A detective duo of apatite and zircon geochronology for East Avalonia, Johnston Complex, Wales

The assembly of Gondwana from 800 to 550 Ma (Meert and Van Der Voo 1997) coincides with a crucial time in the evolution of Earth systems. Neoproterozoic lithologies provide evidence for the emergence of complex multicellular life (Shen et al. 2008), global glaciation (Shields-Zhou et al. 2016) and an important period of continental crust production during the Pan-African Orogeny (Rino et al. 2008). Palaeotectonic reconstructions are key to our understanding of these interrelated processes and constrain the timing, duration and style of magmatism. The construction of Gondwana was a protracted and complex process whereby microcontinents, such as Avalonia, were amalgamated onto, and incorporated within, the emerging land mass (Cawood et al. 2021). The Neoproterozoic of southern Britain represents a prolonged (c. 200 Ma) period of discontinuous magmatism, which necessitates accurate age-constrained isotopic and geochemical signatures to reconstruct the tectonic setting of peri-Gondwanan terranes in this area (Pharaoh and Carney 2000; Compston et al. 2002; Schofield et al. 2016).

Drill core and geophysical data indicate that England and Wales (south of the Iapetus suture) are underlain by a mosaic of Avalonian terranes, which, from west to east, include the Cymru, Wrekin, Charnwood and Fenland terranes (Pharaoh and Carney 2000; McIlroy and Horak 2006; Schofield et al. 2016). However, complications arise from the sparsely distributed, fault-uplifted tectonic inliers (Fig. 1) and the varying degrees of deformation and metamorphism sustained by the Precambrian basement lithologies during the Caledonian, Variscan and Alpine orogenies (Pharaoh and Carney 2000). The Johnston Complex, the Arfon and Sarn groups of North Wales and the Pebidian Supergroup of north Pembrokeshire have been assigned to the Cymru Terrane (Pharaoh and Carney 2000; Schofield et al. 2016) (Fig. 1). However, other workers have questioned the reliability of such broad terrane associations based on limited basement outcrops and the internal complexities of individual inliers (Compston et al. 2002; Schofield et al. 2016). The Johnston Complex typifies this issue; based on published zircon U–Pb ages and its calc-alkaline geochemistry, it has been classified into either the Wrekin Terrane (McIlroy and Horak 2006) or the Cymru Terrane (Fig. 1) (Pharaoh and Carney 2000; Compston et al. 2002; Schofield et al. 2016). East Avalonian terranes are interpreted to represent remnants of a single late Neoproterozoic subduction margin (Pharaoh et al. 1987). As such, constraining the tectonomagmatic processes during this formative period in the evolution of Earth systems relies on accurate terrane designations, which, in turn, underpin palaeocontinental reconstructions, essential to constraining the relationships between different planetary geochemical reservoirs during important intervals in Earth history (Fig. 1) (Strachan et al. 2007; Nance et al. 2008).

As there is a general paucity of recent isotopic studies, previous terrane classification work has relied on sparse isotopic and geochemical data (Fig. 1) (Pharaoh and Carney 2000; Schofield et al. 2016). The Johnston Complex of SW Wales is a case in point, with research into the Johnston Complex being limited to Cantrill et al. (1916), Thorpe (1972) and the isotopic analyses of Patchett and Jocelyn (1979) and Schofield et al. (2016). Despite representing some of the oldest rocks in southern Britain, with a reported zircon U–Pb age of 643 +5/−28 Ma (Patchett and Jocelyn 1979), few other isotopic data are available for the Johnston Complex, aside from a whole-rock Nd analysis undertaken by Schofield et al. (2016). Hence the context of the Johnston Complex within the broader peri-Gondwanan magmatic realm is yet to be fully determined. Several details remain unresolved, including the degree of heterogeneity within the Johnston Complex (single pluton v. composite magmatic body), the character of its source rocks, the relationship of the complex to the surrounding inliers and its ultimate terrane classification.

This paper presents in situ zircon (U–Pb, Lu–Hf), apatite (U–Pb) and trace element data from the Johnston Complex for both minerals. Our results demonstrate that the complex is a composite magmatic body composed of discrete quartz diorite plutons along its c. 20 km length. Our data show that calc-alkaline magmatism persisted in an ensialic setting in East Avalonia to c. 570 Ma, prior to oblique convergence and terrane dispersal. We demonstrate that the source melt of the Johnston Complex was derived from a uniform and broadly chondritic source from 615 to 570 Ma, consistent with evolved $ε$Nd values from East Avalonia. The chemical data presented here reinforce the classification of the Johnston Complex within the Cymru Terrane of East Avalonia and show an S-type component within the melt. We demonstrate that a combined apatite and zircon geochronology approach constrains inherited v. primary zircon within S-type granitoids and clarifies the post-magmatic history of an igneous body.

