No Exploits back-arc basin in the Iapetus suture zone of Ireland

A controversial aspect of the closure history of the Iapetus Ocean concerns the existence of the Exploits basin in Ireland and Britain. The Exploits–Tetagouche back-arc basin of the Canadian Appalachians opened during the Middle Ordovician as the Popelogan–Victoria volcanic arc migrated northwards from the Ganderian margin. The Bellewstown Terrane, within the Iapetus suture zone of Ireland, lies between the Ganderian–Avalonian Leinster Terrane and the Laurentian Grangegeeth Terrane. An early Ordovician ‘Celtic’ shelly fauna is hosted in volcanogenic breccia that we demonstrate is stratigraphically overlain by shales with no volcanic horizons. The lower 24 m of shale are unfossiliferous, but shale above this stratigraphic level has yielded a new graptolite fauna consistent with the upper part of the artus Biozone (Darriwilian, c. 464 Ma). U–Pb zircon dating of the volcanic horizon yields an age of c. 474 Ma. The shale, therefore, appears to represent c. 10 myr of deposition and a significant gap in volcanism. This, together with a Sandbian Anglo-Welsh fauna, suggests that the Bellewstown Terrane remained in a peri-Gondwanan position throughout the Ordovician. Hence, Exploits back-arc basin opening and Ganderian arc migration did not occur in the Irish sector of Iapetus. Supplementary materials: The table of U–Pb zircon data, CL images of zircon grains, concordia plot of all U–Pb zircon analyses, and additional drawings and notes on the graptolite fauna are available at www.geolsoc.org.uk/SUP18858.

Plate tectonic reconstructions indicate that several volcanic arcs were sequentially created and accreted during Ordovician to Silurian subduction and closure of Iapetus (van Staal et al. 1998Chew et al. 2010;Hollis et al. 2012). The timing of volcanism in arc sequences, both flare-up and shut-off, is therefore an important constraint in reconstructing the closure history of Iapetus, especially when combined with faunal provincialism or other palaeogeographical indicators.
An element of the closure history of Iapetus that remains controversial in correlating sequences along the Caledonian-Appalachian orogeny concerns the existence of the Exploits basin in Ireland and Britain. The Exploits-Tetagouche (hereafter Exploits) back-arc basin of the Canadian Appalachians opened during the Middle Ordovician as the Popelogan-Victoria volcanic arc migrated northwards from the Ganderian margin above a retreating, southward subducting plate (van Staal et al. 1998;Zagorevski et al. 2010). This migration carried early Ordovician peri-Gondwanan sequences and faunas to the Laurentian margin by Sandbian times (Williams et al. 1995), while a Middle Ordovician basin fill including basalts and volcaniclastic turbidites was accumulated (O'Brien et al. 1997). Van Staal et al. (1998) proposed that the Ordovician volcanic Grangegeeth Terrane and the late Ordovician to Silurian black shale-turbidite sequences of the Longford-Down Terrane of eastern Ireland were elements of the Exploits basin, placing the Gondwanan-Laurentian boundary in Ireland further north than previously recognized. McConnell et al. (2010), however, found that the Grangegeeth Terrane was Laurentian throughout its history. Furthermore, Waldron et al. (2014) found no Gondwanan detrital zircon provenance signal in the late Ordovician to early Silurian sedimentary rocks of the Southern Uplands of Scotland and argued against the model that a volcanic arc founded on Ganderian crust had migrated to the Laurentian margin by late Ordovician times.
