1. Introduction
It is axiomatic that when regional-scale geological phenomena coincide in space and time, they are likely to represent responses to the same geotectonic cause. Two such are the later stages of Caledonian ‘Newer Granite’ magmatism in Scotland and deposition of the Lower Old Red sandstone, coincidence that has long been recognized (e.g. Thirlwall, 198l) but seen as a paradox: if the magmatism was associated with orogenic convergence, how could its major development coincide with post-collisional sedimentation?
The tectonic setting of the Newer Granite suite of Britain and Ireland is undoubtedly complex. These intrusions occur in the orthotectonic Caledonides to the north of the Highland Border fault and in the slate belts to the south, on both sides of the Iapetus suture (Fig. 1). Their ages lie mainly in the range 435–380 Ma, post-dating the Grampian orogeny (with which the ‘Older Granites’ are associated), overlapping the latest phases of Iapetus closure (the Scandian orogeny and ‘soft-docking’ of Eastern Avalonia) and extending through the subsequent Acadian deformation of the slate belts (see Soper et al. Reference Soper, Strachan, Holdsworth, Gayer and Greiling1992 and Dewey & Strachan, Reference Dewey and Strachan2003 for discussion of the geometry and timing of these events). The granitic magmas were generated and emplaced in a collage of diverse lithospheric segments and it is scarcely surprising that their origin has proved a difficult petrogenetic problem. None of the models proposed for parts of the suite (Iapetan subduction, more complex subduction systems, slab break-off, transpressive crustal thickening and anatexis, post-orogenic decompressional melting) offer a comprehensive explanation for the whole (see discussions by Soper, Reference Soper1986; Brown, Reference Brown and Craig1991; Stephenson & Highton, Reference Highton, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999).
Outline tectonic map of Britain and Ireland showing the main crustal lineaments discussed in the text and the ‘Newer’ Granites and associated Devonian volcanic rocks.
Within the Scottish Newer Granites, chemical and isotopic criteria enabled Stephens & Halliday (Reference Halliday1984) to make a threefold division into the Argyll, Cairngorm and South of Scotland suites, demonstrating the provincialism of source composition (Fig. 1). Stone, Kimbell & Henney (Reference Stone, Kimbell and Henney1997) recognized two distinct subgroups in the South of Scotland suite, those occurring north and those south of the Orlock Bridge fault and associated Moniaive shear zone (Fig. 2). The northern granites are older (>410 Ma) and characterized by low 87Sr/86Sr initial ratios, while to the south such plutons as Fleet, Criffel and Cheviot are younger (<410 Ma) and have more radiogenic initial 87Sr/86Sr ratios. The southern subgroup of Stone, Kimbell & Henney (Reference Stone, Kimbell and Henney1997) was named the Galloway Suite by Highton (Reference Highton, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999). It has much in common with the Lake District granites such as Shap and Skiddaw (see Sections 3.a.6 and 3.a.7) that were intruded south of the Solway Line, the surface trace of the N-dipping Iapetus suture. We propose the recognition of a Trans-Suture Suite to include the Galloway and northern England granitic intrusions, primarily on the basis of age and tectonic setting, and also the common presence of a notable component of sedimentary origin in the genesis of the magmas. The Trans-Suture Suite can be extended along-strike to include representatives in the Isle of Man (Foxdale, Dhoon) and the east of Ireland (principally Newry in the Longford-Down extension of the Southern Uplands and the large Leinster intrusion to the south of the suture; Fig. 2). Their origin has long been a puzzle.
Geological map of the Iapetus Suture region in northern Britain and central Ireland. The Trans-Suture Suite of granites and their probable sub-crops are shown. The principle zone of lamprophyre emplacement is marked with diagonal lines, and rose diagrams for the trends of the lamprophyre dykes are given for three regions.
Obviously, the granites of the Lake District and Leinster could not have been generated by northward subduction beneath Scotland, while the Galloway plutons, although emplaced in the hangingwall of the suture, lie too close to its surface trace to accommodate an arc–trench gap (Thirlwall, 198l; Soper, Reference Soper1986). Furthermore, it is now evident that Iapetan convergence ceased at about the end of Silurian time, 20 to 40 Ma before emplacement of the Trans-Suture Suite granites (Soper & Woodcock, Reference Soper and Woodcock2003). Therefore, if the suite is related to subduction, a post-Iapetan subduction system of Early Devonian age would have to be invoked. Northward Rheic subduction appears to be ruled out as a source by the close spatial association of the Trans-Suture Suite with the (fossil) Iapetus suture zone and its absence further south, in Wales for example.
A non-subduction origin for the Trans-Suture Suite granites thus needs to be considered. The key lies in understanding the tectonic setting of the Iapetus suture zone in Early Devonian time, when the plutons were generated and emplaced. It has recently been recognized that in the northern Caledonides, Early Devonian time was a period of sinistral transtension (Dewey & Strachan, Reference Dewey and Strachan2003). In Britain and Ireland this was associated with widespread Old Red Sandstone magnafacies sedimentation and lamprophyric magmatism, and was terminated by Acadian transpressive deformation towards the end of Early Devonian time (Soper & Woodcock, Reference Soper and Woodcock2003). In this paper we explore genetic links between these phenomena and the intrusions of the Trans-Suture Suite.
2. Siluro-Devonian tectonics of the Iapetus suture zone in Britain and Ireland
Figure 3 is a tectonostratigraphic summary diagram for central Britain through Silurian and Devonian time, simplified from Soper & Woodcock (Reference Soper and Woodcock2003, their fig. 7). It illustrates that, within the sinistrally oblique Caledonian tectonic context, deformation in the suture zone alternated between transpression and transtension.
Tectono-stratigraphic framework for Britain during Siluro-Devonian time (after Soper & Woodcock, Reference Soper and Woodcock2003). Note the onset of transtension after closure of the Iapetus Ocean at c. 420 Ma which ceases at about 400 Ma with Acadian transpression. The emplacement of the Trans-Suture Suite occurs near the end of, or soon after, the cessation of this phase of transtension.
2.a. Silurian convergence to c. 420 Ma
There is a well-established link between northward subduction of Iapetus lithosphere beneath the Laurentian margin (Leggett, McKerrow & Soper, Reference Legget, McKerrow and Soper1983), southward progradation of the Southern Uplands Longford-Down accretionary prism in the hangingwall of the suture (Barnes, Lintern & Stone, Reference Barnes, Lintern and Stone1989) and formation of the Windermere Supergroup flexural basin in northern England in the footwall (Kneller, Reference Kneller1991). More contentious has been the question of when Iapetus convergence ended. Popular interpretation has related the Acadian deformation in southern Britain to continued convergence and southwardly prograding imbrication of the Avalonian footwall through Early Devonian time (Kneller, King & Bell, Reference Kneller, King and Bell1993). It is now apparent that throughout the North Atlantic Caledonides, Early Devonian time was period of sinistral transtension (Dewey & Strachan, Reference Dewey and Strachan2003). Soper & Woodcock (Reference Soper and Woodcock2003) have discussed the ending of Iapetus convergence in Britain in that context. Accretion in the Southern Uplands slowed in mid-Wenlock time (Kemp, Reference Kemp, Leggett and Zuffa1987) and the youngest accreted package exposed in the Southern Belt is of late Wenlock age. Emplacement of late Caledonian minor intrusions spanned the ending of accretion. A suite of K-lamprophyre dykes is concentrated in the Central Belt of the Southern Uplands. Early examples were deformed by the accretionary deformation, but the majority are post-tectonic and have given ages in the range 400–418 Ma (Rock, Gaskarth & Rundle, Reference MacDonald, Rock, Rundle and Russell1986). It is inferred that accretion in the Southern Uplands ended at about 420 Ma (Ludlow) and was succeeded by a period of sinistral transtension, thus marking the end of Iapetus convergence and, by definition, of the Caledonian orogeny.
