-20-29 produces a discordia lower-intercept age of 1038 ± 28 Ma (2σ, n = 30, MSWD = 6.4; Fig. 11h) and an unconstrained 207Pb/206Pb isotopic ratio of 0.88 ± 0.01. RS-20-29 demonstrates evidence for two separate populations of apatite and this is further revealed in Tera-Wasserburg space where analyses from both samples are amalgamated (Fig. 11i). Type-1 analyses across both samples yield an age of 1081 ± 24 Ma (2σ, n = 23, MSWD = 5.0) and an unconstrained 207Pb/206Pb isotopic ratio of 1.06 ± 0.03. Type-2 grains produce an age of 1002 ± 190 Ma (2σ, n = 27, MSWD = 51) and unconstrained 207Pb/206Pb isotopic ratio of 0.89 ± 0.02. The large uncertainty and MSWD associated with this latter age appear to have been strongly influenced by a group of Type-2 analyses from RS-20-29, which retain high common-Pb/Pb (>0.75) and low U/Pb ratios (<1). These analyses also appear to have driven the significantly lower 207Pb/206Pb composition obtained for RS-20-29 (0.88 ± 0.01) relative to RS-20-27 (1.05 ± 0.01) and were therefore likely affected by a fluid phase relatively enriched in radiogenic-Pb during deformation.

Aird Raerinish (RS-20-39 and RS-20-40)

Spot analyses performed on RS-20-39 (n = 23) and RS-20-40 (n = 8) were targeted on both apatite grain types. RS-20-39 yields a discordia lower-intercept age of 1001 ± 48 Ma (2σ, n = 23, MSWD = 8.9; Fig. 11j) and an unconstrained 207Pb/206Pb isotopic ratio of 0.90 ± 0.03. RS-20-40 produces a discordia lower-intercept age of 845 ± 210 Ma (2σ, n = 8, MSWD = 0.86; Fig. 11k) and an unconstrained 207Pb/206Pb isotopic ratio of 0.89 ± 0.01. The reduced sample size was a result of the very fine-grained nature of apatite within this sample and explains the significantly larger uncertainty and the <1 MSWD. The age obtained for RS-20-40 is thus probably unreliable, although it does overlap within error with that of RS-20-39. As undertaken for previous localities, plotting both samples together would appear to reveal a difference in age and common-Pb/Pb ratios between Type-1 and Type-2 apatite (Fig. 11l). Type-1 analyses were performed on RS-20-39 only, although appear to form a steeper isochron line which produces an age of 1047 ± 76 Ma (2σ, n = 10, MSWD = 7.9) and an unconstrained 207Pb/206Pb isotopic ratio of 0.95 ± 0.1. RS-20-40 incorporates Type-2 analyses only which are roughly concordant with the Type-2 analyses in RS-20-39, producing an age of 965 ± 57 Ma (2σ, n = 21, MSWD = 5.4) and an unconstrained 207Pb/206Pb isotopic ratio of 0.89 ± 0.01.

Individual 207Pb/206Pb ages were compared against grain size, laser ablation spot position and REE + Sr enrichment (Table S1 in the supplementary information). However, none of these appear to reveal any meaningful correlation and therefore we do not consider them further.

Micro-textural and trace element characteristics of apatite

Backscatter electron images of apatite grains across the eight samples reveal two apatite grain types, distinguished by their shape and internal structure (Figs 7, 8). The predominantly undeformed Type-1 grains (Fig. 7) are most abundant within the unmodified footwall Lewisian gneisses near Tarbert (∼80–90%) but are considerably less common within samples collected along the mylonite belt (∼5–50%). By contrast, Type-2 apatite, which demonstrates evidence for plastic deformation and/or recrystallization, is scarce within the footwall Lewisian gneisses (∼10–20%) and substantially more abundant within the mylonites (∼50–95%). These observations suggest that the ratio of Type-1 to Type-2 apatite was controlled by the intensity of top-to-the-NW ductile thrusting along the OHFZ. Type-1 grains are therefore interpreted as pre-kinematic porphyroclasts, whereas Type-2 grains are likely to be the deformed and/or recrystallized products of older Type-1 grains and thus syn-kinematic with ductile thrusting. The deformed internal structure commonly observed within coarser grained Type-2 apatite (Fig. 8) is akin to the recrystallization microstructures commonly associated with sub-grain rotation (SGR) of quartz and feldspar (Passchier and Trouw 2005), although this requires confirmation by transmission electron microscopy (TEM) and electron backscatter diffraction (EBSD) techniques.

