Geochemical and Lithostratigraphic Constraints on the Formation of Pillow-Dominated Tindars from Undirhlíðar Quarry, Reykjanes Peninsula, Southwest Iceland

Lithos 200–201 (2014) 317–333

Contents lists available at ScienceDirect

Lithos

journal homepage: www.elsevier.com/locate/lithos

Geochemical and lithostratigraphic constraints on the formation of pillow-dominated tindars from Undirhlíðar quarry, Reykjanes Peninsula, southwest Iceland

Meagen Pollock a,⁎, Benjamin Edwards b,SteinunnHauksdóttirc, Rebecca Alcorn a, Lindsey Bowman a a Department of Geology, The College of Wooster, 944 College Mall, Scovel Hall, Wooster, OH 44691, USA b Department of Earth Sciences, Dickinson College, Carlisle, PA 17013, USA c ÍSOR, Iceland GeoSurvey, Grensávegur 9, 108 Reykjavík, Iceland article info abstract

Article history: Undirhlíðar quarry on the Reykjanes Peninsula in southwest Iceland exposes an almost complete cross-section of Received 1 August 2013 a pillow-dominated tindar. Detailed mapping and geochemical analyses of the quarry walls show that Accepted 23 April 2014 glaciovolcanic lithologies are controlled not only by ice conditions, but also by complex changes in magmatic Available online 14 May 2014 conditions. Undirhlíðar's glaciovolcanic deposits are dominated by pillow lavas (Lp1–3) but also include dikes (Ld1–3) and shallow intrusions, and interbedded tuff (T), lapilli tuff (LT) and tuff-breccia (TB). Lithostratigraphic Keywords: variations record shifts in eruptive vents and styles, including evidence for explosive subaqueous activity. Petro- Iceland Glaciovolcanic graphically, the units can be subdivided into plagioclase-phyric and olivine-phyric, while compositionally the fi Tindar units de ne two trace element populations: (1) incompatible element-enriched (LaN/SmN ~1.6; Nb/Zr ~0.15) Dike rocks comprising the lower (older) pillow units (Lp1–2), and (2) less-enriched (LaN/SmN ~1.3; Nb/Zr ~0.125) Pillow units including dikes (Ld1–3), west wall pillows (LpW), and the upper (younger) pillow units (Lp3). The Hyaloclastite combined lithostratigraphic, petrographic, and geochemical relationships suggest a four-stage model for the formation of Undirhlíðar: (1) an initial effusive phase (Lp1–2) that built the bulk of the ridge, which erupted olivine-free, incompatible element-enriched lava, (2) an explosive phase, which generated lenses of TB and LT, now exposed on the eastern side of the ridge, (3) a second effusive phase (Ld1–2, LpW) exposed in the west side of the quarry, which intruded the initial effusive deposits and erupted pillow lavas that drape over the western edge of the existing ridge; this effusive phase is distinguished from the first as a less-enriched magma bearing large olivine phenocrysts, and (4) a final effusion on the east side of the quarry, which intruded (Ld3) units LT and TB, and erupted a capping layer of pillow lavas (Lp3). The products of this effusive event are genetically related to the previous effusive phase (LpW) but do not contain large olivine phenocrysts. The specificstratigraphic sequence of vitric fragmental units (TB2/LT) cut by dikes (Ld3) that feed pillow lava flows (Lp3) emplaced immediately above the fragmental units (TDP lithofacies association) is consistent with an explosive initiation to the final phase of volcanism evident in the quarry. The explosive unit (TB2/LT) also coincides with the transition from olivine-free, incompatible element-enriched lavas to olivine-phyric, less-enriched lavas. To explain the shift in trace elements between the early and late pillow lavas, we propose a model based on tapping of separate, small magma reservoirs at different depths. © 2014 Elsevier B.V. All rights reserved.