The Bellewstown Terrane is an Ordovician volcanic arc fragment within the broad Iapetus suture zone of eastern Ireland (Fig. 1). The terrane is generally regarded as 'intra-Iapetus' in affinity; it lies tectonically between the Ganderian-Avalonian Leinster Terrane to the south (Murphy 1987) and the Laurentian Grangegeeth Terrane to the north (McConnell et al. 2010) and contains a synvolcanic early Ordovician 'Celtic' shelly fauna that is neither Laurentian nor Gondwanan (Harper et al. 1990). An overlying Sandbian Anglo-Welsh fauna is more typically Avalonian (Parkes & Harper 1996). Harper et al. (1990Harper et al. ( , 1992 compared the Bellewstown 'Celtic' fauna with those of some volcanic arc sequences in the Exploits Subzone of Newfoundland. This paper presents new findings from recently created exposures in Bellewstown quarry. Study of these exposures has provided previously unknown stratigraphical constraints and yielded a new graptolite fauna that, together with a new U-Pb zircon date from a volcaniclastic sandstone, permits a detailed assessment of the timing of volcanism in the Bellewstown Terrane. Our new findings provide evidence against the existence of the Exploits basin in this sector of Iapetus.

Geological setting
The Bellewstown Terrane is bounded to the north by the Slane Fault against the Grangegeeth Terrane, and to the south by the Lowther Lodge Fault against the Leinster Terrane ( Fig. 1; Murphy 1987;Vaughan & Johnston 1992). The oldest exposed rocks are early Ordovician slates and siltstones with slump breccias and coticules of the Prioryland Formation (Harper & Rast 1964;Murphy 1987    1993; McConnell et al. 1999). The Prioryland Formation is overlain, with uncertain contact relations, by the c. 480 m thick Hilltown Formation of early to mid-Ordovician felsic volcanic and finegrained metasedimentary rocks (Harper & Rast 1964;Murphy 1987). Murphy (1987) suggested a calc-alkaline geochemical affinity for the felsic volcanic rocks. Winchester & van Staal (1995) instead proposed that the Hilltown Formation volcanic rocks are tholeiitic, but their few analyses were of cross-cutting minor intrusions of uncertain relationship to the volcanic rocks. An intra-Iapetus 'Celtic' brachiopod fauna (Paralenorthis sp., Jaanussonites sp., 'Ahtiella') has been documented from a volcanogenic breccia near the top of the Hilltown Formation at Bellewstown (Harper et al. 1990). The fauna does not give a tight age constraint, but suggests correlation with parts of the Dunnage Zone of Newfoundland; it is most similar to faunas in rocks of Arenig (Floian-Dapingian) age from the Summerford Group of the Exploits Subzone (Harper et al. 1990(Harper et al. , 1992. The Hilltown Formation was, however, assigned by Harper & Rast (1964) to the D. artus Biozone (Darriwilian: Llanvirn) based on graptolites from both above and below the breccia. The Harper & Rast localities no longer exist. Re-examination of the Geological Survey of Ireland collections from equivalent horizons (Rushton 1991) has found the graptolite faunas to be wide ranging or poorly preserved and not diagnostic of a specific biozone.
The Hilltown Formation is overlain by a 6 m thick unit of calcareous sandstone and bioclastic limestone assigned to the Bellewstown Member of the Carnes Formation. The member is poorly exposed, but Harper & Rast (1964) described a conformable contact with the Hilltown Formation. Although richly fossiliferous, the bioclastic material is fragmentary and a precise age has not been determined; conodonts suggest a late Llanvirn (late Darriwilian) age (Harper & Rast 1964;Parkes & Harper 1996). The limestone member is overlain by at least 300 m of Sandbian tuffaceous shales and dark mudstones of the Carnes Formation (Harper & Rast 1964;Murphy 1987).

Bellewstown quarry
New exposures within the Hilltown Formation in Bellewstown quarry ( Fig. 2) reveal details of the brachiopod-bearing volcanogenic breccia described by Harper et al. (1990). The breccia unit is stratified, with both normal and reverse grading and beds of volcaniclastic sandstone within the sequence. The upper breccia horizon grades through bedded sandstone and siltstone to dark graptolitic shale. This upper transition has been investigated by two cored boreholes through the section (Fig. 3), which demonstrate a continuous fining-upward sequence from breccia to shale. Siltstone and shale in the transitional sequence contain wispy laminae of reworked volcanogenic sand, composed of feldspar crystals and lithic grains of rhyolite, establishing a sedimentary link between the volcanic rocks and shales (Fig. 3).
