U-Pb zircon constraints on obduction initiation of the Unst Ophiolite: an oceanic core

7 The Unst Ophiolite is the best exposed of a chain of early Ordovician ophiolites in the 8 Scottish Caledonides and is widely regarded as having formed in a supra-subduction zone 9 setting within the Iapetus Ocean. Reinterpretation of sheeted dykes suggests that it formed 10 as an oceanic core complex, presumably during subduction roll-back immediately prior to 11 obduction onto the Laurentian margin. A new U-Pb zircon age of 484 ± 4 Ma for 12 development of the metamorphic sole places a lower limit on the timing of obduction, 13 which was subsequently followed by regional-scale crustal thickening and Barrovian 14 metamorphism during the Grampian orogenic event. 15 Supplementary material: Analytical methods, CL images of representative zircon grains, and 16 a LA-ICP-MS U-Th-Pb zircon data table are available at www.geolsoc.org.uk/SUPXXXXX. 17 Ophiolites represent slices of oceanic-type lithosphere that have been tectonically 18 incorporated into continental margins in convergent plate settings (e.g. Dewey & Bird 1970; 19 Coleman 1971). Many ophiolites were formed in supra-subduction zone (SSZ) arc-forearc 20 settings shortly prior to orogenesis and then thrust (obducted) onto a colliding passive 21 margin (e.g. Davies & Jaques 1984; Searle & Cox 1999). The metamorphic soles present 22 beneath many ophiolites are thought to have resulted from high-temperature (± pressure) 23 metamorphism of a subducting oceanic slab beneath hot sub-ophiolitic mantle and later 24 accretion onto the base of the ophiolite during obduction (e.g. Dewey & Casey 2013 and 25 references therein). One of the most significant findings of the last 20 years of research into 26 modern ocean basins has been the identification of oceanic core complexes that expose 27 upper mantle and lower crustal rocks on the seafloor (e.g. Cannat 1993; Tucholke & Lin 28 1994). Although not recorded in many modern fore-arc settings, they have been identified 29 in some SSZ ophiolites (e.g. Suhr & Cawood 2001; Tremblay et al. 2009), thus representing 30 important opportunities for direct study of oceanic core complexes which are otherwise 31 difficult to access. 32 The Unst Ophiolite in the Shetland Islands (Fig 1) is the northernmost and best 33 exposed of a chain of SSZ ophiolites within the Laurentian Caledonides of Scotland and 34 Ireland (e.g. Williams & Smyth 1973; Flinn 1999 and references therein). Other examples 35 include the Bute, Ballantrae and Tyrone ophiolites and the Clew Bay Complex (Fig 1a). These 36 ophiolites were emplaced during the early to mid-Ordovician Grampian orogenic event 37 which records the initial stages in closure of the Iapetus Ocean (Chew et al. 2010). The 38 collision of a juvenile oceanic arc with the Laurentian passive margin resulted in NW39 directed ophiolite obduction (present reference frame) with regional deformation and 40 metamorphism of footwall metasedimentary successions (e.g. Dewey & Shackleton 1984; 41 Chew et al. 2010). The formation and subsequent obduction of the Unst Ophiolite is 42 currently bracketed between its magmatic age of 492 ± 3 Ma (U-Pb zircon, Spray & Dunning 43 1991) and K-Ar mineral ages obtained from the metamorphic sole which range from 479 ± 6 44 Ma to 465 ± 6 Ma (Spray 1988). In this paper we: 1) report the results of U-Pb dating of 45 metamorphic zircon from the sole of the ophiolite, thus placing a lower limit on the timing 46 of obduction , and 2) re-evaluate published geological data from the ophiolite to show that 47 these are consistent with an origin as a SSZ oceanic core complex with limited magma 48 supply. 49 Regional geological setting and internal architecture of the Unst ophiolite The Unst 50 Ophiolite consists of two main nappes of ultramafic and mafic rocks and associated low51 grade metasedimentary rocks and melanges (Fig 1c; Flinn 1999, 2000, 2001 and references 52 therein). The overall structure is a down-folded klippe as suggested by the open synformal 53 folding of the upper nappe on the island of Fetlar, 5 km south of Unst, and modelling of 54 gravity and magnetic anomalies (Flinn 2000). The maximum depth to the tectonic base of 55 the ophiolite is unlikely to be greater than 3 km (Flinn 2000). On Unst, the lower ophiolite 56 nappe rests on metasedimentary rocks of the mid-Neoproterozoic to Cambrian Dalradian 57 Supergroup (Fig 1b; Flinn 1999). A U-Pb monazite age of 462 ± 10 Ma obtained from 58 Dalradian rocks in Unst is thought to date peak metamorphism during the Grampian 59 orogenic event (Cutts et al. 2011). The gently-dipping to steep fault that underlies the lower 60 nappe is not the original obduction thrust but the result of later reworking of the nappe pile 61 (Cannat 1989; Flinn & Oglethorpe 2005; Cutts et al. 2011). 