The presence of early, deformed lamprophyre dykes in the Southern Uplands (Barnes, Rock & Gaskarth, Reference Rock, Cooper and Gaskarth1986) suggests that the switch from transpressive accretion to transtension was not abrupt; either the two deformation modes alternated for a period during late Silurian time, or they were partitioned vertically in the lithosphere so that lamprophyric liquid was generated at depth and emplaced into the active accretionary complex during periods of stress release.
2.b. Devonian transtension c. 420–400 Ma
The evidence for orogen-wide sinistral transtension in Early Devonian time has been outlined by Dewey & Strachan (Reference Dewey and Strachan2003). The great spread of Old Red Sandstone magnafacies sediment that covered much of Britain south of the Scottish Highlands was inferred by Soper & Woodcock (Reference Soper and Woodcock2003) to have been deposited in coalescing transtensional basins for which a β factor of 1.5 was suggested.
In the present context of ‘late Caledonian’ magmatism, critical evidence is provided by the swarm of K-lamprophyre dykes mentioned above. The K–Ar ages of dykes from the suture zone (Southern Uplands and Longford-Down in the hangingwall, eastern Lake District in the footwall) range from 420 Ma to 400 Ma, defining a period of lithospheric (trans)tension in the Early Devonian period that separated the Iapetan and Acadian convergence regimes (Fig. 3).
Lamprophyre dyke trends in the suture zone can be analysed in relation to the general model for transtension developed by Dewey (Reference Dewey2002), which involves the combination of coaxial and non-coaxial strains. Mean orientations are consistent with a sinistral non-coaxial ‘Caledonide-parallel’ strain component plus a coaxial ‘Caledonide-normal’ component (see discussion by Soper & Woodcock, Reference Soper and Woodcock2003). This confirms that both the hanging- and footwalls of the suture zone in Britain experienced the period of sinistral transtension in Early Devonian time that Dewey & Strachan (Reference Dewey and Strachan2003) identified regionally.
2.c. Acadian deformation c. 400–390 Ma
Compared to its major development in the northern Appalachians (e.g. Bradley et al. Reference Bradley, Tucker, Lux, Harris and McGregor2000; Murphy & Keppie, Reference Murphy and Keppie2005), the Acadian deformation in the slate belts of southern Britain and Ireland represents a short-lived event that produced little more than anchimetamorphic rocks at the present surface. Geochronometric evidence for its age has been reviewed by Soper & Woodcock (Reference Soper and Woodcock2003, their Fig. 8). K–Ar dating of illite in cleaved mudrocks in the Central Welsh Basin gave mean of 399±3 Ma (ten determinations: Evans, Reference Evans1996). Ar–Ar dating of cleavage mica in strain fringes in Welsh mudrocks has given a high precision age of 396.1±l.4 Ma (late Emsian: Sherlock et al. Reference Sherlock, Kelly, Zalasiewicz, Scholfield, Evans, Merriman and Kemp2003).
The emplacement of Trans-Suture Suite granite plutons in the Lake District overlapped the Acadian deformation. The structural relationships of individual intrusions are detailed below, together with the somewhat inadequate geochronological evidence (Section 3) (modern zircon studies are lacking) that is currently available for their age of emplacement. However, their mean age of 397±2 Ma (five determinations) is consistent with K–Ar date of 397±7 Ma on cleavage-parallel illite from a Silurian bentonite (Merriman et al. Reference Merriman, Rex, Soper and Peacock1995), indicating a late Emsian age for the Acadian deformation in NW England.
Stratigraphically, the ending of Acadian deformation is not precisely constrained in Britain. The preserved Upper Old Red Sandstone overstep sequence commences in the Famennian, but perhaps earlier in the Lake District (Fig. 3). In the west of Ireland, however, the Acadian unconformity occurs between the Dingle and Smerwick groups, of late Silurian to Pragian or early Emsian age, and the overlying Eifelian Pointagare Group (Richmond & Williams, Reference Richmond, Williams, Friend and Williams2000). The Acadian tectonometamorphism in Britain and Ireland thus appears to have occupied a brief period between 400 and 390 Ma, close to the Lower–Middle Devonian boundary, with no evidence that the deformation prograded in any particular direction.
2.d. Evidence of enhanced heat flow
There is evidence that the slate belt deformation took place under enhanced heat flow. In the Welsh Basin, estimates of the palaeogeothermal gradient at the time of peak Acadian metamorphism range from 36°C km−1 at a depth equivalent to a lithostatic pressure of 0.30–0.36 GPa (based on fluid inclusion measurements: Bottrell et al. Reference Bottrell, Greenwood, Yardley, Shepherd and Spiro1990) to 52°C km−1 (based on metabasite mineral assemblages: Bevins & Merriman, Reference Bevins and Merriman1988). Direct determinations have not been made on Silurian slates in NW England, where the presumption is that low gradients of 25°C km−1 or less are likely to have characterized the Silurian flexural basin phase (Merriman, Reference Merriman2002). However, as argued by Soper & Woodcock (Reference Soper and Woodcock2003), this gradient must have been greatly enhanced during Early Devonian time, otherwise an implausible thickness of Lower Old Red Sandstone cover must be invoked to produce the anchizonal metamorphism seen in late Silurian slates at the present surface. A link between transtension and enhanced heat flow may lie in the Early Devonian lamprophyres; this possibility is explored quantitatively in Section 6.
3. The Trans-Suture Suite of granite intrusions in Britain and Ireland
Numerous authors have remarked on similarities between the granitic intrusions in the north of England and southern Scotland (e.g. Harmon & Halliday, Reference Harmon and Halliday1980; Halliday, Reference Halliday1984; Harmon et al. Reference Harmon, Halliday, Clayburn and Stephens1984; Stephens, Reference Stephens, Harris and Fettes1988; Thirwall, Reference Thirwall1989; Highton, Reference Highton, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999), noting in particular evidence for the importance of an input of sedimentary origin, there being peraluminous, two-mica granites in both areas. It should be noted that the Lake District intrusions of Eskdale and Ennerdale are to be excluded from these comparisons, having been shown to be of Ordovician rather than Acadian age (Millward & Evans, Reference Millward and Evans2003). Information on individual intrusions is considered below, where an unfortunate disparity is evident between the comprehensive amount of data available for the Galloway Suite and the sparser data for those intrusions to the south of the suture line. Highton (Reference Highton, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999) summarized the Galloway plutons as having 87Sr/86Sr ratios typically 0.705– 0.707, δ18O in the range 8 to 12‰, and εNd values suggestive of input from Silurian turbiditic sedimentary sequences.
3.a. Age, composition and structural relationships
Modern zircon-based age determinations are largely lacking for the members of the trans-suture suite of intrusions. Enough is known, however, of their age and structural relationships to place them in their Early Devonian tectonic context (Fig. 3). It has been proposed that large granitic intrusions are emplaced over 106 or even 107 years by the incremental assembly of small batches of magma in sheets or dykes (Coleman, Gray & Glazner, Reference Coleman, Gray and Glazner2004; Glazner et al. Reference Glazner, Bartley, Coleman and Taylor2004). The large Leinster intrusion is thought to be composed of sheets intruded over a substantial period of time (Reavy, Reference Reavy2001). On the other hand, concentrically zoned plutons such as Criffel (Section 3.a.2) are likely to have formed by the continuous or rapidly episodic injection of compositionally changing magma over a time span that lies within the precision errors typically associated with K–Ar and Rb–Sr age determinations. The majority of the trans-suture intrusions were emplaced in anchimetamorphic country rocks, and in our view their ‘cooling ages’ provide reasonable indication of their time of emplacement.
As has become common, though not universally accepted, practice in discussing the Newer Granites, we use the I and S nomenclature (Chappell & White, Reference Chappell and White1974). For the I-type we regard the origin as being either in refused basaltic or lamprophyric material, or in the fractional crystallization of such mantle-derived magmas.