1. Introduction compared to subaerial lava flows, the emplacement dynamics of pillow lavas and the construction of pillow-dominated volcanoes are not as Pillow lavas are one of the most common types of lava morphologies well understood simply because of their relative inaccessibility; on Earth, forming in a wide variety of water-dominated environments the vast majority of present day pillow-dominated volcanoes form in including submarine (e.g., Batiza and White, 2000; Furnes et al., 2007; submarine environments along mid-ocean ridges (MOR) at depths of Johns et al., 2006; Walker, 1992), lacustrine (e.g. McClintock et al., at least 2.5 km. Although a wealth of information about submarine 2008), fluvial (Farrell et al., 2007), and glaciovolcanic (e.g., Edwards pillow lavas has been gleaned from high resolution mapping and sam- et al., 2009; Höskuldsson et al., 2006; Jones, 1969) settings. Yet pling of seafloor exposures (e.g., Ferrini et al., 2008; Soule et al., 2007), deep-sea drill cores (e.g., Tominaga et al., 2009), and analog experiments (Gregg and Fink, 1995, 2000), significant questions remain ⁎ Corresponding author. Tel.: +1 330 263 2202 (office); fax: +1 330 263 2249. about the eruptive processes that control the development of pillow- E-mail address: [email protected] (M. Pollock). dominated volcanic edifices.

M. Pollock et al. / Lithos 200–201 (2014) 317–333 319

Fig. 2. Geological sketch map, Undirhlíðar quarry. A. Sketch map showing the locations of lithofacies described in the text and informal location names. Boxes show approximate locations of sketches in Fig. 4. B. Sketch map showing locations of samples analyzed for major and trace element geochemistry.

For example, most models for the formation of basaltic glacio- Schmincke, 1999). These models are largely derived from ﬁeldwork at volcanoes show a single period of effusive pillow formation, followed Pleistocene glaciovolcanic deposits, which are now exposed due to by a transition to entirely explosive eruptions as water depth shallows Holocene ice retreat in western Canada (Allen et al., 1982; Edwards (e.g., Allen et al., 1982; Hickson, 2000; Jakobsson and Johnson, 2012; et al., 2009, 2010; Hickson et al., 1995; Hickson, 2000; Mathews, 1947; Jones, 1970; Skilling, 1994, 2009; Smellie, 2001, 2008; Werner and Moore et al., 1995), Antarctica (e.g., Smellie and Skilling, 1994; Smellie

Fig. 1. A. Map showing the active volcanic centers in Iceland (after Fig. 1, Edwards et al., 2012). B. Detailed map showing the location of Undirhlíðar (outlined black box) at the northern end of the Sveifluhals ridge (after Saemundsson et al., 2010), in the central part of the Reykjanes Peninsula. C. Aerial overview and general distribution of units in the vicinity of Undirhlíðar quarry. 320 M. Pollock et al. / Lithos 200–201 (2014) 317–333 et al., 2008), and especially in Iceland (Allen, 1980; Bennett et al., 2009; dike-fed pillow units. Based on his field observations, he proposed what Höskuldsson et al., 2006; Jakobsson and Johnson, 2012; Jarosch et al., has become a standard model for many basaltic glaciovolcanic eruptions: 2008; Jones, 1969, 1970; Loughlin, 2002; Moore and Calk, 1991; (1) initiation with effusion of pillow lava at water depths of greater than Saemundsson, 1967; Sigvaldason, 1968; Smellie, 2008; van Bemmelen 500 m, switching to (2) explosive eruptions and production of tephra at and Rutten, 1955; Werner and Schmincke, 1999; Werner et al., 1996). shallow water depths (~100–200 m), leading to (3) possible subaerial However, recent observations of glaciovolcanic pillow exposures in effusion. He noted that, in general, vesicularity of pillows seemed to Iceland (Alcorn et al., 2010; Bennett et al., 2009; Bowman et al., 2011) increase from the base to the top of the pillow pile, which is consistent and western Canada (Edwards et al., 2009) show packages of pillow with a growing edifice in a steady-state englacial lake. lavas separated by deposits of fragmental volcanic material, suggesting Höskuldsson et al. (2006) studied a series of pillow ridges in the that pillow ridges may be produced by a complicated sequence of erup- Kverkfjöll area of north-central Iceland. Their main focus was the varia- tive events (e.g., Johnson and Jakobsson, 1985). tion in vesicularity of pillows within the ridges, and they postulated that The most detailed studies of basaltic tindars, which are volcanic ridges those variations could be explained by sudden pressure releases caused formed by eruptions beneath ice (e.g., Edwards et al., 2009; Jakobsson by floods from the eruption site. Based on measured values of H2Ocon- and Gudmundsson, 2008; Jones, 1969), include the work of Jones centrations in pillow rim glasses, they estimated that the pillows formed (1969, 1970), Höskuldsson et al. (2006), and Edwards et al. (2009). beneath relatively thick ice (N1000 m), and that sudden flood events Jones (1969, 1970) described lithofacies variations at Kalfstindar, dropped the level of the englacial lake at the eruption site by as much Iceland, where he recognized a fundamental separation of basal, pillow- as 400 m. dominated material from an upper sequence of fragmental material. Edwards et al. (2009) described the lithostratigraphy at Pillow Ridge He noted small, intra-pillow zones of volcanic breccia, and possible in northern British Columbia, which is the only detailed description of a