We interpret the breccia units as deposits of fluctuating volcanic mass flow, despite the occurrence of brachiopods in part of the breccia matrix, because (1) clasts in the breccia are exclusively angular rhyolite, including some with jigsaw-fit shapes, or rip-ups of underlying volcanic sandstone; (2) both normal and reverse grading are present, and stratification is diffuse; (3) matrix and sandstone are composed of angular volcanic lithic grains and feldspar crystals in a devitrified glass matrix; (4) zircons extracted from sandstone (see below) are prismatic and magmatic in appearance.
Approximately 40 m of sulphide-rich shale are exposed in continuous quarry section above the volcaniclastic unit, without any further volcaniclastic beds or coarse sediment horizons. The lower 24 m of shale are unfossiliferous. The shale above 24 m has yielded a new graptolite fauna, preserved partly in pyrite.

U-Pb zircon geochronology
Zircons were separated from sandstone within the volcanogenic breccia, to obtain an age of the volcanism and to determine any provenance signal in inherited older grains (sample GSI10/220, col- Analysis of 75 zircon grains in total from the Hilltown volcaniclastic sandstone (Fig. 4) yields a single age-probability peak at c. 474 Ma and a 'TuffZirc' age of 474.1 +1.0/-2.6 Ma. A single grain gave a late Precambrian age (c. 580 Ma).
We interpret the Ordovician age of c. 474 Ma as the age of volcanism at Bellewstown, which is consistent with the age of the associated brachiopods. The late Precambrian age of a single grain may indicate inheritance from an Avalonian crustal source (Pollock et al. 2009), but the paucity of older grains supports the interpretation that the volcanogenic breccia to sandstone unit is a primary volcanic deposit.

New graptolite fauna
The graptolites collected from cleaved dark grey mudrocks of the Hilltown Formation in Bellewstown quarry (Fig. 5) are from a level more than 24 m above the contact with the volcanic rocks. The fauna includes the following: (1) pendent didymograptids typical of Anglo-Welsh Llanvirn faunas from England and Wales; (2) the distinctive sinograptid Nicholsonograptus fasciculatus (Fig. 5a-c); (3) a variety of scandent biserial graptolites, most of them being early diplograptid forms. Many of the graptolites are pyritized and preserved in low or moderate relief, and whereas this may have lessened the effects of tectonic deformation, weathering has commonly caused the pyrite to oxidize, leaving the graptolites difficult to interpret.
This assemblage has features in common with Middle Ordovician (Llanvirn) faunas in England and Wales, especially with the graptolite fauna that Skevington (1970) described from mudrocks of the Skiddaw Group in the eastern Lake District; and also from equivalent strata in Sweden and Norway.
In England and Wales, the Abereiddian Stage (the lower part of the Llanvirn regional series: Fortey et al. 2000) is characterized by pendent Didymograptus species that have been used to divide the succession into the artus graptolite Zone, below, and the murchisoni Zone, above (Zalasiewicz et al. 2009, pp. 817-820). So far, no stratigraphical succession has been reported that is adequate to show details of the contact between these zones. However, there are descriptions of correlative successions in the Upper Didymograptus Shale of Scania, Sweden (Ekström 1937;Hede 1951) and in corresponding shales in Oslo Fjord, Norway (Maletz 1997) that show the succession of graptolites at around this level. Both Ekström (1937) and Hede (1951) recognized the short-ranging species Azygograptus falciformis Ekström that occurs somewhat above the base of the Upper Didymograptus Shale. Skevington (1966, p. 495) recorded Ekström's species as a synonym of N. fasciculatus, and Maletz (1997, p. 36), in agreement with this, formalized a renaming of Ekström's (1937) falciformis Subzone as the fasciculatus Zone.