3.a.1. Cairnsmore of Fleet Intrusion
This pluton was intruded into early Llandovery (Gala Group) strata in the Central Belt of the Southern Uplands, immediately south of the Orlock Bridge fault (Barnes, Phillips & Boland, Reference Barnes, Phillips and Boland1995; Figs 2, 3). A bulk fraction U–Pb zircon age of 396±6 Ma (Pidgeon & Aftalion, Reference Pidgeon and Aftalion1978) is within error of the Rb–Sr mineral–whole rock age of 392±2 Ma (Halliday, Stephens & Harmon, Reference Harmon and Halliday1980). These ages are reported by Barnes, Phillips & Boland (Reference Barnes, Phillips and Boland1995) to have been confirmed by a more recent unpublished U–Pb zircon age.
Emplacement post-dates the accretionary deformation of the Central Belt and initiation of the Moniaive shear zone. A fabric related to ductile reactivation of the shear zone wraps cordierite porphyroblasts in the aureole but is cut by the granite contact (Barnes, Phillips & Boland, Reference Barnes, Phillips and Boland1995; Stone, Kimbell & Henney, Reference Stone, Kimbell and Henney1997). If the reported age reliably reflects the time of emplacement, these relationships indicate that reactivation of the Moniaive shear zone was Acadian, and the Fleet pluton was emplaced syntectonically.
The Fleet two-mica pluton has S-type chemical and isotopic characteristics. A predominantly crustal origin is indicated by an initial 87Sr/86Sr ratio of 0.7062 to 0.7083 and an εNdt of −3.0 to −3.4. The highly evolved oxygen isotopes, with 18O around 11‰, indicate major sedimentary input (Halliday, Stephens & Harmon, Reference Halliday, Stephens and Harmon1980; Halliday, Reference Halliday1984; Stephens & Halliday, Reference Halliday1984). Stephens & Halliday (Reference Halliday1984) suggested that the source of the Fleet magmas was the same as that of the Lake District plutons with the implication that magma genesis took place after closure of the Iapetus Ocean. Derivation from Avalonian crust underthrust beneath the Southern Uplands in the footwall of the suture is also supported by Pb isotopes (Thirwall, Reference Thirwall1989) which match those found in Ordovician Skiddaw Group of the Lake District.
3.a.2. The Criffel pluton
The Criffel pluton was emplaced into Llandovery–Wenlock strata in the southern part of the Southern Uplands accretionary complex, stitching the tract-bounding fault between the Hawick and Riccarton groups (Fig. 2). Features such as enclave alignment and mineral foliations, and rotation of envelope fabrics into parallelism with the margin, indicate diapiric emplacement and ballooning (Stephens, Reference Stephens, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999).
An Rb–Sr age of 397±2 Ma (Halliday, Stephens & Harmon, Reference Halliday, Stephens and Harmon1980) is within error of the 395±9 Ma Rb–Sr age of the nearby Newmains lamprophyre dyke. This dyke has the same initial 87Sr/86Sr ratio (0.70514±5) as the outer mafic part of the pluton (Macdonald et al. Reference MacDonald, Rock, Rundle and Russell1986) and the relationship between coeval lamprophyric and granitic magmatism gives important insights into the petrogenesis of the Trans-Suture Suite, as is discussed in Sections 3 and 4.
The Criffel pluton exhibits strong concentric zoning from an early outer hornblendic granodiorite of I-type to an inner two-mica granite with marked S-type affinity (Stephens et al. Reference Stephens, Whitely, Thirwall and Halliday1985; Stephens, Reference Stephens1992). Isotopic ratios (Halliday, Stephens & Harmon, Reference Halliday, Stephens and Harmon1980; Halliday, Reference Halliday1984) vary from the outer zone inwards, 87Sr/86Sr from 0.7052 to 0.7073, εNd from −0.6 to −3.1 and δ18O from 8.5 to 11.9‰. Stephens et al. (Reference Stephens, Whitely, Thirwall and Halliday1985) demonstrated correlated behaviour of rare earth element abundances with isotopic variability and showed that the data can only be interpreted in terms of substantial and progressively increasing input of crustally derived anatectic magma towards the pluton interior.
Mafic inclusions are common in the outer part of the Criffel intrusion and it has been shown that these represent syn-plutonic injections of lamprophyric magma into the host granodioritic magma (Holden, Halliday & Stephens, Reference Holden, Halliday and Stephens1987; Holden et al. Reference Holden, Halliday and Stephens1991; Stephens, Holden & Henney, Reference Holden, Halliday, Stephens and Henney1991). The simultaneous existence of mantle and crustal melts demonstrates the importance of magmatic advection in the generation and ascent of anatectic granitic magma in the Criffel pluton.
3.a.3. The Cheviot intrusion
This intrusion is located close to the trace of the Iapetus suture (Fig. 2) and intrudes comagmatic lava flows of the Cheviot Volcanic Formation (Thirwall, Reference Thirwall1981). Thirwall (Reference Thirwall1988) reported a mean Rb–Sr biotite age of 395±2.9 Ma for three quartz monzonites from the intrusion, while the most reliable age obtained for the extrusive rocks is a biotite–apatite age of 395±3.8 Ma on a trachyte flow, confirming the contemporaneity of the intrusive and extrusive rocks. Thirwall regarded the Cheviot Volcanic Formation as chemically distinct from the Old Red Sandstone volcanic rocks of the Midland Valley which he related to subduction.
Harmon & Halliday (Reference Harmon and Halliday1980) attributed the low δ18O of 4.71–4.78 of the Cheviot granite to post-emplacement exchange with depleted meteoric groundwater. The initial 87Sr/86Sr of 0.7061 was presumed to have been unaffected and if so to limit the input of radiogenic Sr from crustal source. The εNdt of −4.2 (Halliday, Reference Halliday1984) is, however, somewhat more negative than for the S-type Fleet two-mica granite or the central part of the Criffel body.
3.a.4. Minor granites
Small and poorly known granitic bodies at Portencorkie in the Rhinns of Galloway and at Kirkmabreck near Creetown, were also included by Highton (Reference Highton, Stephenson, Bevins, Millward, Highton, Parsons, Stone and Wadsworth1999) in the Galloway Suite of intrusions.
3.a.5. The Skiddaw granite
The Skiddaw biotite granite is the most northerly of the Lake District granite intrusions and has a wide metamorphic aureole in Ordovician country rocks of the Skiddaw Group. It is a boitite granite with lesser amounts of muscovite. In the aureole, andalusite overgrows the S1 Acadian cleavage but is locally weakly wrapped by it and is further deformed by S2 (Soper & Roberts, Reference Soper and Roberts1971). The emplacement age is less well constrained geochronologically, but a K–Ar biotite age of 392±4 Ma (Shepherd et al. Reference Shepherd, Beckinsale, Rundle and Durham1976) and an Rb–Sr isochron of 399±8 (Rundle, Reference Rundle1992) are consistent with an ‘intra-Acadian’ age setting for the intrusion. There is a paucity of other isotopic data for the intrusion.
3.a.6. The Shap granite
This well-known K-feldspar megacryst granite was emplaced in Caradoc volcanic rocks of the Borrowdale Group and its aureole affects strata as young as Ludlow in the overlying Windermere Supergroup. Contact metamorphism post-dates the Acadian cleavage but cogenetic microgranite dykes are locally cleaved at their margins, indicating that the magmatism was also ‘inter-Acadian’ (Soper & Kneller, Reference Soper and Kneller1990).