The Menai Strait Fault separates the Precambrian geology of southern Britain into the Avalonian and Monian tectonic blocks (Fig. 1) (Strachan et al. 2007; Woodcock et al. 2012; Schofield et al. 2016). Several terranes within these blocks (e.g. the Cymru, Charnwood, Fenland and Wrekin terranes) comprise the Avalon Composite Terrane, which is primarily concealed by Phanerozoic rocks (Pharaoh and Carney 2000; Schofield et al. 2016). Major fault zones, such as the Malvern fault system and the Welsh Borderland fault system, represent the broad structural controls on the distribution of Neoproterozoic inliers in southern Britain (Fig. 1) (Pharaoh and Carney 2000; Woodcock et al. 2012). The ultimate relationship between the Avalonian and Monian terranes remains unresolved. However, both share a history of calc-alkaline magmatism, accretionary prism and basement gneiss amalgamation during the late Neoproterozoic (Woodcock et al. 2012; Schofield et al. 2016). Isotopic and structural data point towards a shared genesis for both units (Horak et al. 1996) and, taken together, they are termed the Avalon Superterrrane.

The Precambrian rocks of SW Wales crop out in two areas north and south of St Brides Bay (Fig. 1). To the south is the Johnston Complex, an east–west-trending intrusive suite of felsic plutons consisting of diorites, tonalites and granodiorites (Thorpe 1972). These rocks are cross-cut by metre-scale dolerite dykes emplaced during the Variscan Orogeny (Thorpe 1972, 1979; Patchett and Jocelyn 1979). The Johnston Complex is exposed in two outcrops: a coastal section on the southern end of St Brides Bay and an inland area within the Bolton Hill Quarry near the eponymous town of Johnston (Fig. 1). The Johnston Complex is thrust on top of Paleozoic sedimentary rocks and is entrained within the hanging wall of the Johnston–Benton thrust zone (Woodcock et al. 2012). The coastal outcrop of the Johnston Complex consists of quartz diorite, which is dominated by plagioclase (oligoclase to andesine), quartz, partially altered primary mafic minerals (hornblende and biotite) now occurring as chlorite, sericite and illite, and minor iron–titanium oxides (Fig. 2).

There is some lithological variability between the inland portion and coastal outcrops. The inland lithologies have an increased abundance of hornblende and biotite, but alkali feldspar is mainly absent (Thorpe 1972, 1979). By contrast, the coastal outcrops have a more homogenous quartz diorite composition and contain orthoclase at <2% modal abundance (Thorpe 1972).

A sample of quartz dolerite from the Johnston Complex was collected from south of St Brides Bay (51° 46′ 21.6′ N, 5° 08′ 57.1′ W). Zircon and apatite grains were separated following standard procedures at the John De Laeter Centre (JdLC), Curtin University, Perth, Australia. The sample was crushed using a ring mill and sieved to yield the ≤425 $μ$m fraction. This fraction underwent heavy mineral separation via a Jasper Canyon Research water-shaking table (Dumitru 2016). Lithium heteropolytungstate heavy liquid separation at 2.85 g cm⁻³ and Frantz isodynamic separation followed to produce a final fraction that was mounted in entirety onto a 25 mm epoxy resin round.

A polished 4 cm × 2 cm block was also prepared and analysed via a Tescan Integrated Mineral Analyser (TIMA) to provide insights into texture and mineralogy. Both the round mount and block were polished using a 1 $μ$m diamond suspension fluid and imaged using reflected and transmitted light. TIMA analysis was used to identify mineral phases on the mount and block and a Clara field-emission scanning electron microscope was then used to provide high-resolution images of the internal grain morphology, aiding spot placement during laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS).

A single session of split-stream zircon U–Pb and Lu–Hf analysis was undertaken at the GeoHistory Facility at the JdLC. A RESOlution LE193 nm ArF laser and a Lauren Technic S155 cell were used to ablate the zircon and an Agilent 8900 inductively coupled plasma mass spectrometer was used to collect U–Pb data. The Lu–Hf isotopes were collected simultaneously via a Nu Plasma II multi-collector inductively coupled plasma mass spectrometer. The relatively small grain size of zircon (typically <70 $μ$m in length) and internal complexities necessitated a comparatively small spot size of c. 38 $μ$m diameter.

The matrix-matched primary reference material used for zircon U–Pb data reduction was GJ1 (Jackson et al. 2004). The Plešovice (337.13 ± 0.37 Ma) (Sláma et al. 2008) and 91500 (1063.78 ± 0.65 Ma) (Wiedenbeck et al. 1995) secondary reference materials were used to monitor the accuracy of the ²³⁸U/²⁰⁶Pb ratio measurement. To check the accuracy of the ²⁰⁷Pb/²⁰⁶Pb ages, OG1 (3465.4 ± 0.6 Ma) (Stern et al. 2009) was reduced against Maniitsoq (3008.70 ± 0.72 Ma) (Marsh et al. 2019). All the weighted mean ages for the secondary reference materials were within 2$σ$ uncertainty of the accepted values (Appendix 1, Supplementary Material).