Based on the ranges of graptolites in Scandinavia, the fauna from Bellewstown quarry is assigned to the Nicholsonograptus fasciculatus Zone of Maletz (1997). Furthermore, the presence of Haddingograptus oliveri (Fig. 5c and d), which appears in the fasciculatus Zone in Norway and ranges up through the overlying Pterograptus elegans Zone (Maletz 1997, fig. 2, p. 10), shows that the Bellewstown fauna is not older than the upper part of the British artus Zone (Loydell 2012). In terms of the Australasian graptolitic succession, the Bellewstown fauna may be correlated with the middle of the Darriwilian Series. The lowest division, Da1, is correlated with the upper part of the British regional Arenig Series, and Da2 approximates to the lower part of the Didymograptus artus Zone at the base of the Llanvirn Series. Nicholsonograptus fasciculatus is recorded from the subdivision Da3 (VandenBerg & Cooper 1992, p. 62).

Discussion
Our U-Pb zircon age of 474.1 +1.0/-2.6 Ma for the Hilltown volcanic breccia is significantly older than the upper D. artus Biozone age of  Sadler et al. (2009) (Fig. 6). We use the Sadler et al. (2009) timescale because the subsequent Cooper & Sadler (2012) timescale includes a calibration point from Slieve Gallion (Cooper et al. 2008) that has since been shown to be unreliable (Hollis et al. 2013) and that distorts the position of the stage boundaries in the section around the zircon age from Bellewstown. Given that we have established a conformable strati-graphical transition between the two, the simplest explanation appears to be that the barren shale between the volcanic unit and the graptolitic shale represents c. 10 myr of deposition. This yields an average sedimentation rate of 2.4 m Ma −1 , which is consistent with rates for graptolitic shale in other Ordovician sequences (e.g. Churkin et al. 1977) and deep-sea sedimentation generally (Stow et al. 1996).
Uninterrupted shale deposition for c. 10 myr, without any tuff or bentonite layers, implies that volcanic activity in the Bellewstown Terrane ceased after emplacement of the c. 474 Ma volcaniclastic deposits. Volcanism in the Bellewstown sequence did not resume until after deposition of the Bellewstown Limestone Member, during early Sandbian deposition of the tuffaceous shales of the Carnes Formation. The section above the new graptolitic shale locality, through the Bellewstown limestone and into the Carnes Formation is poorly exposed, so it is not possible to be sure if volcanic horizons occur within that part of the sequence that would permit constraining the onset of renewed volcanism. The gap in volcanism is, however, at least throughout the upper Dapingian and lower Darriwilian (Fig. 6).
Upper Dapingian to lower Darriwilian volcanism is a feature of terranes of Laurentian affinity in Ireland (e.g. Dewey 1963;Graham et al. 1989;Hollis et al. 2012Hollis et al. , 2013, and the Exploits back-arc basin of Newfoundland includes a Middle Ordovician basin fill including basalts and volcaniclastic turbidites (O'Brien et al. 1997). In contrast, the Hilltown Formation records a late Dapingian to Darriwilian gap in volcanism at Bellewstown. The overlying stratigraphy and fauna suggest a geological history shared with the Ganderian-Avalonian Leinster Terrane. At Bellewstown, the Llanvirn limestone member below renewed Sandbian volcanism and Anglo-Welsh fauna in the Carnes Formation is comparable with the situation in the Leinster Terrane, where the late Llanvirn Courtown-Tramore limestone at the base of the volcanic Duncannon Group rests unconformably on Floian to possibly Dapingian Ribband Group metasedimentary rocks ( Fig. 6; Brenchley & Treagus 1970;McConnell 2000). Age control on the Ribband Group is poor (McConnell et al. 1999); a varicosus Biozone age for the Oaklands Formation (Rushton 1996) suggests a correlation with the Prioryland Formation at Bellewstown, in agreement with correlation on lithological grounds (Murphy 1987;Kennan & Murphy 1993). Thus it appears that, in the southern part of Leinster, the sequence equivalent to the Hilltown Formation of Bellewstown is missing below the unconformity, which has been related to a Monian orogenic event to the south (Tietzsch-Tyler 1996). In the small Kildare inlier, in the northern part of the Leinster Terrane, the fine-grained metasedimentary Conlanstown Formation contains D. artus Biozone graptolites and is overlain unconformably by the fossiliferous base of the volcanogenic mid-Sandbian Grange Allen Formation (Parkes & Palmer 1994). No volcanic rocks have been recorded from the Conlanstown Formation and it may be equivalent to the upper part of the Hilltown Formation at Bellewstown (Fig. 6). Sandbian volcanic rocks at Balbriggan and Lambay are correlated to the Kildare sequence as parts of a northern belt of the Upper Ordovician volcanic arc of SE Ireland (McConnell 2000), but no pre-Caradoc rocks are exposed in those places. The Lowther Lodge Fault separates the Balbriggan rocks from the Bellewstown Terrane to the north (Murphy 1987), but this fault records displacements within the active Ganderian margin of SE Ireland rather than accretion of exotic terranes.