The Shap intrusion was the subject of early Rb–Sr and K–Ar dating which yielded ages in the range 380–410 Ma with large precision errors (summarized by Brown, Miller & Soper, Reference Brown, Miller and Soper1964). A discordant bulk fraction U–Pb zircon age of 390±6 Ma was reported by Pidgeon & Aftalion (Reference Pidgeon and Aftalion1978). The best available estimate of the emplacement age of the intrusion seems to be the 394±3 Ma obtained by Wadge et al. (Reference Wadge, Gale, Beckinsale and Rundle1978) from whole rock–feldspar isochron and supported by K–Ar biotite age of 397±7 Ma.
The Shap granite has high δ18O of 11‰ (Harmon & Halliday, Reference Harmon and Halliday1980), moderately high 87Sr/86Sr of 0.707 (Wadge et al. Reference Wadge, Gale, Beckinsale and Rundle1978) and εNd of −2.0. These figures have been variously interpreted as being due to hydrothermal crustal fluids acting on a granite with I-type mineralogy (Halliday, Reference Halliday1984) and as obvious evidence of strong sedimentary input (Harmon & Halliday, Reference Harmon and Halliday1980).
The intrusion contains abundant megacryst-bearing microdioritic inclusions which are distinct from the country rock xenoliths that are also sparsely present in the granite. The microdiorite clots are thought to result from injection of contemporaneous basic magma (Grantham, Reference Grantham1928) which Harker, as long ago as Reference Harker1909, suggested was related to the regional K-lamprophyre dyke swarm. Unfortunately, subsequent studies have not pursued this important aspect, concentrating on the feldspar megacrysts (e.g. Cox et al. Reference Cox, Dempster, Bell and Rogers1996; Lee & Parsons, Reference Lee and Parsons1997).
3.a.7. The Crummock intrusion
The presence of a subsurface granitic intrusion in the western Lake District is inferred from an extensive zone of bleaching and tourmaline mineralization in the Skiddaw Group (Cooper et al. Reference Cooper, Lee, Fortey, Cooper and Rundle1988). The alteration is later than the main deformation but pre-dates reverse displacement on the associated Causey Pike fault, so the intrusion is thought to be ‘late Acadian’. Cooper et al. (Reference Cooper, Lee, Fortey, Cooper and Rundle1988) reported that 15 whole-rock samples gave a Rb–Sr regression line equivalent to 395±12 Ma. The quoted isochron age of 401±3 Ma obtained by progressive elimination of outlying points is more precise but not necessarily a more accurate estimate of the emplacement age.
3.a.8. The Weardale intrusion
The presence of a sub-Carboniferous granitic pluton in the northern Pennines was detected geophysically and proved by drilling (Rookhope borehole: Dunham et al. Reference Dunham, Dunham, Hodger and Johnson1965). It is compositionally similar to the Shap and Skiddaw intrusions (Millward, Reference Millward2002), but unlike them its upper part has shallow-dipping gneissose foliation (Dunham, Reference Dunham1990). Fitch & Miller (Reference Fitch and Miller1965) reported a K–Ar muscovite age for the Weardale intrusion of 396±6 Ma. A Rb–Sr age of 410±10 Ma was obtained by Holland & Lambert (Reference Holland, Lambert and Johnson1970) and recalculated by Dunham (Reference Dunham1974) as 394±34 Ma. A reliable emplacement age would show whether the Weardale intrusion should be included in the Trans-Suture Suite, perhaps as an early (pre-Acadian) member. It may be noted that the other large subsurface Pennine granite, Wensleydale, which gave a Rb–Sr age of 410±10 Ma (Dunham, Reference Dunham1974), is now thought to be Ordovician on the grounds of its similarity to the sub-volcanic Lake District intrusions of Eskdale and Ennerdale (Millward, Reference Millward2002).
3.a.9. Isle of Man granites
The Dhoon granodiorite and Foxdale and Oatlands granites have small outcrops immediately south of the trace of the suture (Fig. 2) and are intruded into Ordovician slates of the Manx Group. Fabric studies on their aureoles indicate that the Dhoon body was emplaced during D1 and Foxdale after D2 in the local deformation sequence, which is presumed to be Acadian (Power & Barnes, Reference Power, Barnes, Woodcock, Quirk, Fitches and Barnes1999).
3.a.10. Eastern Irish granites
The Leinster granite is the largest Caledonian granite exposed in the British Isles. It is a peraluminous two-mica granite believed to have been derived from partial melting of an immature, dominantly volcaniclastic protolith (Elsdon & Kennan, Reference Elsdon, Kennan, Harris, Holland and Leake1979). U–Pb and monazite geochronology dates the intrusion at 405±2 Ma (O'Connor, Aftalion & Kennan, Reference O'Connor, Aftalion and Kennan1989). Structural studies of the batholith (M. O'Mahony, unpub. Ph.D. thesis, NUI, Cork, Ireland, 2001) have shown that it was emplaced into low-grade Lower Palaeozoic metasediments in an active extensional crustal lineament by process of amalgamation of multiple sheets of magma. Petrographic studies (Grogan & Reavy, Reference Grogan and Reavy2002) have shown magma mixing between successive sheets. Appinites and lamprophyres are locally abundant along the northwestern margin of the pluton and are also associated with a major shear zone, the East Carlow deformation zone, that bounds its eastern margin (McConnell et al. Reference McConnell, Philcox, Sleeman, Stanley, Flegg, Daly and Warren1994). The penetrative fabric in the East Carlow deformation zone deforms the regional foliation, post-dates the emplacement of the appinites and lamprophyres, and is contemporaneous with contact metamorphism associated with the Leinster granite (McArdle & Kennedy, Reference McArdle and Kennedy1987).
The Newry igneous complex intruded and contact metamorphosed low-grade metasediments of the Longford-Down massif. The central part of the intrusion has given a U–Pb zircon age of 423±7 Ma (Meighan et al. Reference Meighan, Hamilton, Gamble, Ellam and Cooper2003). The geologically older northeastern part of the complex has a concordant titanite age of 410.4±1.3 that has been interpreted as a cooling age (Meighan et al. Reference Meighan, Hamilton, Gamble, Ellam and Cooper2003). The Newry complex probably pre-dates the final closure of the Iapetus ocean and we do not consider it as part of the Trans-Suture Suite.
3.b. Tectonic setting of the Trans-Suture Suite granites
In summary, the isotopic ages and structural relationships of the Trans-Suture Suite of granites show that they were emplaced at relatively shallow (anchimetamorphic) crustal levels in both the hanging- and footwalls of the Iapetus suture during Acadian deformation. This occurred late in Early Devonian time, within the period 390–400 Ma. The deformation was transpressive and no doubt episodic; stress release by seismic slip may have facilitated the ascent of granite magma. However, although the Trans-Suture Suite granites were ‘synorogenic’ as regards their final emplacement, their association with the Early Devonian lamprophyre suite, explored below, suggests that their generation and ascent were associated with the preceding period of transtension.
4. Lamprophyre dykes
Mica-phyric and hornblende-phyric lamprophyre dykes are present throughout Scotland and extend into northern England (e.g. Rock et al. Reference Rock, Gaskarth, Henney and Shand1988; Shand et al. Reference Shand, Gaskarth, Thirwall and Rock1994; Canning et al. Reference Canning, Henney, Morrison and Gaskarth1996, Reference Canning, Henney, Morrison, Van Calsteren, Gaskarth and Swarbrick1998). In the region of the Trans-Suture Suite of plutons the main concentration of dykes is in a 10 km wide zone extending from the Central Belt of the Southern Uplands into Eastern Ireland (Rock, Cooper & Gaskarth, Reference MacDonald, Rock, Rundle and Russell1986; Vaughan, Reference Vaughan1996). The zone trends significantly anticlockwise to the regional strike (Fig. 2) and within it, upper crustal extension averages 1% and locally as much as 6% (Barnes, Rock & Gaskarth, Reference Barnes, Rock and Gaskarth1986). Regionally, mica lamprophyres dominate the zone, but close to the Criffel pluton, hornblende lamprophyres are common.