A D

Pillow Lava (Lp1) (TB1)

Pillow Lava (LpW) B Intrusion E (Li) (LT)

~10 m

Pillow Lava (LpW)

C F Pillow lava Pillow Lava (Lp1) (Lp3)

Dike (Ld3) (TB2/LT)

Dike (Ld2) (T)

Fig. 3. Field images of lithofacies described in the text. A. Pillow lavas (Lp1), south-central quarry with person for scale. B. Intrusion, northwest quarry wall. C. Dike (Ld3) intruded through tuff-breccia (TB1) feeding pillow lavas (Lp3), southeastern quarry wall with person for scale. D. Tuff-breccia (TB1) comprising dominantly angular lithic fragments derived from Lp units; hammer for scale is 45 cm long. E. Vitric tuff-breccia/lapilli tuff (TB2/LT) comprising vitric clasts; horizontal ruler segment is 42 cm long. F. Tuff (T) interbedded with Lp1 and Lp2, intruded by a small dike (Ld2); lens cap for scale is 6 cm in diameter. M. Pollock et al. / Lithos 200–201 (2014) 317–333 321 large tindar outside of Iceland. They demonstrated that long-lived suggests that glaciovolcanic lithologies are controlled by the complex eruptions could produce significant lithostratigraphic complexity in a interaction between ice/water and magmatic conditions. glaciovolcanic environment, and they recognized a sequence of five different pillow-dominated effusive lithofacies units interstratified 2. Geological setting with 14 other lithofacies to form five lithofacies associations. Edwards et al. (2009) argued that the variations in stratigraphy and H2Oconcen- The Reykjanes Peninsula in southwest Iceland connects the Western trations from pillow rims recorded a complex history of water level Volcanic Zone to the submarine portion of the slow-spreading changes over a prolonged glaciovolcanic eruption. They also described Reykjanes Ridge (Fig. 1). This active and subaerially exposed portion of a unique lithofacies association of tuff-breccia, dikes and pillow lava the mid-ocean ridge system is broken into en echelon NE–SW-trending that was preserved in the upper part of the ridge; they interpreted the volcanic segments (from west to east; Fig. 1: Reykjanes, Krisuvik, association as recording collapse of a growing proximal pillow mound, Brennisteinsfjöll and Hengill; Jakobsson et al., 1978; Saemundsson, with subsequent pillow flows advancing over the tuff-breccia. 1980). Pleistocene volcanism focused primarily in the present-day This study presents a high-resolution investigation of a section volcanic systems produced NE-trending glaciovolcanic ridges reaching through a pillow ridge at Undirhlíðar quarry on the Reykjanes Peninsula elevations of 300–400 m a.s.l. (Saemundsson et al., 2010; Schopka in southwest Iceland (Fig. 1), which highlights the petrogenetic com- et al., 2006) that, today, protrude through post-glacial lava fields (Fig. 1). plexity possible during pillow ridge formation. We demonstrate that a Undirhlíðar is a quarry located at 63°59′46″ Nand21°53′24″ Watthe combination of lithostratigraphic mapping and detailed petrographic northern extension of the Sveifluhals ridge, which is part of the Krisuvik and geochemical observations in areas of exceptional exposure yields fissure system (Fig. 1). Glaciovolcanic eruptions along a ~7 km fissure new insights on the formation of pillow-dominated tindars. We produced the volcanic ridge, a ~150 m thick sequence of interstratified describe the lithostratigraphy, petrology and geochemistry for part of pillow lavas, tuff-breccias and tuffs overlain by glaciofluvial diamict. the quarry, and discuss a possible scenario that explains the observed The Pleistocene-age pillow lavas, intrusions and volcanic tuff-breccias geochemical and lithostratigraphic variations. The proposed model were mined within the quarry, which is 0.6 km long and 0.25 km wide

CDSouth

Lp1

Ld2

T Ld1

[Talus/unconsolidated cover]