The Type-2 grains associated with trails of ultra-fine-grained apatite (Fig. 8e, f) are remarkably similar to ‘type-2’ apatite described by Ribeiro et al. (2020a) in a study of the deformational behaviour of apatite in petrologically similar upper-greenschist facies mylonites within the Taxaquara Shear Zone, SE Brazil. Ribeiro et al. (2020a) utilized electron backscatter diffraction (EBSD) imaging, coupled with U–Pb and trace element isotopic analysis across the separate domains and discovered that tailed grains displayed a high crystallographic misorientation and a depletion of REEs and common-Pb enrichment relative to their parent grains. These authors concluded that the fine-grained apatite was the product of fluid-assisted dissolution–precipitation recrystallization (termed ‘dynamic precipitation mechanism’) and this mechanism of apatite recrystallization and its influence on Pb and REE diffusion has been further reported elsewhere (e.g. Kirkland et al. 2018; Odlum and Stockli 2020; Odlum et al. 2022). The numerous textural similarities with the tailed Type-2 grains described in the present study suggest that the latter may also have formed by this process.

Trace element analysis of apatite reveals that all REEs analysed are most enriched within Type-1 grains, whilst the plastically deformed and/or recrystallized Type-2 grains are marginally depleted in REEs relative to Type-1 but demonstrate slightly elevated Sr concentrations (Fig. 9). This is likely a factor of both grain types producing two separate enrichment profiles in variable proportions: those with relatively high REE enrichment and lower Sr, and those with lower REE and higher Sr. The low REE/high Sr analyses are highlighted in Figure 9 and were determined based on Sr > Ce or Nd (chondrite-normalized). Whilst not mutually exclusive, these analyses are notably more prominent within Type-2 apatite suggesting a broad correlation between textural and trace element characteristics of apatite, where deformed and recrystallized grains produce overall lower REE and higher Sr enrichment values. Figure 9i and j also provides a comparison of our dataset with apatite REE + Sr enrichment from different rock types and/or metamorphic grade (O'Sullivan et al. 2020; see references within), in addition to a direct comparison with the comparative study of Ribeiro et al. (2020b). High REE/low Sr analyses from both grain types fit best within the greenschist facies mylonite of Ribeiro et al. (2020b) and somewhat within the felsic granitoids and high metamorphic grade data compiled by O'Sullivan et al. (2020) but are variably depleted in Y and Yb. Low REE/high Sr analyses conform much better with the low/medium metamorphic grade data of O'Sullivan et al. (2020). Overall, the chondrite-normalized REE + Sr diagrams show a broad chemical disequilibrium across both grain types, producing two separate enrichment profiles of low REE/high Sr and high REE/low Sr. The latter are more abundant within Type-2 grains and consistent with apatite affected by medium grade metamorphism, which correlates well with the clear textural characteristics of these grains.

The PCA (Fig. 10) corroborates this trend by demonstrating the widest spread on PC2 (towards high La, Ce and high Yb, Y) in Type-1 analyses. By contrast, both radiogenic- and common-Pb demonstrate a strong association with PC1 and are linked with a group of Type-2 apatite analyses. This follows a pattern of increasing deformation where Type-1 analyses load positively on PC1 and Type-2 analyses load negatively. This indicates that OHFZ deformation has induced Pb isotope diffusion within apatite. In their comparable study, Ribeiro et al. (2020a) presented trace element data demonstrating a similar enrichment in REEs in ‘type-1’ apatite, which retain their igneous internal texture, relative to the plastically deformed and recrystallized ‘type-2’ apatite previously described. Ribeiro et al. (2020a) thus suggested that the fluid-assisted dissolution-precipitation thought to have deformed ‘type-2’ apatite also likely resulted in REE fractionation. A similar conclusion could therefore be inferred for this study, where fluid-assisted deformation has altered REE content within Type-2 grains, whilst Type-1 grains have remained relatively unaffected.

U–Pb isotopic data: direct dating of apatite recrystallization or cooling?

U–Pb apatite geochronology performed across all eight samples produces a broad spread of ages which mostly range between c. 1100 and 900 Ma. The absence of any documented magmatism in the Outer Hebrides during Stenian–Tonian times almost certainly rules out a magmatic crystallization age for these apatites and therefore the ages obtained must mark a period of resetting of the U–Pb apatite system. However, what is the temporal relationship between the apatite ages and the timing of ductile thrusting? For example, could the resetting have occurred significantly earlier than (Caledonian?) ductile thrusting? However, this can be ruled out given that, in agreement with previous studies (e.g. Walker 1990; Fettes et al. 1992; Imber 1998; Imber et al. 2001, 2002; Szulc et al. 2008), we have demonstrated that temperatures during early ductile thrusting reached 400–500°C which is within the closure temperature range of the U–Pb apatite system (350–575°C; Chew and Spikings 2021). It is highly implausible that apatite within the mylonite belt could have undergone widespread recrystallization at such temperatures to produce Type-2 grains and yet still retain older ages. Conversely, could the apatite ages represent a thermal event that was significantly younger than (Laxfordian?) ductile thrusting? This possibility can also be excluded since any later thermal event sufficient to reset apatite would be expected to have produced an overprinting metamorphic mineral assemblage. Since this is not observed, we interpret the c. 1100–900 Ma age range to broadly represent the timing of early ductile thrusting along the OHFZ.