Dated examples from the Lake District and Southern Uplands (Nixon, Rex & Condliffe, Reference Nixon, Rex and Condliffe1984; Rock, Gaskarth & Rundle, Reference Rock, Gaskarth and Rundle1986) have given Early Devonian K–Ar ages, but some undated early dykes in the Southern Uplands overlapped late stages of the accretionary deformation and are likely to be of latest Silurian age. However, as outlined above, their main development was associated with the Early Devonian period of sinistral transtension that followed Iapetus closure.
The biotite lamprophyres include examples with high Mg numbers suggesting derivation from primitive mantle melts unaffected by fractionation or assimilation. The origin of such volatile-rich potassic magmas is thought to be in small degrees of melting of metasomatized sub-continental lithospheric mantle (e.g. Harry & Leeman, Reference Harry and Leeman1995; Canning et al. Reference Canning, Henney, Morrison and Gaskarth1996). In an environment of extensional tectonics and lithospheric thinning, melting may be triggered by decompression and the input of asthenospheric heat. Also, in the Iapetus suture zone, decompression melting may be attributed to the influence of lithosphere-penetrating shear zones (Vaughan, Reference Vaughan1996), which both concentrated transtensional strain in early Devonian time and provided channels for H2O and CO2 metasomatism.
A genetic link between lamprophyric/appinitic magmatism and the Scottish Newer Granites has long been considered. Atherton & Ghani (Reference Atherton and Ghani2002) specifically suggested remelting of a lamprophyric crustal underplate as a source of I-type magmas in the Argyll Suite of the Newer Granites. However, fractionation of primitive lamprophyric/appinitic magma cannot alone have generated the huge volume of Newer Granite plutons in the Highlands; it is necessary to invoke substantial assimilation of crustal material (e.g. Macdonald et al. Reference MacDonald, Rock, Rundle and Russell1986; Rock & Hunter, Reference Rock and Hunter1987; Fowler, Reference Fowler1988; Fowler & Henney, Reference Fowler and Henney1996; Canning et al. Reference Canning, Henney, Morrison and Gaskarth1996; Fowler et al. Reference Fowler, Henney, Derbyshire and Greenwood2001). In the case of the Trans-Suture Suite intrusions, the volumetric importance of S-type magma, to the point of being the major component of the Fleet pluton, demonstrates the existence of independent mantle (I-type) and crustal (S-type) magmas, and mingling and mixing of these is evident in many members of the Trans-Suture Suite.
A mantle input to the Criffel pluton is seen in the mafic inclusions which are common in the outer parts of the intrusion. These have been interpreted by Holden, Halliday & Stephens (Reference Holden, Halliday and Stephens1987) and Stephens, Holden & Henney (Reference Holden, Halliday, Stephens and Henney1991) as syn-plutonic injections of mantle-derived lamprophyric magma; they are evidence of the simultaneous existence of mantle and crustal melts and, by inference, for the heat source responsible for crustal melting and facilitating the rise of S-type magmas. Similar, but less well-researched, relationships between the Shap granite and its mafic inclusions, point to the same dual magmatic source.
A mechanism is thus required for the simultaneous production of mantle and crustal melts during the Early Devonian period of transtension. While decompression melting of lithospheric mantle is a well-established mechanism, under realistic boundary conditions, significant melting of the lower crust is not (Ryan & Soper, Reference Ryan and Soper2001). It is not possible to quantify the relative importance of fractionation of lamprophyric magma versus fusion of lamprophyric underplate in the genesis of the I-type component of the Trans-Suture Suite, but as explored below, it is our proposal that advection of heat into the lower crust by lamprophyric magmas (Section 6) was responsible for crustal melting and the I-type transitional to S-type character of the Trans-Suture Suite plutons.
5. End-Caledonian lithospheric domains
A starting point is required for thermal modelling to investigate the proposed petrogenetic relationship between Early Devonian transtension, lamprophyric magmatism and generation of the Trans-Suture Suite granites. Figure 4 is a lithospheric cross-section of Scotland and northern England that attempts to depict mantle domains and their thermal structure at the end of Caledonian convergence, say 420 Ma. The section is notionally along the UK Geotraverse North line (Pharaoh et al. Reference Pharaoh, Morris, Long and Ryan1996). It represents a two-dimensional time slice through a complex convergence system and is necessarily speculative. While the upper part of such a section can be based on surface geology and geophysical data, the configuration of the lithospheric mantle, which is critical to the modelling, has to be inferred from Caledonian convergence history, itself inferential.
Lithospheric cross-section drawn approximately along the line of the UK Geotraverse North line, depicting mantle domains and their thermal structure at c. 420 Ma. The geometry of the Avalonian lithosphere is derived from assuming that the ridge was destroyed around 450 Ma (Woodcock, Reference Woodcock, Woodcock and Strachan2000). The southern extent of Iapetan lithosphere is unknown and may be as far north as the Great Glen Fault or as far south as the Southern Uplands Fault. The base of the lithosphere is taken as 1300°C. The local depression of the 300°C and 700°C geotherms is drawn to reflect the cessation of subduction at this time.
5.a. Caledonian convergence history
The later stages of the Caledonian orogeny are attributed to convergence between three plates: Laurentia, Baltica and Avalonia (Fig. 5). The timing and geometry of Baltica–Avalonia interaction is debated but not of critical importance here. It is widely agreed that in the British–Irish sector, Iapetus closure involved subduction to the south (in the present frame) beneath Eastern Avalonia until late Ordovician time and northward beneath Laurentia until late Silurian time.
Tectonic diagram showing the likely positions of the three plates: Avalonia, Baltica and Laurentia, during late Silurian–early Devonian time.
Generally southward subduction in the Ordovician beneath Eastern Avalonia (the Borrowdale subduction zone) led to the development of arc-marginal basin pairs in the Lake District and Wales (Kokelaar, Reference Kokelaar1988) and Leinster (McConnell, Reference McConnell2000), culminating in a burst of volcanicity in the Caradoc at about 450 Ma (Millward & Evans, Reference Millward and Evans2003). This and the subsequent volcanic shut-down have been attributed by Woodcock (Reference Woodcock, Woodcock and Strachan2000, figs 10.5, 10.9) to subduction of an Iapetan ridge-transform system, probably followed by slab break-off. The Avalonian lithospheric mantle lay in the hangingwall of this system and, as such, would have been fluxed by volatiles during these events. The northwards extent of the Avalonian lithospheric mantle is unknown, but it is assumed to extend at least as far as the Midland Valley.
Lithospheric convergence geometry during the ensuing closure of Iapetus in the Silurian period presents a four-dimensional problem in which the main uncertainty is whether Baltica and Avalonia had different displacement vectors with respect to Laurentia, or whether they had already combined into a single plate during late Ordovician collision (e.g. Trench & Torsvik, Reference Trench and Torsvik1992). A resolution of the problem is not essential in the present context, because we are concerned principally with Avalonia–Laurentia interaction in the British–Irish sector. Here it is widely agreed that northward convergence of Iapetan lithosphere with Laurentia was partitioned into NW-dipping subduction system and sinistral slip on NE–SW (Caledonoid) terrane-bounding faults in the hangingwall, principally the Great Glen and Highland Border.
5.b. Seismic evidence
The Iapetus suture zone has been imaged on BIRPS deep seismic reflection profiles and can be traced for some 900 km from the Atlantic continental shelf west of Ireland to the North Sea (Klemperer, Ryan & Snyder, Reference Klemperer, Ryan and Snyder1991 and references therein). It is associated with a N-dipping zone of middle and lower crustal reflectivity, consistent with northward underthrusting of Avalonian crust beneath the Laurentian margin during final closure. The suture zone is perhaps best imaged on the NEC line located close offshore of eastern Britain (Freeman, Klemperer & Hobbs, Reference Freeman, Klemperer and Hobbs1988). A boundary (IN on Fig. 4) can be recognized between reflective, presumably Avalonian crust, to the south and unreflective mid-crust to the north. On a depth-converted section the boundary dips at about 20° to the north and projects up-section to intersect the base of the Upper Palaeozoic cover on the Solway Line, the geologically defined trace of the Iapetus suture (Soper et al. Reference Soper, Strachan, Holdsworth, Gayer and Greiling1992). A similar reflectivity boundary (IW) is seen on the WINCH line to the west in the North Channel, which had previously been identified by Hall et al. (Reference Hall, Brown, Mathews and Warnėr1984) as marking the suture. We have interpolated between IW and IN to fix the crustal position of the suture on Figure 5.