After deposition of the early Ordovician rocks with 'Celtic' fauna, the Bellewstown Terrane became volcanically quiescent and remained in a peri-Gondwanan position, rather than migrating across Iapetus with the active Popelogan-Victoria volcanic arc sequences of the Exploits subzone of the Appalachians (see van Staal et al. 1998). The Bellewstown Terrane provides evidence that Exploits back-arc basin opening and arc migration did not occur in the Irish sector of Iapetus. Volcanic quiescence in the Hilltown Formation may equate to a pause in subduction related to initiation of Exploits basin extension. Renewed subduction produced abundant Upper Ordovician andesitic and felsic volcanism on the Ganderian margin in SE Ireland (McConnell 2000), which includes peralkaline rhyolites indicative of supra-subduction extension but no formation of basaltic crust (McConnell et al. 1991). Upper Ordovician volcanic rocks are absent from the equivalent margin to the south of the Exploits basin in Newfoundland because the active volcanic arc had migrated away. It is worth noting that extension-related peralkaline rhyolites at Avoca host the only significant economic volcanogenic massive sulphide deposit in the Ordovician rocks of SE Ireland, in contrast to the Central Mobile Belt of Newfoundland, where the Exploits back-arc basin hosts several such deposits (Rogers et al. 2007).
The Grangegeeth Terrane, to the north of Bellewstown, was formed at the Laurentian margin (McConnell et al. 2010) and so the Red Indian Line, the Laurentian-Gondwanan boundary, in eastern Ireland lies between Grangegeeth and Bellewstown along the Slane Fault and probably its extension along the Navan-Silvermines fault zone, the traditional 'Iapetus suture' in Ireland ( Fig. 1; Phillips et al. 1976;Todd et al. 1991). Furthermore, because there was no Exploits back-arc basin, Silurian ocean closure in Ireland occurred along the same structure (Vaughan & Johnston 1992), unlike in the Canadian Appalachians where there are separate Ordovician and Silurian sutures. Ir is h G a n d e ri a Iapetus slab roll-back  Our interpretation from Bellewstown and the Ganderian margin of Ireland, that the Exploits back-arc basin is absent from that sector of Iapetus, is supported by the interpretation of Waldron et al. (2014), based on zircon provenance analysis, that a volcanic arc founded on Ganderian basement is not present in the Southern Uplands Terrane of the Scottish Laurentian margin. We conclude that Iapetus closure was not cylindrical along the orogen between Newfoundland and Ireland and that arc migration and Exploits back-arc basin opening did not occur along the Ireland and Britain sector of the Ganderian margin. We suggest that a hinge or an oceanic structure such as a transform fault allowed decoupling of the southern Iapetus subduction zone between Newfoundland and Ireland (Fig. 7), so that the active Ganderian margin in Ireland was isolated from slab roll-back and Exploits back-arc extension at the margin in Newfoundland.