Mud Tank zircon (Woodhead et al. 2004) was the primary reference material for the zircon Lu–Hf analyses. Several secondary reference zircons, including GJ1 (0.282000 ± 0.000025) (Morel et al. 2008), 91500 (0.282306 ± 0.00004) (Woodhead et al. 2004; Woodhead and Hergt 2005) and R33 (0.282764 ± 0.000014) (Fisher et al. 2014a) were measured to confirm the accuracy. All the reference materials gave ¹⁷⁶Hf/¹⁷⁷Hf ratios within 2$σ$ uncertainty of the accepted values and the stable ¹⁷⁸Hf/¹⁷⁷Hf ratio was also within 2$σ$ uncertainty of the expected ratio (Appendix 1, Supplementary Material) (Sláma et al. 2008; Fisher et al. 2014b).

Two sessions of apatite U–Pb isotopic analyses were completed at the GeoHistory Facility, JdLC, using a RESOlution 193 nm ArF laser with a Lauren Technic S155 cell and an Agilent 8900 inductively coupled plasma mass spectrometer. Given the small grain size (<50 $μ$m) of apatite from the Johnston Complex, a 38 $μ$m spot size was ablated using a 5 Hz repetition rate and a fluence of c. 2 J cm⁻². Madagascar (MAD-2) (Thompson et al. 2016) and Mt McClure (MMC) (Schoene and Bowring 2006) apatite were used as the primary reference material for sessions 1 and 2, respectively. The secondary reference material analysed included the MMC apatite (523.51 ± 1.47 Ma), the Duluth Complex (FC) apatite (1099.1 ± 0.2 Ma) (Schmitz et al. 2003) and the Durango apatite (31.44 ± 0.18 Ma) (McDowell et al. 2005). The ages obtained via unanchored regressions through the secondary reference apatite were within 2$σ$ uncertainty of accepted ages (Appendix 2, Supplementary Material).

A single session of zircon and apatite trace element analysis was performed at the GeoHistory Facility at the JdLC. In situ trace element measurements were made on some grains not previously analysed during earlier U–Pb sessions using an Agilent 8900 inductively coupled plasma mass spectrometer. The size of the zircon and apatite grains necessitated a 38 $μ$m spot size at a fluence of 2.5 J cm⁻² and a 5 Hz laser pulse rate.

NIST 610 was used as the primary reference material for zircon analyses to calculate elemental abundance and correct for instrumental drift. The internal reference isotope used was ²⁹Si, assuming a stoichiometric zircon composition of 14.76 wt% silicon. Secondary reference materials included 91500 zircon, using the reported values of Sano et al. (2002), and NIST 612 (Jochum et al. 2005). For both reference materials, the measured abundances were within 5–10% of the reported values.

NIST 610 was the primary reference material for apatite trace element analyses. ²⁹Si was used as the internal reference isotope, assuming an apatite Si content of 0.41 wt% (Azadbakht et al. 2018). Secondary reference materials included the Durango apatite and NIST 621 glass; the measured abundances were generally within 10% of the reported values (Jochum et al. 2005; McDowell et al. 2005).

The results of the automated mineralogy undertaken via TIMA are consistent with the previous petrographic accounts of Cantrill et al. (1916) and Thorpe (1972). The Johnston Complex sample is a mesocratic medium-grained (1–5 mm) quartz diorite to tonalite principally composed of plagioclase feldspar (oligoclase to andesine) and quartz (Fig. 2). The primary mafic minerals (hornblende and biotite) are variably altered to hydrous secondary phases, including chlorite, sericite and illite. Accessory minerals within this rock include apatite, zircon, muscovite and iron–titanium oxides, including rutile and hematite (Fig. 2). Orthoclase occurs at low modal abundances (<2%) within antiperthite textures. Veinlets of pumpellyite cross-cut the sample (Fig. 2), consistent with the low-grade of metamorphism experienced by the Johnston Complex and other Neoproterozoic inliers in SW Wales (Thorpe 1972; Patchett and Jocelyn 1979). There is a general lack of significant metamorphic and deformational overprint in the sample, with only limited strained quartz, consistent with the rock fabric being dominantly magmatic.

Zircon grains from the Johnston Complex are colourless to light brown, tabular, euhedral to subhedral and are mainly intact crystals. Pristine grains have aspect ratios of ≥3:1. The grain size is variable, but does not typically exceed 200 $μ$m in length. Sharp oscillatory zoning, with well-defined core–rim relationships (based on the cathodoluminescence response), is apparent in the zircon population. Small anhedral ≤10 $μ$m apatite inclusions are found within some zircon crystals.

The zircon grains are predominantly situated on the margins of quartz or are contained within plagioclase and quartz (Fig. 3a, b). Zircon is also associated with altered hydrous phases such as chlorite (Fig. 3a). A petrographic relationship between zircon and mafic minerals has been observed in many granitoids and attributed to enhanced local Zr saturation (Bacon 1989; Elburg 1996) or greater zircon growth potential in localized volatile-rich melt pockets (Miller et al. 2007; Erdmann et al. 2013). Zircon is absent in the late pumpellyite veinlets (Fig. 3a).

Over two analytical sessions, 124 U–Pb isotopic analyses were performed on zircon grains, from which 29 were <10% discordant. In the Tera–Wasserburg plot, some zircon analyses scatter towards lower ²³⁸U/²⁰⁶Pb ratios at constant ²⁰⁷Pb/²⁰⁶Pb, interpreted to reflect recent radiogenic Pb loss (Fig. 4).