The deeper trajectory of the suture has not been imaged directly, but there are grounds to suppose that in the present-day lithosphere it flattens northwards. On the NEC line, the IN boundary appears to merge with a zone of strong subhorizontal reflectivity in the lower crust which includes two bright, gently N-dipping reflections (P1 and P2, Fig. 4) that are coincident with the Moho at 30 km depth. These have been interpreted as Iapetus crust (Freeman, Klemperer & Hobbs, Reference Freeman, Klemperer and Hobbs1988) or imbricated basement and sedimentary cover from the continent–ocean transition at the leading edge of Avalonia (Soper et al. Reference Soper, Strachan, Holdsworth, Gayer and Greiling1992). We adopt the latter interpretation in Figure 4 and assume, therefore, that Avalonian mantle must extend at least as far north as the IN, P1 and P2 reflectors. Avalonian sedimentary rocks could be the source of the high 207Pb/204Pb ratios of certain Southern Uplands granites which, as mentioned above, Thirlwall (1989) matched with the Ordovician Skiddaw Group of northern England.
On the WIRE profiles, offshore of western Ireland (Klemperer, Ryan & Snyder, Reference Klemperer, Ryan and Snyder1991), strong crustal reflections in the suture zone dip north at about 30°, and further north there are patches of gently dipping sub-Moho reflectivity which presumably relate to subhorizontal mantle fabric (Klemperer, Ryan & Snyder, Reference Klemperer, Ryan and Snyder1991, their Fig. 3). However, flat subduction could not have generated the Llandovery–Wenlock magmatism of northern Connemara (Menuge, Williams & O'Connor, Reference Menuge, Williams and O'Connor1995), now some 100 km north of the suture. Even if the Silurian arc–trench gap was significantly greater, a relatively steep subduction zone is required.
Our inference is that normally inclined subduction in early Silurian time became shallower as progressively younger lithosphere entered the system, and convergence slowed to zero at 420 Ma with modest collisional deformation due, in part, to the buoyancy effect of the Avalonian margin. The dip of the subduction zone at 420 Ma adopted in Figure 4 assumes that further flattening of the uppermost mantle fabric occurred during subsequent episodes of lithospheric extension, principally in Early Devonian and Permo-Triassic time.
6. Thermal modelling
Our proposal is that the Trans-Suture Suite granites were produced by melting in the lower part of the Avalonian crust, where it forms the footwall of the Iapetus suture, by the advection of heat from rising lamprophyric magma. The lamprophyre melts were generated by decompression melting of deep, enriched and hydrated Avalonian lithospheric mantle during Early Devonian transtension. We use two-dimensional transient finite element models to test whether this proposal is physically realistic. The modelling is performed in three stages: thermal relaxation of the initial 420 Ma template to 400 Ma; followed by lithospheric extension to produce a pull-apart basin 21 km wide to determine the volume of associated lamprophyric melt produced under realistic strain rates; and then the emplacement of this melt into the lower crust to explore the generation of granite melt. The duration of the rifting events and strain rates were varied to produce the required Beta factor of 1.5 (Table 3), a reasonable value for the overlying Devonian basins (Soper & Woodcock, Reference Soper and Woodcock2003). We follow the methodology outlined in Ryan & Soper (Reference Ryan and Soper2001) with some amendments, discussed in this section. Assumed values of the input parameters used in the modelling are listed in Table 1.
Parameters used in the modelling
List of assumptions used in this analysis. All densities are given in values at NTP. The asthenospheric densities were adjusted so that the basins initiated at or near sea-level; these basins were then sediment loaded to produce a maximum value for the rift sediment deposited.
Figure 6a represents the initial lithospheric thermal structure at 420 Ma derived from Figure 4, and Figure 6b shows the thermal structure after static relaxation for 20 Ma, following the method of Ryan & Dewey (Reference Ryan and Dewey1997). This provides the starting conditions for subsequent finite element models of extension, lamprophyre intrusion and consequent generation of granite magma. Models were run using a crust with 30 km, 35 km and 40 km thickness; these bracket the likely values estimated by adding the eroded material onto the present crustal thickness (see Soper & Woodcock, Reference Soper and Woodcock2003). The models predict that the Moho temperature at the centre of the site of the rift will rise due to thermal relaxation and radiogenic heating between 420 and 400 Ma from 424°C to 481°C or from 494°C to 551°C and from 499°C to 577°C for continental crust of 30 km, 35 km and 40 km thickness, respectively. This is an important prediction because the temperature of the lower crust significantly affects the amount of granitic melt that can be generated by a given aliquot of injected mafic magma (see, for example, Annen & Sparks, Reference Annen and Sparks2002). If the crust had remained at the refrigerated temperatures associated with an accretionary prism, then it would be hard to generate any substantial volume of granite by our chosen mechanism.
(a) Assumed initial thermal structure of the lithosphere 110 km either side of the Iapetus suture at 420 Ma based upon Figure 4. (b) Modelled thermal structure of the lithosphere at 400 Ma at the start of rifting. These diagrams were prepared using GMT software (Wessel & Smith, Reference Wessel and Smith1991).
Lamprophyric magma is regarded as resulting from a low-fraction melting of metasomatized subcontinental lithospheric mantle or of metasomatic veins within such mantle (e.g. McKenzie, Reference McKenzie1989); they are volatile-rich and have high abundances of light rare earth and large ion lithophile elements. Such magmas are produced by low-fraction melting of hydrous peridotite (see, for example, Hirose & Kawamoto, Reference Hirose and Kawamoto1995). The concept of decompression melting as used to produce lamprophyre magma in our modelling requires that the lower Avalonian lithosphere was enriched and hydrated. The Avalonian subcontinental lithosphere was above the southerly Iapetan subduction zone during the Ordovician and may have been enriched at that time. Also, as proposed by Vaughan (Reference Vaughan1996), deep-seated transtensional shear zones (reactivated transform faults?) are thought to have facilitated decompression and movement of volatiles in the lower lithosphere. The main concentration of dykes lies anticlockwise to the regional structural trend but clockwise to the dyke maximum (Fig. 2), suggesting that the controlling shear zone was neither inherited from crustal structure nor initiated under the Early Devonian transtensional regime but existed as a zone of weakness in the deep lithosphere which was reactivated in transtension.
A further problem is that the lamprophyre solidus has not been fully determined experimentally. However, experimental studies (Wyllie, Reference Wyllie1995; Hirose & Kawamoto, Reference Hirose and Kawamoto1995) show that solidus temperatures for peridotite with only 2.5 to 3.0% of water are significantly lowered (Fig. 7). A hydrated basaltic melt results as the water is partitioned into the initial melt increments and the source peridotite then becomes dehydrated. It is difficult to select an exact solidus temperature to model, as this depends on the degree of water saturation. However, Hirose & Kawamoto (Reference Hirose and Kawamoto1995) showed that 5% melt fraction will be generated at 1100°C and 1 GPa. Thus, for wet basaltic magma to ascend to the Moho, about 1.0 GPa, it must have a temperature of at least 1100°C. A significant amount of the lower lithospheric mantle is at or above 1100°C; melt should, therefore, originate as soon as hydration takes place, and subsequent batches of melt may be produced in regions below the solidus in response to other tectonic events.