Fig. 4. A. Annotated photomosaic of the east wall of Undirhlíðar (Central Wall on Fig. 1) showing interbedded pillow (Lp1–3) and tuff-breccia units (TB1, TB2/LT). B. Annotated image of the south wall of Undirhlíðar (South Wall on Fig. 1) showing two olivine-phyric dikes (Ld1–2) intruding through the lowermost pillow unit (Lp1) and a lens of stratiﬁed tuff (T). 322 M. Pollock et al. / Lithos 200–201 (2014) 317–333

(Fig. 2A). Regional mapping has assigned most of the volcanic ridge the quarry, but are most obvious on the western wall (Fig. 3B). It is to the Weichselian (118–10 kyr; Saemundsson et al., 2010), with the unclear from the two dimensional quarry walls if these intrusions northernmost end mapped as Early Weichselian (118–80 kyr; Lambeck extend laterally or are relatively small apophyses. All of the bodies et al., 2006). show well-developed cooling joints with spacings of 10s of cm. This study focuses on the well exposed southern and eastern walls of Only three intrusive bodies occur in the southern half of the the quarry, where a variety of volcanic and volcaniclastic lithofacies are quarry, all of which can clearly be seen to cross-cut other identified, mapped, and geochemically characterized. The southern wall, lithofacies (Figs. 2 A; 3 C, F; 4 C).Onthewesternpartofthesouth which is ~10 m tall and ~200 m long, is oriented east–west, providing wall, two tabular bodies, ~2.5 m (Ld1) and ~0.25 m wide (Ld2), in- a cross-section of the pillow ridge. The eastern walls, which are up trude units Lp1 and a tuffaceous unit (Fig. 4C, D). Both bodies con- to ~20 m tall and ~350 m long, are generally oriented north–south, pro- tain plagioclase and olivine crystals up to 1 cm in size. The third viding a longitudinal section. The western wall parallels the eastern wall. tabular body, ~3.5 m wide (Ld3), occurs on the southern end of the east wall and clearly intersects the overlying pillow lava flow (Lp3). The 3. Physical and petrological characteristics of undirhlíðar third body does not contain large olivine phenocrysts like the other two bodies, but its center has an unusual glassy, pillow-like appearance. The quarry exposes an almost complete cross-section across the All of these units are considered to be the result of either shallow width of the pillow-dominated ridge, with excellent exposures of level intrusion into the growing tindar (e.g., Ld1–3), or parts of the dikes, shallow intrusions, and interbedded fragmental units (Fig. 2). within-ridge lava transport network. The units in the southern half Three coherent (pillow lava, dikes, irregular intrusions; Fig. 3A–C) and of the quarry are clearly dikes and Ld3 appears to have been the lava three volcaniclastic [seno lato] (tuff, lapilli tuff, tuff breccia; Fig. 3D–F) supply for Lp3 pillows. lithofacies have been identified in the quarry, which form a minimum of eight different lithostratigraphic units. All of the units are basaltic 3.1.3. Tuff-breccia and lapilli tuff (TB1–2/LT) with varying proportions of plagioclase, olivine, and clinopyroxene as The dominant fragmental lithofacies is tuff-breccia (TB; Figs. 2A; phenocrysts and in the groundmass. 3D, E), which is locally gradational to lapilli tuff (LT; Fig. 3E). Tuff- breccia has three different modes of occurrence: i) intermixed with 3.1. Lithofacies descriptions and interpretations steeply plunging pillow lavas in Lp2 and LpW; ii) as massive deposits with isolated pillow clasts (TB1; Fig. 3D); and iii) as massive deposits 3.1.1. Pillow lava (Lp1–3, LpW) with vitric, highly vesiculated clasts in a matrix that is transitional to Plagioclase-phyric basaltic pillow lava with or without olivine is lapilli tuff (TB2/LT; Fig. 3E). TB1 and TB2/LT occur in deposits along the dominant lithofacies exposed in the quarry walls (Figs. 2A; 3A); the eastern side of the quarry. Where seen in contact along the South- four units are distinguished based on spatial and stratigraphic eastern Wall, TB2/LT unconformably overlies TB1. criteria. Lp1 forms the lowest exposed unit in the quarry floor in The two TB units are texturally distinct. While all of the TB units the southern half of Undirhlíðar and occurs along the Central and South- lack well-defined sorting, TB1 is distinctly heterogeneous with local ern walls. It is overlain sequentially by Lp2 and Lp3 along the Central Wall large pillow fragments (N0.5 m). Compared to TB1, TB2/LT shows (Figs. 2 A; 4A). Unit LpW is not seen in direct contact with the other pillow units, but is exposed within the western wall of the quarry (Fig. 2A). Individual pillows range in size from 0.3 to 2 m, with average pillow A dimensions of ~1 m (Fig. 3A). Where exposed, the majority of pillow surfaces are not strongly corrugated. Pillows generally have lithic interiors with poorly preserved glassy rims on the order of 1 cm in thickness. Most pillows are vesiculated with a range of gross-scale vesicle arrangements, varying from uniformly distributed vesicles, to concentric layers, to highly vesiculated cores and poorly vesiculated rims; no systematic variations within individual units have been identified yet. Pipe vesicles are present but not common. Concentric fracture patterns are well-developed in pillows without highly vesiculated cores. Most pillows are ovoid in cross-section with locally sagged pillows (e.g. Walker, 1992). Plunges of individual pillows in Lp1,3 tend to be relatively low (b20°), whereas pillows in Lp2 and LpW plunge more steeply (N20°). Lp1–3 are similar in hand sample mineralogy. They are fine- B grained with few plagioclase crystals up to 5 mm long. Upon thin- section examination (see Section 3.2), Lp3 was also found to contain sparse microphenocrysts (b0.25 mm) of olivine. LpW contains large (up to 1 cm) olivine phenocrysts. The pillow units are interpreted as having formed during four discrete eruptions, with some diachroneity evident. Based on standard relationships between effusion rates and subaqueous lava morphologies (e.g., Gregg and Fink, 2000), we interpret the pillow units as having formed at relatively low effusion rates.