The next issue to address is whether, in more detail, the U–Pb ages correspond to syn-kinematic apatite recrystallization at 400–500°C or cooling past the apatite Tc threshold. Apatite closure temperatures range between 350 and 575°C (Chew and Spikings 2021). Cherniak et al. (1991) showed that for apatite Tc of ∼350°C, grain size must be extremely fine-grained (≤10 μm) and the cooling rate must be ∼0.1°C/Myr. In the present study, spot analyses were targeted on grains with a diameter of ∼40–600 μm, which using the dataset of Cherniak et al. (1991), yields a Tc between ∼400–600°C. This suggests significant overlap between the inferred deformation temperatures and apatite closure temperatures within the analysed samples, and hence thermally driven Pb diffusion is inferred to have been important in the resetting of the U–Pb apatite ages. However, we note that the dataset does not produce a set of completely overlapping ages and the samples host two distinguishable apatite ‘types’. Type-1 grains are dominant within the gneissic protoliths, their relatively coarse grain-size and euhedral to subhedral form consistent with inheritance from igneous crystallization and/or high-grade metamorphism (Fig. 7). In contrast, Type-2 grains record plastic deformation or recrystallization (Fig. 8). Type-1 apatite produces slightly older ages than that of Type-2 apatite although the respective ages produced vary between localities (Fig. 12). Loch Sgiobacleit and Aird Raerinish demonstrate Type-1 ages of 1081 ± 24 Ma and 1047 ± 76 Ma, respectively, and are thus slightly younger, although within uncertainty, of the Type-1 age obtained from samples at Tarbert (1096 ± 29 Ma). By contrast, samples from Scalpay produce a Type-1 age of 986 ± 43 Ma, thus plotting outside of this uncertainty. Type-2 apatite produces ages which range from between 100 and 60 million years younger than the Type-1 ages at their respective localities and moreover demonstrate a similar age distribution between the four localities to that observed across the Type-1 ages (Fig. 12). The oldest Type-2 age obtained was from Tarbert, at 1034 ± 35 Ma, whilst Loch Sgiobacleit and Aird Raerinish produced ages of 1002 ± 190 and 965 ± 57 Ma, respectively. Samples from Scalpay produced a younger Type-2 age of 889 ± 95 Ma. A comparison of Pb isotope ratios between Type-1 and Type-2 apatite also reveals a contrast between both grain types, notably within samples collected at Loch Sgiobacleit and Aird Raerinish. The near identical and significantly lower Pb isotope ratios of Type-2 (0.89 ± 0.01–0.02) compared to Type-1 grains (1.06 ± 0.03 (Loch Sgiobacleit); 0.95 ± 0.1 (Aird Raerinish)) within these samples supports a different age and origin between the two grain types as alluded to by their different micro-textural characteristics, REE + Sr concentrations and U–Pb ages. Within samples collected from Tarbert and Scalpay, Type-1 and Type-2 grains yield very similar and overlapping common-Pb/Pb isotope ratios of c. 1.05 (Tarbert) and 1.00 (Scalpay). This may be best explained by localized variations in the intensity of fluid-assisted deformation during ductile thrusting. Metamorphism and apatite recrystallization at Loch Sgibobacleit and Aird Raerinish may have been accompanied by a fluid phase relatively enriched in radiogenic-Pb which, by contrast, was either absent or had only a minor effect on the rocks further south at Tarbert and Scalpay. This is supported by the more advanced growth of epidote and sericite and higher abundance of fluid-related Type-2 apatite observed within samples from Loch Sgiobacleit and Aird Raerinish compared to those from Tarbert and Scalpay.