Solidus curve(s) for wet and dry mantle peridotites from 0–120 km (4.0 GPa). The arrow shows the interrupted ascent path of magma produced at 1080°C at 4 GPa. It freezes again at 2 GPa. However, if this magma were produced at 1.3 GPa it could ascend to a Moho at 1 GPa (dashed arrow).
An alternative approach is to use the wet peridotite solidus (Wyllie, Reference Wyllie and Perchuk1991, Reference Wyllie1995) (Fig. 7). Again we assume that 5% melt is produced when the temperature exceeds that of the solidus for a given pressure. However, this solidus has both positive slope at low pressure and negative slope at high pressures with maximum temperature of 1100°C at about 1.5 GPa (Fig. 7). Magma produced below this point will freeze on ascent unless it has temperature at or in excess of 1100°C. In our model, it is then registered as being trapped and available for re-melting if suitable conditions arise. Both solidi shown in Figure 7 produce similar amounts of lamprophyric melt for the conditions assumed. This is because during stretching, with Beta factors of about 1.5, temperatures of 1100°C are never exceeded at pressures of less than 1.4 GPa. Therefore, all melt produced at >1.4 GPa above 1100°C can ascend. If temperatures of 1050°C were exceeded between 1.0 GPa and 1.4 GPa then the models would differ (Fig. 7).
The next step is to assume that this region is then subjected to transtension developing pull-aparts of 21 km width. We present models for one such rift that initiates at 100 km into the model space. The predicted crustal temperature structures across the model are fairly uniform by 400 Ma, but the mantle has not fully thermally relaxed. Selecting 100 km gives a conservative estimate as this area has the thinnest lithosphere and will produce the least lamprophyre on rifting. In modern transtension zones, strain rates can approach 10−14 s−1 (Dewey, Reference Dewey2002). Consequently, we assume two relatively high strain rates of 6.6×10−15 s−1 and 3.3×10−15 s−1 orthogonal to the zone's boundaries in addition to the more ‘normal’ rate of 10−15 s−1. We then run two-dimensional finite element models following the method of Ryan & Soper (Reference Ryan and Soper2001) to calculate the amount of lamprophyric melt generated during rifting for each strain rate (Table 2). The area of lamprophyric melt is calculated by re-evaluating the temperature of the nodes bounding each element between 1 GPa and 4 GPa after each time step. If the node temperature exceeds the selected solidus at that pressure, then that element is assumed to produce 5% melt at that temperature and become refractory; the melt is moved up instantaneously until it freezes (see Fig. 7) or reaches the Moho. Such a model assumes that there is no heat exchange between the melt and the surrounding mantle rocks.
Two models for the hydration of the mantle were used. Firstly, the mantle was hydrated at 400 Ma, at the onset of transtensional rifting and development of deep penetrating shear zones. Secondly, the mantle was hydrated during final closure of the ocean at 420 Ma. The appropriate area of lamprophyric melt generated at given times (Table 3, Fig. 8) is then emplaced as 20 km long sills, 4 km above the base of the crust (Fig. 9a). Lamprophyres generated outside these time windows were assumed to have ascended to higher levels, in other words, they were not emplaced in the lower crust and did not contribute to the generation of granitic magma. However, they may have assisted in the ascent of the magma by increasing heat flow. We identified six phases of intrusion for all models in which the locus of intrusion was assumed to move downwards with time. The nodes which accommodated the lamprophyre ‘sill’ had their vertical spacing increased and the intervening elements received the appropriate amount of advected heat at given time intervals (Table 2).
Lamprophyre produced with time by rift models
Thicknesses and temperatures of lamprophyric magma emplaced as 20 km wide sills into the rift zone at the nodes equivalent to 4 km above the Moho before rifting. The thickness decreases by 1/βt and the width increases by βt as rifting progresses. βt is the beta factor at a given time. Each successive package was emplaced below the previous one.
Areas of lamprophyric melt generated and able to ascend to the Moho with time, assuming 5% melting of lithospheric mantle. Melt model assumes the wet peridotite solidus of Wyllie (Reference Wyllie and Perchuk1991, Reference Wyllie1995). Melt areas are shown in a logarithmic scale. (a) Hydration of the mantle at 420 Ma and a strain rate of 3.3×10−15 s−1; most melt is generated during rifting over a period of some 2.0 Ma after the onset of rifting. (b) Hydration of the mantle at 420 Ma and a strain rate of 10−15 s−1; most melt is generated during rifting over a period of some 7.0 Ma after the onset of rifting. (c) Hydration of the mantle at 400 Ma and a strain rate of 6.6×10−15 s−1; note that the majority of melt is generated during the first time step after the onset of rifting, the time of hydration.
Results of melting models for 21 km wide basin
Results of melting models used in this analysis based upon on the method of Ryan & Soper (Reference Ryan and Soper2001), but using variable time steps to estimate the likely maximum melt volume. The granite models were constructed using the method of Zen (Reference Zen1995) and also based upon data from Johannes & Holtz (Reference Johannes, Holtz, Ashworth and Brown1990).
(a) Thermal structure of the lithosphere 0.025 Ma after emplacement of a sill of lamprophyre 20 km long and 2.9 km thick at about 1200°C at a position 4 km above the Moho. The model uses a strain rate of 6.6×10−15 s−1 affecting a pull-apart of 21 km initial width and an initial crustal thickness of 35 km (see Table 2 for more details). (b) Contours showing total proportion of melt formed using a biotite melting model 2.5 Ma after the start of rifting at a strain rate of 6.6×10−15 s−1. (c) Contours showing proportion of melt formed using a muscovite melting model, 2.5 Ma after the start of rifting at a strain rate of 6.6×10−15 s−1.
The crust is assumed to contain considerable amounts of hydrated juvenile material and both muscovite- and biotite-dominated melting models (see Figs 9b, c, 10b) are evaluated (Zen, Reference Zen1995; Ryan & Soper, Reference Ryan and Soper2001). However, tests showed that the area of granitic melt produced by a given model was affected by the size of the time step used. If a time step is too long, a crustal node may pass into and out of the melt field during this step and no melt will be recorded at the end of the step. This effect is illustrated in Figure 10a, which shows a speculative temperature versus time plot for a crustal node adjacent to a lamprophyre intrusion. To record the maximum possible melt, the time step should coincide with the maximum temperature for a given node. For example: time steps of 1.5, 0.5 or 0.25 would record the maximum possible melt of 35% at time 1.5, while steps of 1 or 2 would record maximum melt of 15% at time 2, but a step of 3 would record no melt (Fig. 10a). A plot of maximum melt produced versus time step is given in Figure 10c. There are several hundred nodes that melt in one step, therefore, it would be expected that such behaviour would be approximately linear over a range of time steps, excluding the very short and the very long. To allow for this effect, a range of time steps (0.05 Ma to 0.00625 Ma) was used and the area of granite magma produced in an infinitely small time step was estimated using simple linear extrapolation (Fig. 10d). The linear fit in Figure 10c slightly overestimates the maximum melt for this node. It is, therefore, assumed that the amount of melt produced in any one set of models lies between the value for the minimum time step (0.00625 Ma) and the value extrapolated to a zero-duration time step. All twelve models show strongly linear relationships with correlation coefficients approaching −1.00 and little difference between the area of melt modelled for time steps of 0.00625 Ma and 0.0 Ma.
(a) Hypothetical thermal evolution of a node adjacent to a lamprophyre intrusion. The time axis is linear, but with arbitrary values. (b) Melt fraction curves for a muscovite-dominated melt and a biotite-dominated melt system. (c) Plot of maximum melt yielded from the region adjacent to the node shown in (a) plotted against the size of the time step; the time interval is in arbitrary units. (d) Plots of area of melt generated versus size of time step (0.05–0.00625 Ma) for rifts with: a strain rate of 6.6×10−15 s−1 in 35 km crust for muscovite and biotite melt models; a strain rate of 10−15 s−1 in 30 km crust for muscovite and biotite melt models. Linear interpolation lines are plotted for each set of models and results are given in Table 3.