62 The lower nappe consists of three steeply-inclined to sub-vertical layers with a total 63 maximum thickness of c. 7 km measured normal to lithological boundaries (Fig 1c; Flinn 64 1999, 2001). From northwest (structurally lowest) to southeast (highest) these comprise 65 metaharzburgite, metadunite (both extensively serpentinized) and metagabbro (Fig 1c), 66 which have been affected by greenschist facies metamorphism. A plagiogranite within a 67 metagabbro yielded a U-Pb zircon age of 492 ± 3 Ma (Spray & Dunning 1991). Uppermost 68 parts of the metagabbro in SE Unst are intruded by sub-parallel mafic dykes, separated by 69 narrow screens of host metagabbro (Prichard 1985; Flinn 1999). The dykes have been 70 interpreted as the base of a sheeted dyke complex (Prichard 1985; Spray 1988). The 71 boninitic chemistry of some of these dykes provides evidence that the ophiolite formed in a 72 SSZ setting (Prichard & Lord 1988; Spray 1988; Flinn 2001). Trace element discrimination 73 diagrams indicate a range of volcanic arc basalt/island arc-like compositions (Flinn 2001). 74 However, these dykes depart from classic sheeted dykes described from many other 75 ophiolites because they never form >50% of the rock by volume in any area and are nearly 76 parallel to the ophiolite layering than normal to it, so at present they occur as sub-vertical 77 sheets (Flinn 2001). 78 The Unst Ophiolite does not contain any pillow lavas. Instead, the metagabbro layer 79 of the lower nappe, together with its ‘sheeted dykes’, is overlain by dark, locally graphitic, 80 quartz-sericite-chlorite meta-siltstones of the Muness Phyllites (Fig 1c). The presence of 81 metagabbro and plagiogranite clasts, as well as elevated concentrations of Cr, suggests that 82 the underlying ophiolitic rocks contributed detritus to the sedimentary protoliths of the 83 Muness Phyllites. These phyllites are therefore thought to have been deposited 84 unconformably on the lower nappe (Flinn 1985). Exposed contacts between the Muness 85 Phyllites and the metagabbro occur as faults, but none of these are thought to be regionally 86 significant structures. The structurally overlying upper nappe is dominated by 87 metaharzburgite, and thought to result from the tectonic duplication of the lower nappe, 88 most probably during the c. 435-430 Ma Scandian orogenic event (Flinn & Oglethorpe 2005). 89 Both nappes are underlain by discontinuous tectonic slices of amphibolite and 90 hornblende schist which were interpreted by Williams & Smyth (1973) as a metamorphic 91 sole to the ophiolite. Geochemical studies indicate a MORB-type basaltic origin for the 92 igneous precursors (Spray 1988; Flinn 2001). On Unst an early amphibolite facies mineral 93 assemblage of hornblende + plagioclase + titanite ± apatite shows evidence for greenschist 94 facies retrogression to actinolite + albite + epidote (Spray 1988). On Fetlar, texturally early 95 garnet-clinopyroxene assemblages preserve evidence for upper amphibolite facies P-T 96 conditions of 700-780°C and 9-11 kbar (Spray 1988; Flinn et al. 1991). At several localities, 97 the metabasic rocks contain a well-developed gneissic fabric defined by cm-scale 98 trondjhemitic layers interpreted to be the result of high-temperature segregation. K-Ar 99 hornblende ages obtained from various samples of the metamorphic sole (lower and upper 100 nappe) range from 479 ± 6 Ma to 465 ± 6 Ma. 101 Sample descriptions To further constrain the timing of formation of the metamorphic sole, 102 we investigated two samples: 11-SH-10 (collected from beneath the lower nappe on Unst at 103 HP 5635 0067) and 11-SH-12 (collected from beneath the upper nappe on Fetlar at HU 6450 104 9206). Both samples are amphibolites, with 11-SH-10 being medium-grained and 11-SH-12 105 being medium to fine grained. Both amphibolites posess a well-developed foliation defined 106 by alternating mafic and felsic layers. Mafic layers are dominated by aligned grains of 107 hornblende, which in the case of sample 11-SH-10 wrap skeletal garnets up to 5 mm in 108 diameter. Titanite and apatite occur as visible accessory minerals. Felsic layers are 109 dominantly composed of sericitised plagioclase with minor quartz. There is evidence for 110 widespread replacement of hornblende by actinolite and biotite, and garnet (where 111 present) by chlorite. Scattered grains of secondary epidote are also common. 