3.1.2. Intrusions (Li)/dikes (Ld1–3) Larger, more massive bodies of basalt are found throughout the quarry, although they are more abundant in the northern half (Fig. 3B). The relationship of the bodies to surrounding lithofacies appears to be intrusive but not necessarily cross-cutting (Li). These ir- Fig. 5. Textures in handsamples of olivine- and plagioclase-phyric units. A. Olivine- and regular intrusions are exposed on both sides of the northern half of plagioclase phyric intrusion (Li). B. Plagioclase-phyric pillow lava (Lp1–2). M. Pollock et al. / Lithos 200–201 (2014) 317–333 323 some crude sorting and has smaller clasts. The outsides of many of the Where TB1 occurs without intact pillow lavas, it likely represents clasts in TB2/LT are vitric. The clasts have ﬂuidal shapes and vesiculated deposits somewhat distal to the advancing pillow mound front interiors when broken open. or resulted from larger-scale collapse of entire mound fronts. In con- The TB units apparently formed by two distinct processes. trast, the distinctly smaller, vitric clasts in TB2/LT do not appear to be Occurrences of TB1 intermixed with Lp2 and LpW most likely resulted broken fragments of pillow lavas; instead, they are interpreted to from gravitational collapse of pillow lavas plunging down relatively be subaqueously erupted bombs that formed during brief periods of steep slopes, as shown by the steeper plunges of the adjacent pillows. subaqueous Strombolian activity (e.g., Head and Wilson, 2003).

A Dikes Pillow Lavas i ii iii plag No Visible Ol in

Hand ol Sample ol vesicle vesicle vesicle

vesicle iv vesicle v vi ox Ol plag ol cpx Observed ol in cpx ol vesicle Hand Sample

Ol Chr Chr Ol

Chr

Fig. 6. Photomicrographs representative of the lithostratigraphic units. Images are in crossed-polars and scale bars are 1 mm long. The ﬁrst column shows dikes; columns 2 and 3 show pillow lavas. The top row shows units in which no olivine phenocrysts are visible in hand sample. The bottom row shows units in which olivine phenocrysts are observed in hand sample. Vesicles and mineral phases are labeled (ol — olivine; cpx — clinopyroxene; plag — plagioclase; ox — FeTi oxide). A. Ld3 (S-10-21); B. Lp3 (S-10-23); C. Lp1 (S-10-10). D. Ld2 (S-10-06); E. LpW (W-08-02); F. Lp1 (S-10-24). 324 M. Pollock et al. / Lithos 200–201 (2014) 317–333

Table 1 Major and trace element compositions of Undirhlíðar dikes and lavas. Values in italics were analyzed by DCP. Bold values were analyzed by XRF. All other trace element values were determined by ICP-MS. LOI was not determined where no value is given.