In a muscovite-dominated melting system the area of granite melt produced ranged from 12.5 km2 to 30.0 km2 for hydration at 400 Ma, and 1.3 km2 to 5.1 km2 for hydration at 420 Ma. In a biotite-dominated system it ranged from 4.2 km2 to 9.4 km2 for a system hydrated at 400 Ma and from 1.3 km2 to 0.4 km2 for one hydrated at 420 Ma. Although similar amounts of lamprophyre were emplaced for models with strain rates 6.6×10−15 s−1 and 3.3×10−15 s−1, the former produced about 30% more melt. This was because the longer time interval between sill emplacements for the lower strain rate model allowed the heat to dissipate (Table 2), consequently fewer nodes reached the minimum melting temperature for granite (675°C). This result mirrors that of the one-dimensional models of Annen & Sparks (Reference Annen and Sparks2002), where an increased frequency of emplacement promotes greater crustal melting. The runs with 40 km Moho produced about twice the amount of melt as those with 30 km Moho, where other conditions were equal. This reflects the higher temperature at the Moho with thicker crust (577°C; see above) and emphasizes the importance of the 20 Ma thermal relaxation stage. Had this period been shorter, we would predict that correspondingly less granite magma would have been produced.
The total volume of granite produced by such models depends upon the third dimension of the rift. As pull-aparts typically have length to breadth ratio between 2.1 to 5.1 (Aydin & Nur, Reference Aydin and Nur1982), the maximum volume of melt produced within a 20 km wide rift would be up to 3000 km3 for a strain rate of 6.6×10−15 s−1 and hydrated underlying mantle. If the magma ascended vertically, this would form a pluton of approximately 20 km width by 100 km length and 1.5 km thickness. Other geometries are possible if lateral transport of the granitic magma occurs.
The model used in this analysis is capable of taking a large range of parameters into account that may well vary the final amount of melt produced. Ryan & Soper (Reference Ryan and Soper2001) showed that dykes near the margin of rift can produce up to 30% more melt than sills at the depocentre. Annen & Sparks (Reference Annen and Sparks2002) showed that both the rate of magma emplacement and the nature of the lower crust can influence the melt volume. However, it is not the purpose of this analysis to investigate the subtle influences that various tectonic and geological factors can have on the precise volume of granitic melt produced. Rather, it is our intention to demonstrate that under certain conditions the emplacement of lamprophyres into the lower crust can produce sufficient granite volumes to account for the Trans-Suture Suite.
These models suggest that the whole of the relatively small elongated Crummock Water intrusion of the Lake District (Fig. 2), with volume of perhaps 102 km3 could be produced at strain rates equal to or in excess of 10−15 s−1 and crust of 30 km or more thickness, regardless of when hydration took place. ‘I’-type components from fractionation or mixing of lamprophyric magma would create an S+I-type complex which would promote the rise of the granitic melts to higher crustal levels. The large isolated plutons of the Southern Uplands such as Criffel and Fleet, which have volumes of the order of 103 km3, can only be produced in muscovite-dominated systems with a strain rate ≥10−15 s−1 and a crust ≥30 km in thickness. The 104 km3 Leinster pluton could only be produced by this model with a strain rate of 6.6×10−15 s−1, a crustal thickness of ≥40 km and a rift of 120 km length with lateral migration of magma over 10 km or more.
The Trans-Suture Suite plutons typically comprise both S- and I-type components but it has not been possible, even in the best-studied example of Criffel, to quantify the relative proportions of these inputs. If the crustal component was only 50%, then according to our model it must represent granitic melt gathered from considerable length of shear zone, unless the strain rate was very high and assuming that the mantle was hydrated at the time of rifting. The same applies to the differentiated and contaminated lamprophyric melt in the mafic outer parts of the pluton. In such situation our models predict a ratio of muscovite- to biotite-dominated melt of about 3:1 and volume of excess lamprophyre of the order of 1–4 km3. Emplacement of the Trans-Suture Suite plutons thus involved substantial lateral magmatic flow and/or very high strain rates within the putative transtensional shear zones.
The model produces the most granite if hydration was coeval with transtension and penetrated the lithosphere along pre-existing lines of weakness. It is important to note that at the onset of granite formation the mantle was hydrated, a fact supported by the presence of the lamprophyre swarms. We fully accept that the lack of an experimentally determined lamprophyre solidus to parameterize our model places constraints on the accuracy of our models. However, in a test run using the dry peridotite solidus for 35 km thick crust at a strain rate of 6.6×10−15 s−1, a sill of basaltic melt of only 0.0024 km thickness was produced and no granite magma. This supports our contention that a melting regime similar to those modelled is required to produce the observed volumes of lamprophyric magma during transtension.
7. Conclusions
A group of granitic intrusions occurring to the south of the Moniaive shear zone in the south of Scotland, in northern England and in Ireland, and which span the trace of the Iapetus suture is recognized as the Trans-Suture Suite. The ages of these intrusions are in the range 400–390 Ma, significantly younger than the Newer Granite intrusions in the Caledonide orthotectonic zone to the north. In the orthotectonic zone a range of rock types is represented, mainly of I-type, whereas in the Trans-Suture Suite, members with significant S-type characteristics are more conspicuous. The origins of the intrusions occurring in both the footwall and the hangingwall of the Iapetus suture have been particularly problematical. It is concluded that the genesis of these granitic intrusions can be explained in the context of Siluro-Devonian plate tectonics and the recognition of orogen-wide sinistral transtension in the early Devonian period. This episode of extensional tectonics is linked to the deposition of Old Red Sandstone sediments in coalescing basins across much of Britain south of the Scottish Highlands. Significantly, the granitic intrusions of the Trans-Suture Suite are accompanied by an intense suite of lamprophyre dykes, the origin of which was in extension, decompression and melting of subducted, enriched and hydrated Avalonian lithospheric mantle. Following the ‘soft’ collision of Avalonia with Laurentia there was period of thermal relaxation which removed subduction refrigeration. Lamprophyre melting triggered by Devonian transtension led to the advection of heat into the lower crust of the Trans-Suture Suite zone which comprised juvenile volcanogenic material and generated the Trans-Suture Suite. Numerical models suggest that if hydration occurred during transtension then large I+S-type granitic bodies could be generated. If hydration pre-dated transtension then only small granitic bodies would be produced, unless the zone of lamprophyre generation extends beyond the rift zone. The volumes of lamprophyre and granite are consistent with volatile fluxing of the Avalonian lithosphere above the Borrowdale subduction zone during Ordovician time and hydration along deep-seated transtensional shear zones during the Early Devonian period. A similar multiphase model, which required both subduction and post-collision processes, was proposed by MacDonald et al. (Reference MacDonald, Thorpe, Gaskarth and Grindrod1985) for the lamprophyres of NW England.
The generation of I+S granite suites within a transtensive pull-apart zone with a high strain rate will assist emplacement at higher levels. For example, in the model for a 35 km thick crust at a strain rate of 6.6×10−15 s−1, the upper surface of the zone of granite generation was transported tectonically upwards from about 31 km to about 23 km by crustal thinning beneath the sedimentary basin. The higher the strain rate, the faster this process and the more likely it is to transport a magma body before final consolidation.
The model may be generalized to account for post-orogenic S-type granites in other ‘soft’ collision zones. However, the amount of I-type melt produced by this model was limited and thus some other mechanism must be sought for the generation of the northern I-type suite of the Newer Granites.
Acknowledgements
We would like to acknowledge a review by M. B. Fowler which helped improve the manuscript. P. E. B. wishes to acknowledge much help and discussion on this problem over the years from W. E. Stephens.