112 U-Pb zircon data: results and interpretation. Cathodoluminescence (CL) images of zircon 113 grains show sector and fir-tree zoning patterns and large parts of grains which are 114 homogeneous in CL, all of which is consistent with a single phase of metamorphic growth 115 (Corfu et al. 2003), which most likely occurred during upper amphibolite facies P-T 116 conditions as previously recognised from the early mineral assemblage (Spray 1988) . No 117 inherited cores and no igneous zircon were evident, compatible with a MORB-type protolith 118 (Spray 1988; Flinn 2001). Representative zircons from both samples were analysed by LA119 ICP-MS following the operating conditions of Crowley et al. (2014 see supplementary files 120 for data table) and were found to have low U concentrations (generally <3ppm for sample 121 11-SH-10 and <0.3ppm for sample 11-SH-12). 69 zircon analyses from 11-SH-10 are 122 concordant to near-concordant (Fig 2) and do not appear to have suffered any significant 123 Pb-loss. A weighted mean Pb/U age calculation from these data returns a date of 483.7 124 ± 4.4 Ma (MSWD=1.18), interpreted as the age of formation of the metamorphic zircon. 125 Despite a large number of analyses, only one age determination was possible from 11-SH126 12. This grain was found to have relatively elevated U concentration (c. 5ppm) compared to 127 other zircons from the same sample and yielded a Pb/U age of c. 482 ± 18 Ma (i.e. 128 within error of the age determination for sample 11-SH-10). 129 An oceanic core complex model for the Unst Ophiolite Oceanic core complexes are known 130 to develop in conjunction with some intermediate to slow spreading centres and represent 131 exhumed lower crustal and upper mantle rocks in the footwalls of extensional detachments 132 (Whitney et al. 2013; Platt et al. in press). They typically exhibit a dome-shaped massif, 133 which may display considerable topographic relief relative to the surrounding sea-floor. 134 Oceanic core complexes are usually cored by gabbro, or gabbro and peridotite and do not 135 exhibit the classic Penrose-type ophiolite sequence (Miranda & Dilek 2010). They have been 136 described from segments of the Australian-Antarctic Discordance, Caribbean–North 137 American Ridge, Mid-Atlantic Ridge, Southwest Indian Ridge and from several back-arc 138 spreading centres (Whitney et al. 2013 and references therein). Several oceanic core 139 complexes occur in the inner-bend of ridge-transform intersections, so there is likely an 140 interplay of magmatic and tectonic processes, the mutual rates of which are critical to 141 initiating accelerated extension in magma-poor areas of divergent zones (Whitney et al., 142 2013). Slow spreading rates of <55 mm yr may result in episodic and spatially variable 143 magma supply, which in turn facilitates an increase in tectonic partitioning, accelerated 144 lithospheric thinning and denudation of oceanic crust (Miranda & Dilek, 2010). Furthermore, 145 the rate of magma supply (i.e. not too rapid and not too slow) appears to be a critical factor 146 in the development of oceanic core complexes (Tucholke et al. 2008). From a structural 147 perspective, the presence of pre-existing inward-dipping normal fault systems, coupled with 148 lithospheric thinning results in a passive rotation or flexure of faults (Morris et al. 2009), 149 resulting in the formation of low angle detachments. Tucholke et al. (2008) state that if 150 dykes and plutons collectively account for 50% or more of the total extension, then 151 detachment faults initiated at spreading centres may accumulate large displacements. 152 Any model for the formation of the Unst Ophiolite has to account for the steep 153 orientation of the internal layering relative to the assumed gently-dipping basal thrust (Fig 154 1c), the intrusion of the boninitic dykes sub-parallel to the steep layering within host upper 155 gabbro and their absence from underlying units. In contrast to Spray (1988) who envisaged 156 formation within an SSZ setting sensu stricto, Flinn (2001) suggested that it was derived 157 from a subducting oceanic slab. In this latter interpretation, the SSZ chemistry was derived 158 from flushing the subduction zone with fluids from the dehydrating descending slab. These 159 fluids promoted partial melting of the overlying wedge, forming basaltic melts that were 160 intruded into the upper gabbro as dykes from above and to the side. Flinn (2001) envisaged 161 that during arc-continent collision, a thrust that initiated in the overlying mantle wedge cut 162 down through the ‘stratigraphy’ of the subducting plate, then more parallel to it, eventually 163 acting as the obduction thrust for the ophiolite as it was thrust onto the Laurentian margin. 164 The Muness Phyllites were thought to have been deposited on the ophiolite after it was 165 obducted. 