The concordant analyses define two coherent groups of ages (populations 1 and 2). Group 1 yields a concordia age of 569 ± 2 Ma [n = 27, MSWD = 1.9, p(x²) = 0.079], which is interpreted as the magmatic crystallization age for the quartz diorite, consistent with the oscillatory zoned zircon textures for this component (Fig. 2).

Population 2 is defined by two zircon core analyses and is statistically distinct from population 1, yielding an older concordia age of 615 ± 11 Ma [MSWD = 0.28, p(x²) = 0.89]. Population 2 is interpreted to represent older cores and was exclusively collected from larger zircon grains, which allowed individual cores to be analysed (Fig. 2). Given the low number of analyses, these grains could represent either rare inherited zircons assimilated from the wall rock or an antecrystic component from an earlier magmatic episode within the same longer-lived igneous system. The similarity of the zircon rare earth elements (REEs) and the Lu–Hf chemistry between populations 1 and 2 supports the latter interpretation.

Zircon REE chondrite-normalized (Boynton 1984) plots show positively sloping gradients with cerium enrichment and europium depletion (Fig. 5). The mean cerium anomaly (Ce/Ce*) for zircon from the Johnston Complex is +1.48, whereas the Eu anomaly (Eu/Eu*)_CN is strongly negative at 0.28. Some analyses show flattening of the light REE (LREE) v. medium REE (MREE) slope [LREE/MREE, (La/Sm)_CN], consistent with a gain of non-stoichiometric elements into metamict zircon. These analyses typically yield discordant U–Pb isotopic results (Fig. 4).

A total of 44 zircon Lu–Hf analyses were obtained (Fig. 6). Analyses of concordant zircon yield an $ε$Hf of −4.0 to 2.9 for population 1 with a weighted mean of −0.41 ± 1.38 (MSWD = 0.68). The population 2 $ε$Hf values range from −4.0 to −1.2 (Fig. 6). Both populations yield similar model ages with a mean Lu–Hf T_DM² of 1521 Ma.

Apatite from the Johnston Complex typically occurs as small (c. 50 $μ$m) (Fig. 3) subhedral equant grains, although the largest euhedral apatite is >120 $μ$m long. Most grains are intact with hexagonal basal sections. Apatite grains are largely colourless to light brown and internally homogenous without inclusions. Apatite is primarily associated with hydrous minerals (i.e. chlorite) and some grains occur within quartz, oligoclase and andesine (Fig. 3).

A total of 94 apatite U–Pb analyses define a discordant array on the Tera–Wasserburg concordia diagram, spanning between a ²⁰⁷Pb/²⁰⁶Pb value of 0.85 ± 0.0086 at the upper intercept to an apparent age of 576 ± 11 Ma at the radiogenic lower intercept [MSWD = 1.2; p(x²) = 0.064]. The F207% of the apatite analyses ranges from 12 to 99% with a mean value of 69% (Fig. 4). F207% defines the distance along a mixing line between the non-radiogenic (common Pb) and radiogenic Pb components, where a higher percentage implies a greater common Pb component (Kirkland et al. 2017).

The uranium concentration of the apatite ranges from 1.3 to 151.6 ppm and the mean concentration is 9.2 ppm. Thorium concentrations range from 2.9 to 203.5 ppm with a mean value of 25.8 ppm. The REE chondrite-normalized plots (Boynton 1984) show a characteristic negative slope from light to heavy REEs, with a median (La/Sm)_CN value of 2.6. The europium anomaly (Eu/Eu*) of the apatite is comparable with that in zircon at 0.35 (Fig. 7).

The ages yielded by the population 1 zircon and apatite analyses are identical within uncertainty and yield a combined weighted mean age of 570 ± 3 Ma (MSWD = 1.5) (Fig. 4), interpreted as the crystallization age of the quartz diorite component of the Johnston Complex. This c. 570 Ma age contrasts with the only previously reported U–Pb age for the Johnston Complex of 643 +5/−28 Ma (Patchett and Jocelyn 1979) (Fig. 1). Previous workers, including Patchett and Jocelyn (1979) and Thorpe (1979), concluded that the entire Johnston Complex (aside from the dolerite dykes) was emplaced at c. 643 Ma. Assuming that the multi-grain zircon fractions reported by Patchett and Jocelyn (1979) reflect a magmatic event, the apparent c. 643 Ma U–Pb age implies an earlier stage of magmatism for an inland component of the Johnston Complex. This interpretation is consistent with the observation that different multi-grain size fractions of the Patchett and Jocelyn (1979) dataset define a single lower intercept with no excess dispersion (Fig. 4). Here, we also report an older ≥615 Ma antecrystic zircon core component (population 2) within the coastal outcrop of the Johnston Complex (Fig. 4).

We analysed a sample collected 9 km to the west of the study site of Patchett and Jocelyn (1979). Geophysical data and field observations show that both outcrops of the Johnston Complex are bound within the Johnston–Benton thrust zone and share a quartz diorite composition (Thorpe 1979; Woodcock et al. 2012). Our isotopic data therefore indicate that the Johnston Complex represents discrete magmatic units with ages spanning over 40 Ma (Fig. 4). Hence previous accounts that the Johnston Complex represented a single magmatic episode at c. 650 Ma (Pharaoh and Carney 2000; Woodcock et al. 2012; Valley et al. 2014) now appear inconsistent with the younger Cryogenian history of plutonism recorded within the apatite and zircon of the western Johnston Complex.