ID S-08-04-gl S-08-04 S-10-05 S-08-01-gl S-08-01 S-10-08 S-10-09 S-10-10 S-10-14 S-10-15 S-10-16 S-10-17 S-10-18 S-10-22 S-10-26 E-11-02

Materiala g wr wr g wr wr wr wr wr wr wr wr wr wr wr wr

Unitb Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1 Lp1

Major elements (wt.%)

SiO2 49.68 48.91 49.68 50.61 50.57 49.18 49.65 50.14 49.69 52.55 49.40 49.13 49.38 49.87 54.12 47.83

TiO2 1.61 1.50 1.53 1.59 1.48 1.54 1.55 1.55 1.55 1.45 1.55 1.51 1.48 1.56 1.34 1.71

Al2O3 14.65 15.53 15.05 14.60 16.38 14.73 14.67 14.58 14.77 13.88 14.64 14.69 14.69 14.58 13.94 15.34 FeO*c 11.59 11.21 11.23 11.58 11.03 11.24 11.33 11.43 11.28 10.59 11.37 11.15 11.18 11.45 9.94 12.12 MnO 0.20 0.19 0.19 0.20 0.19 0.20 0.20 0.20 0.20 0.18 0.20 0.19 0.19 0.20 0.17 0.20 MgO 7.33 7.89 7.85 7.60 7.85 7.82 7.83 7.87 7.96 7.33 8.27 7.94 8.77 8.22 7.08 8.52 CaO 12.34 12.12 12.46 12.63 12.01 12.49 12.52 12.62 12.60 11.81 12.96 12.41 12.74 12.79 11.11 11.79

Na2O 2.23 2.13 2.19 2.20 2.26 2.14 2.16 2.19 2.14 2.46 2.06 2.12 2.06 2.05 2.69 2.18

K2O 0.33 0.19 0.36 0.28 0.39 0.31 0.26 0.30 0.27 0.30 0.26 0.31 0.31 0.30 0.17 0.17

P2O5 0.13 0.13 0.19 0.23 0.15 0.18 0.19 0.17 0.17 0.17 0.20 0.17 0.17 0.21 0.15 0.15 Total 100.09 99.81 100.75 101.52 102.32 99.82 100.36 101.05 100.62 100.71 100.91 99.62 100.99 101.23 100.73 100.00 LOI (wt.%) ––0.2 ––1.4 0.3 0.0 0.1 0.0 0.5 0.0 0.0 0.0 0.3 0.1