All of the U–Pb ages and three out of four of the common-Pb/Pb ratios obtained for Type-1 apatite are within the 95% confidence interval of those of their respective Type-2 grains. Therefore, a possibility remains that the different grain types do not correspond to different age populations and origins and instead record a single tectono-thermal event. However, whilst higher precision techniques (i.e. ID-TIMS) would be required to conclusively resolve any multi-stage history, we consider it more likely that such differences in age and Pb isotope ratios are geologically meaningful since they are detected across up to four separate localities and can be directly correlated with clear textural heterogeneities. Type-1 grains are interpreted as relicts of the gneissic protolith, preserved as porphyroclasts which underwent thermally driven diffusion during initial overthrusting at c. 1100–1000 Ma. By contrast, Type-2 grains represent the result of progressive and continued recrystallization during protracted ductile shearing. Hence, these preserve younger U–Pb ages of c. 1030–900 Ma and generally lower common-Pb/Pb isotopic ratios than the Type-1 grains. The oldest age population (RS-20-01 & RS-20-03; Fig. 11c), which were sampled from the footwall Lewisian gneisses and yielded an age of 1316 ± 110 Ma, are interpreted as recording incomplete thermal resetting, possibly linked to post-Laxfordian regional cooling.

Timing of ductile thrusting along the OHFZ: implications for regional tectonic models

Our conclusion that ductile thrusting occurred between 1100 and 900 Ma places initiation of the OHFZ firmly within the time window of the tripartite collision of Laurentia, Baltica, and Amazonia, which culminated in the Grenville–Sveconorwegian orogeny and assembly of Rodinia. Hitherto the only evidence for deformation or metamorphism of this age on the Laurentian foreland of the Scottish Caledonides has been limited to: (1) the resetting of Rb–Sr biotite ages at c. 1.1 Ga in the footwall of the OHFZ on Lewis and Harris (Cliff and Rex 1989); and (2) 40Ar–39Ar ages of between 1024 ± 30 Ma and 980 ± 39 Ma obtained from pseudotachylytes within Laxfordian shear zones near Gairloch on the Scottish mainland (Sherlock et al. 2008). In contrast, east of the Moine Thrust, within the Caledonian orogen (Fig. 2a), inliers of Neoarchean basement were reworked at high metamorphic grade during the latest Mesoproterozoic as indicated by: (1) Lu–Hf garnet ages of c. 1200 Ma obtained from eclogites within the Eastern Glenelg inlier (Bird et al. 2023), (2) isotopic disturbance of U–Pb systems at c. 1008 Ma within the Swordly inlier (Strachan et al. 2020), and (3) a Lu–Hf garnet age of c. 1050 Ma within the Cullivoe inlier in Shetland, ∼170 km NNE of mainland Scotland (Walker et al. 2020). However, whether these basement inliers acquired their latest Mesoproterozoic overprint in their present positions relative to the Laurentian foreland in Scotland is uncertain. The present study is therefore significant as it provides unambiguous evidence that Grenville–Sveconorwegian events affected this sector of Laurentia.

Most recent reconstructions of the Grenville–Sveconorwegian orogen view it as the result of the collision of Amazonia with Laurentia and Baltica (Fig. 13; Li et al. 2008; Cawood and Pisarevsky 2017; Bingen et al. 2021, see however Slagstad et al. 2017). An extrapolated orogenic front between the Grenvillian and Sveconorwegian sectors of the orogen might reasonably pass north of the Annagh Gneiss Complex in NW Ireland which has long been known to have been affected by high-grade Grenvillian metamorphism (Daly 1996). However, evidence for c. 1.1–1.0 Ga events further north in northern Scotland and Shetland is less easy to attribute directly to the collision of Amazonia. Palaeomagnetic data imply an alternative interpretation (Cawood and Pisarevsky 2006; Cawood et al. 2010). Although Laurentia and Baltica are widely thought to have been adjacent until c. 1270 Ma, palaeomagnetic evidence indicates that Baltica then rifted away and rotated ∼90° clockwise. Baltica is thought to have returned to a position close to Laurentia by c. 990 Ma during the collision of Amazonia. The evidence for initiation of the OHFZ as well as high-grade metamorphism further east within the Scottish Caledonides at c. 1.1–1.0 Ga suggests that Laurentia and Baltica must have not only been in close proximity at that time but must have collided to form a short northern arm of the main Grenville–Sveconorwegian system (Fig. 13; Li et al. 2008; Strachan et al. 2020).