Neoproterozoic igneous activity in southern Britain has been summarized via a three-stage model (Pharaoh and Carney 2000; Woodcock et al. 2012). Based on the new geochronology reported herein, the Johnston Complex retains information from each stage (Fig. 8). An early c. 700–640 Ma period of magmatism is recorded throughout the basement of southern Britain and is characterized by pluton emplacement at mid-crustal depths within the Wrekin Terrane – for example, the Stanner–Hanter Complex and the Malverns (Fig. 1) (Pharaoh and Carney 2000). The inland c. 643 Ma pluton of the Johnston Complex (Patchett and Jocelyn 1979) thus brackets the end of this earliest stage of magmatism.

Calc-alkaline arc plutons (the Sarn Complex, Arfon Group) and volcanic sequences (the Pebidian Supergroup) dominated a later period of magmatism from c. 620 to 585 Ma during the so-called Main Magmatic Event within the British Avalonian terranes (Pharaoh and Carney 2000). The c. 615 Ma zircon core age and c. 570 Ma zircon and apatite U–Pb age (Fig. 4) constrain the age of the western portion of the Johnston Complex to a hitherto unidentified younger expression of the Main Magmatic Event. The c. 615 Ma zircon core age is coeval with a 614 ± 2 Ma age reported for the Arfon Group (Fig. 8) (Tucker and Pharaoh 1991), indicating an interval of calc-alkaline plutonism in the Cymru Terrane at this time (Schofield et al. 2016). Furthermore, zircon U–Pb data from the Fenland Terrane also record magmatism at 615 ± 6 Ma (Noble et al. 1993). Based on the shared c. 615 Ma age and $ε$Nd signatures, Schofield et al. (2016) posited that the Fenland Terrane, Arfon Group and Sarn Complex were the disrupted remains of a single basement unit. The context of the Johnston Complex to this purported basement block remains enigmatic as we report only two c. 615 Ma grain cores; nonetheless, it is noteworthy that magmatism of this age has not been documented from the Cymru, Wrekin and Charnwood terranes (Schofield et al. 2016).

Despite their close geographical proximity (Fig. 1), the relationship between the Pebidian Group and Johnston Complex remains unresolved and this has led to these two inliers having variable terrane classifications; see McIlroy and Horak (2006) cf. Pharaoh and Carney (2000). A minimum age of c. 587+25/−14 Ma is reported for the Pebidian Group from the cross-cutting St David's granophyre (Patchett and Jocelyn 1979). The c. 570 Ma age obtained from zircon and apatite in this study, combined with the strongly calc-alkaline affinities of both inliers, points towards a genetic relationship between the Pebidian Supergroup and Johnston Complex, thereby supporting the position of the Johnston Complex within the Cymru Terrane. If so, the coastal pluton of the Johnston Complex would be constrained to the final stage of magmatism, defined by the transition from a magmatic-dominated regime to one of sedimentation and oblique terrane dispersal (Pharaoh and Carney 2000; Woodcock et al. 2012).

McIlroy and Horak (2006) posited that the Welsh Borderland fault system (Fig. 1) represented a terrane boundary between the Wrekin and Cymru terranes. The combined c. 570 Ma apatite and zircon magmatic crystallization age and their continental arc geochemical signatures support the association of the Johnston Complex with the Cymru Terrane. Northern constituents of the Cymru Terrane (e.g. the Arfon Group and Sarn Complex) share a geochemical affinity with the Pebidian Supergroup and Johnston Complex. To the east, the undated Coomb Volcanic Formation differs significantly and has intra-continental bimodal geochemical signatures (Fig. 1) (Bevins et al. 1995). Other components of the Wrekin Terrane, such as the Longmyndian and Uriconian groups, also share a pronounced intra-plate geochemistry, suggesting a decreasing influence of subduction-related magmatism towards the east (McIlroy and Horak 2006). We therefore suggest that the eastern margin of the Johnston Complex, the inferred extension of the Carreg Cennen Fault, is a terrane boundary separating the Cymru and Wrekin terranes (Fig. 1).

The relationship between the Benton Volcanic Group and the Johnston Complex is uncertain (Fig. 1). Nonetheless, the entrainment of both inliers within the Johnston–Benton thrust zone and their shared calc-alkaline chemistry (Thorpe 1972; Bevins et al. 1995) implies a common magmatic source for both units. The rhyolitic lavas and tuffs of the Benton Group are posited to represent more evolved expressions of eruptives in the Pebidian Supergroup to the north (Baker 1982; Bevins et al. 1995). Given the potential linkage between the Pebidian Supergroup and the Johnston Complex, the Benton Group warrants further analysis to determine its broader terrane affinity and relationship to the Johnston Complex. The Johnston Complex now provides constraints on the evolution of the Cymru Terrane itself, representing some of the oldest (c. 640 Ma) (Patchett and Jocelyn 1979) and youngest (c. 615–570 Ma) magmatism in the area.