Trace elements (ppm) Sc 43.2 43.5 44.9 43.2 42.0 42.7 43.6 43.9 43.4 42.1 44.7 44.3 43.5 41.6 44.6 V 320 316 358 324 309 330 336 338 332 329 341 342 326 317 340 294 Cr 162 179 179 162 170 162 170 167 168 166 182 180 202 171 175 217 Ni 86 99 88 86 97 80 82 82 87 81 94 93 116 91 89 113 Cu 177 172 139 170 159 124 131 136 139 125 136 152 122 152 140 116 Zn 96 89 94 96 90 87 90 91 87 98 93 91 87 86 91 81 Rb 5.65 5.48 5.66 4.92 5.05 5.11 4.69 5.04 4.36 4.77 4.50 4.86 4.84 Sr 197 190 200 199 183 185 191 189 185 188 198 198 191 184 191 195 Y 26 23 24 27 22 23 23 23 23 23 24 24 22 23 23 21 Zr 88 92 87 86 89 88 87 87 90 90 84 85 87 75 Nb 12.97 13.33 12.68 12.68 12.98 12.97 12.60 12.92 13.06 13.03 12.22 12.57 12.81 13 Mo 0.62 0.55 0.55 0.52 0.53 0.56 0.51 0.49 0.49 0.52 0.51 0.53 0.53 Cs 0.051 0.067 0.057 0.056 0.055 0.058 0.055 0.057 0.051 0.054 0.053 0.053 0.053 Ba 79 76.04 74.76 80 74.40 70.25 73.17 72.92 70.05 72.68 72.78 71.65 67.78 71.65 69.98 68 La 8.13 8.62 7.97 8.39 8.67 8.60 8.52 8.63 8.76 8.50 8.17 8.29 8.40 Ce 19.90 20.27 19.30 19.35 20.24 20.03 19.76 20.03 20.08 20.07 18.93 19.44 19.67 Pr 2.85 2.97 2.75 2.87 2.94 2.93 2.89 2.95 2.97 2.91 2.77 2.86 2.87 Nd 12.73 13.17 12.30 12.72 13.01 13.05 12.79 13.06 13.34 13.01 12.42 12.85 12.76 Sm 3.33 3.51 3.18 3.33 3.47 3.46 3.41 3.47 3.54 3.46 3.21 3.38 3.38 Eu 1.18 1.23 1.15 1.19 1.19 1.21 1.18 1.19 1.21 1.21 1.14 1.20 1.17 Tb 0.66 0.68 0.64 0.65 0.67 0.65 0.65 0.67 0.67 0.66 0.64 0.64 0.64 Gd 3.91 4.02 3.81 3.89 3.90 4.00 3.90 3.93 4.01 3.99 3.78 3.85 3.84 Dy 4.03 4.18 3.89 4.05 4.10 4.20 4.05 4.14 4.21 4.16 3.93 3.98 4.03 Ho 0.85 0.86 0.81 0.84 0.82 0.85 0.82 0.84 0.86 0.84 0.82 0.81 0.82 Er 2.32 2.42 2.25 2.29 2.29 2.34 2.29 2.32 2.35 2.31 2.22 2.23 2.26 Yb 2.17 2.28 2.05 2.23 2.21 2.31 2.27 2.24 2.31 2.22 2.15 2.23 2.18 Lu 0.34 0.35 0.33 0.35 0.35 0.35 0.35 0.34 0.35 0.34 0.34 0.34 0.35 Hf 2.33 2.29 2.28 2.25 2.21 2.27 2.24 2.22 2.27 2.26 2.19 2.17 2.23 Ta 0.73 0.74 1.02 0.72 0.74 0.75 0.73 0.73 0.75 0.72 0.71 0.71 0.72 Pb 0.623 0.643 0.676 0.575 0.562 0.599 0.558 0.584 0.561 0.561 0.538 0.585 0.580 Th 0.626 0.672 0.609 0.664 0.669 0.685 0.659 0.668 0.672 0.663 0.653 0.630 0.662 U 0.185 0.192 0.187 0.197 0.199 0.214 0.184 0.197 0.191 0.193 0.184 0.196 0.192

a Material is designated as glass (g) or whole-rock powder (wr). b Unit refers to pillow units (Lp1 to Lp3), dikes (Ld1 to Ld3), and west wall pillows (LpW). c FeO* is total Fe recalculated as FeO.

3.1.4. Tuff (T) occurrence in the Southeastern Wall may indicate deposition during a Relatively thin lenses of moderately sorted, vitric ash-sized tuff (T) short hiatus from local pillow formation, while the more laterally exten- occur mainly in the southern half of the quarry. Along the Central Wall sive T unit along the Central Wall appears to demarcate a longer period (Fig. 4A), a more or less continuous lens of T separates Lp1 from Lp2, of quiescence between effusion of Lp1 and Lp2. while along the Southeastern Wall, a lens of T is interstratiﬁed with Lp1 for approximately 10 m and is intruded by Ld1 and Ld2 (Fig. 4C). 3.2. Petrography The lowermost T unit is an indurated yellow vitric tuff that separates pillow units Lp1 and Lp2 (Figs. 3F; 4A, B). The tuff is laterally continuous Samples from all lithofacies were examined using standard polariz- with thicknesses generally b30 cm, and locally has bombs of brownish ing light microscopy, and selected samples were also examined using sideromelane. The unit appears to be subhorizontal in orientation. a JEOL 5900 Scanning Electron Microscope equipped with an Oxford The T units likely represent small granules of pillow rim glass Energy Dispersive Spectrometer (EDS); the EDS was used to obtain and can be interpreted as hyaloclastites sensu stricto. The limited semi-quantitative estimates of major cations in olivine and plagioclase M. Pollock et al. / Lithos 200–201 (2014) 317–333 325