Insights into the sequence of collisional events within the three-plate model outlined above are potentially provided by the late Mesoproterozoic to early Neoproterozoic ‘Torridonian’ continental sedimentary successions in NW Scotland (Fig. 2a). The oldest is the Stoer Group, a 2.5 km thick succession of alluvial red siliciclastics which rest unconformably on the Lewisian Gneiss Complex and were thought to have been deposited in a rift basin (Stewart 1982, 2002). A depositional age of c. 1.2 Ga has been proposed on the basis of 40Ar–39Ar dating of diagenetic K-feldspar grains and Pb–Pb dating of calcite within a limestone which yielded ages of 1177 ± 5 Ma (Parnell et al. 2011) and 1199 ± 70 Ma (Turnbull et al. 1996), respectively. However, a younger age is implied by the U–Pb ages of c. 1100 ± 40 Ma obtained from two detrital apatite grains (Kenny et al. 2019). If the latter constraint is more correct as seems likely, deposition probably overlapped c. 1.1–1.0 Ga tectonic activity and ductile thrusting along the OHFZ. The Stoer Group might therefore be better interpreted as the sedimentary fill of a ‘thrust top’ basin. No sedimentological data require a rift basin interpretation: palaeocurrent data indicate that sediment was transported mainly westward, with variations a consequence of local palaeotopographic highs (Simms 2015; Ielpi et al. 2016). The Stoer Group was tilted before deposition of the unconformably overlying Sleat-Torridon succession (Fig. 2a) which is widely interpreted to have occurred in a foreland basin developed along the margin of the Grenville orogen, with some detritus derived from NE Canada (Rainbird et al. 2001; Krabbendam et al. 2008, 2017). Hence the ‘Torridonian’ successions of NW Scotland might represent the results of deposition in two foreland basins, the Stoer Group relating to a Laurentia–Baltica collision that occurred slightly earlier than the culminating collision of Amazonia and formation of the foreland basin in which the Sleat–Torridon groups accumulated.

  1. Thrust-related mylonites within the OHFZ comprise a syn-kinematic mineral assemblage of quartz + albite/oligoclase + sericite + actinolite + epidote + apatite ± calcite. Metamorphic conditions during ductile thrusting were within the temperature range of 400–500°C, equivalent to mid- to upper-greenschist facies.

  2. Two texturally different apatite types occur within the mylonites and the footwall Lewisian gneisses. Type-1 apatite shows little evidence for recrystallization or plastic deformation and is most abundant within the footwall Lewisian gneisses. Type-2 apatite is preferred within the thrust-related mylonites and demonstrates textures indicative of plastic deformation, sub-grain rotation and/or fluid-assisted recrystallization, overall REE depletion/Sr enrichment relative to Type-1 grains and increasing association with radiogenic- and common-Pb.

  3. Type-1 grains produce variably older U–Pb ages and generally higher common-Pb/Pb isotopic ratios than that of Type-2 grains from their respective localities, although remain within uncertainty. Type-1 grains are interpreted as relicts of the gneissic protolith, preserved as porphyroclasts which record initial overthrusting between c. 1100 and 1000 Ma. These were variably preserved as ductile strain intensified and became progressively partitioned at all scales. Coupled with the influx of fluids, this intensification is suggested to have led to the development of the mylonite belt and syn-kinematic recrystallization of Type-2 apatite between c. 1030 Ma and an age possibly as young as 900 Ma.

  4. Ductile thrusting likely resulted from the collision of Laurentia and Baltica shortly prior to their collision with Amazonia. This explains the OHFZ and the sub-parallel belt of high-grade c. 1.1–1.0 Ga metamorphism preserved within basement inliers within the Scottish Caledonides. The initiation of the OHFZ is thus firmly tied to the Rodinia supercontinent cycle.

The authors thank Geoff Long for sample preparation, Bruno Ribeiro for discussions, Jack Mulder and other (anonymous) reviewers for detailed comments which improved the manuscript, and Kathryn Cutts for editorial handling.

JHM: conceptualization (equal), formal analysis (lead), investigation (lead), methodology (equal), visualization (lead), writing – original draft (lead), writing – review & editing (equal); RAS: conceptualization (equal), investigation (supporting), methodology (equal), resources (lead), supervision (lead), visualization (supporting), writing – original draft (supporting), writing – review & editing (equal); JRD: conceptualization (supporting), formal analysis (supporting), investigation (supporting), methodology (equal), supervision (supporting), writing – original draft (supporting), writing – review & editing (equal); MF: formal analysis (supporting), investigation (supporting), methodology (supporting), visualization (supporting), writing – original draft (supporting), writing – review & editing (supporting); GC: formal analysis (equal), investigation (equal); JD: conceptualization (supporting), formal analysis (supporting), investigation (supporting), methodology (equal), supervision (supporting), writing – original draft (supporting), writing – review & editing (equal)

Funding of fieldwork and sample collection from the University of Portsmouth is gratefully acknowledged.

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.

All data generated or analysed during this study are included in this published article (and if present, its supplementary information files).

This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License (http://creativecommons.org/licenses/by/4.0/)