Previous accounts of the geochemistry of the Cymru Terrane (Thorpe 1972; Pharaoh et al. 1987; Bevins et al. 1995) have reported calc-alkaline geochemical signatures consistent with subduction-related continental arc magmatism. Our new zircon and apatite data reinforce these earlier interpretations and provide insights into the geochemistry of the late Main Stage magmatism in the Cymru Terrane. The chondrite-normalized (Boynton 1984) REE plots for zircon from the Johnston Complex show positively sloping trends with Ce and Sm enrichment and Pr and Eu depletion (Fig. 5). Oxidized Ce⁴⁺ is more compatible with zircon than the more reduced cations of cerium (Belousova et al. 2002a). Accordingly, the positive cerium anomaly (+1.48, Fig. 5) is a broad reflection of the oxidized nature of the source magma for the Johnston Complex (Belousova et al. 2002a). Subduction-derived melts are typically oxidized, with the zircon Ce enrichment of the Johnston Complex analogous to the modern Indonesian subduction system (Woodcock et al. 2012; Bénard et al. 2018).

Zircon crystals derived from fractionated granitoid melts generally show a negative Eu anomaly, representing Eu²⁺ sequestration during oligoclase and andesine crystallization. The zircon Eu anomaly reported here (0.28) is consistent with an evolved source melt and the high abundance of plagioclase within the quartz diorite of the Johnston Complex (Fig. 2) (Appendix 1, Supplementary Material).

The geochemical data suggest that evolved calc-alkaline plutonism persisted in the Cymru Terrane to at least c. 570 Ma and the emplacement of the Johnston Complex was therefore contemporaneous with the terrane dispersal and oblique convergence associated with the final stage of Neoproterozoic magmatism in southern Britain (Pharaoh and Carney 2000). The geochemical profiles of the Johnston Complex and the Pebidian Supergroup represent an archive of the evolving tectonic setting of the Cymru Terrane from a time of mid-crustal plutonism at c. 640–615 Ma (Johnston Complex) to intercalated volcanism and sedimentation within the Welsh Basin during the Paleozoic (Pebidian Supergroup). Late Neoproterozoic <580 Ma components of the Wrekin Terrane, such as the Uriconian and Warren House groups, have geochemical affinities that shifted towards more intra-plate bimodal components. Such geochemical trends point towards later arc rifting, suggesting a more distal location from the subducting front for these units (Bevins et al. 1995; Pharaoh and Carney 2000). Our geochemical data (Fig. 5) are consistent with the Johnston Complex retaining a uniform evolved calc-alkaline magmatic signature from c. 640 to 570 Ma, suggesting a lack of source magma modification prior to terrane dispersal after c. 570 Ma.

Unlike the zircon from the Johnston Complex, the apatites do not show a pronounced cerium anomaly (Ce/Ce* = 0.02) (Fig. 7). Such a trend is typical of evolved felsic melts, where cerium is sequestered into fractionating plagioclase (London 1992). Although cerium sequestration can occur into epidote group minerals (Glorie et al. 2019), this process appears minimal in the Johnston Complex given the absence of primary epidote group minerals or allanite (Fig. 2).

Sha and Chappell (1999) demonstrated that apatites in S-type and I-type granites show a chondrite-normalized depletion in Th and LREEs and, therefore, a shallower slope, which flattens with increasing atomic mass in REE plots (Fig. 7). Conversely, I-type granites can typically show a distinct enrichment in the LREEs and heavy REEs compared with U and the MREEs – a trend not evident in apatite from the Johnston Complex. Instead, the apatite has a negatively sloping REE trend with a moderate negative Eu anomaly of 0.35 and a mean (Ce/Yb)_cn of 7.3, typical of apatite grains derived from evolved granitoids (Fig. 7) (Sha and Chappell 1999; Belousova et al. 2002b; Chu et al. 2009). Accordingly, the Johnston Complex apatite falls within the range of S-type granites compiled by O'Sullivan et al. (2020) (Fig. 7). The mineralogy of the Johnston Complex is also consistent with an S-type source, including apatite being relatively abundant (c. 0.20% modal abundance) and an absence of diagnostic minerals for I-type granites, such as allanite and magnetite (Chappell and White 2001). The predominant association of zircon within altered mafic minerals (chlorite and hornblende) in the Johnston Complex (Fig. 3) is also consistent with an S-type classification (Clemens et al. 2016).

Continental arc granitoids, such as the Johnston Complex, typically form by melting pre-existing crust, with potential input from juvenile material (Hawkesworth et al. 2010). I-type felsic melts can transition to a more peraluminous composition following assimilation with material derived from subduction, such as greywackes and shales (Chappell and White 2001). Extensive continental collision and orogenesis, associated with the assembly of Gondwana, were posited to have led to the uplift and production of giant turbidite fans (Zhu et al. 2020). The erosion of the uplifted hinterland supported the prolonged delivery of supracrustal detritus via subduction-related erosion followed by transfer to the upper plate. Our data suggest that the Johnston Complex, an S-type granite from a peri-Gondwanan terrane, is a consequence of this process.