166 In contrast, we suggest that all the main features of the Unst Ophiolite are more 167 easily accounted for if it is viewed as having originated as an oceanic core complex that 168 developed in association with a spreading ridge in the SSZ plate. It should be noted that 169 Flinn (1999) in considering field relationships within the ophiolite suggested: “It is possible 170 that the slice of crust and upper mantle forming the Lower Nappe slipped and rotated into a 171 recumbent position on a listric fault at the constructive margin prior to the intrusion of the 172 dykes” but did not develop this model any further. We justify our interpretation as follows; 173 firstly, the steep orientation of the internal layering of the ophiolite (Fig 1c) can largely be 174 accounted for by rotation of the oceanic lithosphere in the footwall of a major detachment 175 that was initiated as a steep normal fault but was rotated to a gently-dipping orientation 176 during progressive extension. Rotations of 50-80° have been demonstrated at slow177 spreading sectors of mid-ocean ridges (e.g. Garcés & Gee 2007; Morris et al. 2009) and 178 could also account for the otherwise enigmatic steep orientation of the earliest and 179 dominant magnetization direction within the Unst Ophiolite (Taylor 1988). Secondly, the 180 occurrence of dykes sub-parallel to the layering within the upper gabbros, and their absence 181 from underlying units, can be accounted for by ongoing intrusion and dyke emplacement 182 into this part of the ophiolite after rotation had been substantially completed. Garcés & 183 Gee (2007) documented this process along the Fifteen-Twenty Fracture Zone in the Central 184 Atlantic Ocean. Thirdly, no evidence precludes the possibility that the Muness Phyllites were 185 deposited on the ophiolite before obduction. The generally low energy nature of the fine186 grained sedimentary protoliths, albeit interrupted occasionally by high-energy debris flows, 187 is entirely consistent with deposition in a deep-water oceanic setting on (and derived from) 188 gabbros that were exposed on the sea-floor following detachment faulting and initiation of 189 core complex formation. Such a predominantly low-energy sedimentary facies is unlikely to 190 be deposited on top of an obducted ophiolite. 191 192 Discussion and implications The new interpretations and data presented here support a 193 new tectonic model for the Unst Ophiolite. The timing of initiation of oceanward-subduction 194 is poorly constrained but is likely to have occurred by c. 510 Ma (Chew et al. 2010), followed 195 by subduction roll-back and development of the precursor to the SSZ ophiolite at 492 ± 3 196 Ma (Spray & Dunning, 1991). A c. 18 m.y. duration between subduction initiation and 197 formation of the Unst Ophiolite and its boninites can be explained by ophiolite development 198 in an infant arc environment which subsequently became the fore-arc after initiation of 199 steady-state normal oceanic subduction. Spreading rates within the Unst Ophiolite 200 precursor are likely to have been low to moderate, as documented by the relatively low 201 volume of sheeted dykes. Crystallisation of the Unst Ophiolite protolith is immediately prior 202 to development of the metamorphic sole (484 ± 4 Ma; this paper) suggesting that they 203 formed as part of the same subduction system. Such a continuum of magmatism associated 204 with sea-floor spreading events followed by obduction initiation within a few m.y. is 205 documented by other SSZ ophiolites (e.g. Semail Ophiolite; Styles et al., 2006). Continued 206 subduction roll-back resulted in development of detachment faults, rotation of the oceanic 207 lithosphere, and exhumation of mantle and lower crust on the sea floor, accompanied by 208 deposition of the sedimentary protoliths of the Muness Phyllites. We agree with Spray 209 (1988) that arc-continent collision and ophiolite obduction onto the Laurentian passive 210 margin was probably initiated by c. 480 Ma, as constrained by our new zircon age for 211 subduction initiation in the Unst Ophiolite. There is a c. 20 m.y. gap between formation of 212 the metamorphic sole to the ophiolite and peak Barrovian metamorphism in the footwall 213 Dalradian Supergroup metasediments in Shetland at c. 462 Ma (Cutts et al. 2011). This 214 confirms the notion of Chew et al (2010) and Cutts et al (2011) that peak P-T conditions 215 recorded in Dalradian pelites are more likely to have been caused by Grampian orogenic 216 collisional thickening and convective heat transfer within this over-thickened crust, rather 217 than being directly related to burial following ophiolite obduction. 218