Obtaining primary magmatic crystallization ages can be challenging for S-type granites due to the common presence of inherited zircons, with difficulties in definitively identifying neoblastic growth in such rocks (Collins et al. 2021; Kelsey et al. 2022). Differentiating primary v. inherited zircon has important implications for accurate terrane classification, where zircon U–Pb crystallization ages are often the key discrimination or association parameter between disparate inliers (Pharaoh and Carney 2000; Schofield et al. 2016).

Given solubility considerations, apatite is much less likely to be an inherited phase in felsic melts than zircon (Wolf and London 1994). Where both zircon and apatite yield similar crystallization ages, and the host rock shows little evidence of recrystallization, this provides a compelling case for coeval magmatic growth. Coeval ages for both minerals also imply rapid cooling (i.e. shallow epizonal emplacement) through the radiogenic Pb closure temperature of both minerals (zircon ≥900°C and apatite ≥350°C) (Fig. 4) (Chew and Spikings 2021).

Apatite and zircon ages at c. 570 Ma for the Johnston Complex imply that calc-alkaline magmatism was ongoing during the final stages of Eastern Avalonian arc magmatism in the Cymru Terrane (Woodcock et al. 2012). Across the inliers of southern Britain, an absence of calc-alkaline magmatism after c. 570 Ma has been attributed to the increasingly oblique convergence of East Avalonia with Laurentia and the detachment of the coherent geology into tectonic slivers and subsequent terrane dispersal (Fig. 9) (Horak et al. 1996; Woodcock et al. 2012). The rapid cooling of the Johnston Complex at c. 570 Ma and its low metamorphic grade is consistent with structurally controlled emplacement into the shallow crust. Increasing rates of oblique convergence and transpression along strike-slip faults have been proposed to reduce the geothermal gradient of the mid- to upper crustal sections (Alvarado et al. 2016; Collett et al. 2019). Hence the 570 Ma age of the Johnston Complex is consistent with dispersal of the Eastern Avalonian Terrane along the subduction front (Woodcock et al. 2012). During this process, strike-slip faults would have provided conduits for melt emplacement and crystallization of the Johnston Complex at shallow epizonal depths (Murphy et al. 2022).

In addition, these new data imply that the Johnston Complex underwent minimal thermal disturbance following magmatic crystallization at c. 570 Ma. The U–Pb closure temperature of apatite is highly dependent on microstructures, grain size and cooling rates. Consequently, the reported closure temperatures range between 350 and 600°C (Chew and Spikings 2015; Kirkland et al. 2018). Nonetheless, apatite U–Pb ages typically represent metamorphic perturbances under mid- to upper crustal conditions or protracted cooling through the closure temperature range (Chew and Spikings 2015; Kirkland et al. 2018; Gillespie et al. 2022). The similarity between apatite and zircon U–Pb ages implies that the Johnston Complex cannot have resided at temperatures ≥350°C for a protracted period. Rather, the isotopic data imply rapid cooling through the U–Pb closure temperatures of both minerals. Rapid cooling at a relatively shallow depth of emplacement is consistent with the coeval crystallization of apatite and zircon within this constituent body of the Johnston Complex (Fig. 1).

K–Ar hornblende data from another coastal outcrop of the Johnston Complex have yielded c. 624–612 Ma ages and biotite from the Dutch Gin series, a so-called metre-scale band of localized deformation at the western margin of the Johnston Complex, yields a K–Ar date of c. 612 Ma (Thorpe 1979). The relatively lower closure temperature of the K–Ar system in both minerals (300–600°C) (Harrison 1982) further suggests that the Johnston Complex experienced minimal post-magmatic thermal perturbations. The Johnston Complex is now bound within a Variscan thrust zone and was also emplaced prior to the Caledonian Orogeny, but neither tectonic event caused extensive metamorphism of the complex (Thorpe 1972; Patchett and Jocelyn 1979). Indeed, across the Neoproterozoic rocks of SW Wales, metamorphism is limited to prehnite–pumpellyite facies (Fig. 1) (Thorpe 1972).

The zircon $ε$Hf data for c. 570 Ma crystals (population 1) forms a broadly sub-chondritic cluster, with a mean $ε$Hf of −0.41 ± 1.38. The older c. 615 Ma (population 2) cores have ɛHf values that are not significantly different (Fig. 5), suggesting that both magmatic episodes were derived from compositionally similar sources (at least in terms of the Lu/Hf ratio) that did not undergo significant evolution or modification during the c. 615–570 Ma interval.

A Lu/Hf ratio of 0.05 was chosen to reflect a felsic source for the parental melt of the Johnston Complex, consistent with other S-type Proterozoic granitoids from the database compiled by Bea et al. (2018). Two-stage depleted mantle Lu–Hf model ages (T_DM²) imply late Mesoproterozoic 1011–1892 Ma crust-extraction dates, with a mean T_DM² of 1521 Ma. Model ages, especially in the case of S-type granitoids, may only provide a minimum age for the oldest component within a mixed magma. Nonetheless, the zircon Hf isotopic signature, with a source crustal residence time of at least 800–500 Ma, is similar to other Avalonian granitoids (Fig. 6) (Pollock et al. 2015; Henderson et al. 2016; Schofield et al. 2016).

A whole-rock $ε$Nd value of −1.8 with a T_DM² of 1437 Ma was reported for the Johnston Complex from the inland outcrop (Schofield et al. 2016). Along with the Lu–Hf T_DM² for zircon reported here, these data imply variable degrees of crustal contamination. Complementary ɛHf data from Western Avalonia granitoids yield analogous late Mesoproterozoic T_DM² ages, implying a similar source for most Avalonian microcontinental blocks during the late Neoproterozoic (Fig. 6) (Willner et al. 2013; Pollock et al. 2015; Henderson et al. 2016).

Although localized areas of considerably older Proterozoic ≥1.6 Ga crust in West Avalonia have been inferred by $ε$Hf data (Willner et al. 2013; Pollock et al. 2015; Henderson et al. 2016), there is no evidence of similarly ancient crust based on the sparse Lu–Hf dataset for the Cymru Terrane. Nonetheless, the presence of older, mid-Proterozoic crustal components within the Cymru Terrane is implied by Sm–Nd model ages of c. >1.35 Ga for the Parwyd Gneisses and Sarn Granites (Fig. 1) (McIlroy and Horak 2006). Other areas show more definitive evidence of older underlying crust. For example, the Malverns Complex (Fig. 1) has yielded inherited zircon age maxima of c. 1.6 Ga and whole-rock Sm–Nd model ages of c. 1.6–1.4 Ga (Murphy et al. 2000; McIlroy and Horak 2006). Such older crustal components provide an important distinction between the Cymru and Wrekin terranes, in which the latter appears to incorporate an older crustal component.

Zircon data from other East Avalonian lithologies also show similar $ε$Hf patterns to the Johnston Complex. For example, zircon grains from the Stavelot–Venn Massif of East Avalonia imply felsic melt generation from a chondritic source of similar age and composition to the Johnston Complex (Fig. 6) (Willner et al. 2013). Hence $ε$Hf data from the Johnston Complex, when considered with data from other terranes, support the interpretation that Avalonian terranes formed from a relatively uniform Neoproterozoic continental arc system with a broadly chondritic melt source (Fig. 6) (Pollock et al. 2015) variably contaminated with older Meso- to Paleoproterozoic crust (e.g. the Wrekin Terrane) (Thorpe et al. 1984). This contamination could represent localized input into the subduction system of allochthonous structurally emplaced older crust during the Proterozoic (Pollock et al. 2015).

New time-constrained (U–Pb) isotopic (Lu–Hf) and trace element data from the Johnston Complex, SW Wales help to constrain magmatism in East Avalonia during the late Neoproterozoic. We report ages for a coastal pluton of the Johnston Complex from coeval apatite and zircon at 570 ± 3 Ma. Combined with a previously reported age of c. 643 Ma for an inland component, we demonstrate that the Johnston Complex is a composite suite of quartz diorite plutons along its c. 20 km length. Our data necessitate that calc-alkaline igneous activity was ongoing until c. 570 Ma in East Avalonia.

Apatite and zircon geochemistry provide complementary insights into late Neoproterozoic magmatism in southern Britain. Zircon trace elements demonstrate an oxidized calc-alkaline source magma typical of an ensialic subduction belt setting for the Johnston Complex. A sedimentary component within the melt is revealed through apatite trace element profiles, consistent with voluminous S-type granitoid production during the assembly of Gondwana. $ε$Hf values imply a broadly chondritic source melt for both zircon populations with a mean T_DM² of 1.5 Ga.

Classification of the Precambrian terranes of southern Britain has primarily been facilitated through correlating geochemical signatures, structural relationships and Nd (model) ages. Generally, sparse U–Pb magmatic crystallization dates have underpinned associations of otherwise disparate lithologies into unified terranes. Given this approach, the work presented here offers a cautionary note. On the kilometre scale, the Johnston Complex varies in age by more than c. 40 Ma (Fig. 3). Although individual zircon grains trace a multi-stage history with components at c. 615 and 570 Ma, the combined zircon–apatite approach gives confidence in magmatic age assignment and facilitates terrane association in partially obscured basement areas. Thus, in situations when the closure temperature of Pb in apatite has not been exceeded, the application of combined zircon and apatite geochronology may remove some of the ambiguity related to zircon inheritance when dating rocks.

The authors thank Bradley McDonald and Noreen Evans for assistance with the LA-ICP-MS data collection. We thank Brendan Murphy and Damian Nance for reviews that improved this work and Yildirim Dilek for facilitating a constructive review process.

AJIC: formal analysis (lead), visualization (lead), writing – original draft (lead); CLK: conceptualization (lead), funding acquisition (lead), resources (lead), supervision (lead), writing – review and editing (lead); SG: funding acquisition (equal), supervision (supporting), writing – review and editing (supporting).

This work was funded by an Australian Research Council Discovery Project (DP200101881). The instruments in the John deLaeter Centre, Curtin University, were funded via AuScope, the Australian Education Investment Fund, the National Collaborative Research Infrastructure Strategy and the Australian Government.

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).