S-10-25 S-10-24 E-08-01-gl E-08-03-gl E-08-04 E-11-01 E-11-03 E-11-05 E-11-06 S-10-23 S-11-02 E-08-02-gl 12BRE632 W-08-01 W-11-01 12BRE631

wrwrg g wrwrwrwrwrwrwrg Wrwrwrwr

Lp1 Lp1 Lp2 Lp2 Lp2 Lp3 Lp3 Lp3 Lp3 Lp3 Lp3 Lp3 Lp3 LpW LpW LpW

49.30 49.25 49.86 49.56 50.18 47.88 47.95 48.78 48.04 47.57 47.98 48.51 47.99 48.74 47.47 47.99 1.56 1.55 1.57 1.55 1.58 1.72 1.68 1.44 1.75 1.79 1.74 1.75 1.56 1.56 1.61 1.64 14.58 14.71 14.72 14.75 14.62 15.40 15.45 15.15 15.32 15.57 15.39 15.09 16.67 15.47 15.70 16.16 11.48 11.25 11.54 11.50 11.64 12.13 11.98 10.86 12.21 12.03 12.14 12.36 11.45 11.53 11.48 11.36 0.20 0.20 0.20 0.20 0.20 0.20 0.20 0.19 0.20 0.20 0.20 0.21 0.19 0.19 0.19 0.19 8.00 7.75 7.79 7.81 7.99 8.37 8.50 8.74 8.15 7.95 8.25 8.01 8.18 9.44 9.40 8.44 12.41 12.39 12.37 12.35 12.33 11.76 11.72 12.48 11.79 11.30 11.76 11.77 11.68 12.00 11.77 11.83 2.17 2.15 2.17 2.18 2.18 2.21 2.20 2.07 2.20 2.18 2.20 2.25 2.21 2.09 2.09 2.12 0.34 0.32 0.21 0.24 0.52 0.16 0.17 0.23 0.17 0.27 0.18 0.25 0.13 0.11 0.14 0.15 0.19 0.16 0.22 0.22 0.16 0.15 0.15 0.14 0.15 0.19 0.16 0.17 0.13 0.13 0.15 0.15 100.22 99.74 100.66 100.35 101.40 100.00 100.00 99.99 100.00 98.95 99.99 100.36 99.99 101.28 100.00 99.99 0.0 0.5 –– –0.0 0.0 0.0 0.0 0.4 0.0 – 0.1 0.4 0.0 0.1

43.2 45.8 43.8 47.3 43.2 42.6 41.1 37.1 332 360 322 354 326 341 299 323 338 375 327 352 320 305 315 332 170 176 164 199 171 273 244 299 287 284 284 265 304 303 379 378 84 90 91 102 95 122 110 121 113 121 117 117 131 176 174 148 134 146 188 197 158 142 95 96 135 165 132 173 160 145 115 163 90 92 90 124 94 93 83 88 94 103 92 99 94 91 91 95 4.83 6.12 5.91 4.01 5.32 2.38 3.04 1.87 188 203 190 206 188 209 195 204 201 208 198 191 226 200 214 221 24 24 23 26 24 25 23 23 25 26 24 25 23 22 23 24 90 93 91 95 92 77 71 73 78 86 76 86 71 79 78 80 13.44 13.28 13.38 14.08 13.57 11 11 10 11 10.76 11 10.44 11 9.69 12 12 0.52 0.56 0.58 0.56 0.58 0.39 0.60 0.35 0.051 0.052 0.056 bDet. limit 0.056 0.027 0.021 0.016 76.44 76.64 79.07 78.83 74.88 55 44 45 51 53.38 47 50.44 51 38.89 41 42 9.09 8.66 8.46 8.82 8.45 6.85 6.57 5.82 20.85 22.24 20.59 19.61 20.82 17.22 16.90 15.37 3.04 2.74 2.94 3.05 2.92 2.67 2.53 2.30 13.54 13.13 13.08 13.97 12.96 12.55 11.95 10.68 3.63 3.52 3.37 3.78 3.39 3.60 3.27 3.06 1.26 1.28 1.21 1.31 1.19 1.32 1.20 1.12 0.68 0.69 0.68 0.74 0.66 0.73 0.71 0.63 4.13 4.12 4.03 4.36 3.97 4.38 4.13 3.72 4.23 4.44 4.19 4.44 4.18 4.58 4.48 3.94 0.87 0.87 0.87 0.94 0.87 0.95 0.95 0.82 2.40 2.43 2.38 2.55 2.32 2.65 2.52 2.19 2.34 2.30 2.27 2.36 2.21 2.53 2.35 2.07 0.36 0.36 0.36 0.41 0.34 0.39 0.36 0.32 2.38 2.33 2.36 2.41 2.39 2.29 2.39 2.13 0.77 0.70 0.84 6.93 1.50 0.61 0.65 0.75 0.587 0.603 1.488 bDet. limit 0.644 0.441 0.591 0.401 0.701 0.698 0.621 0.723 0.643 0.401 0.362 0.284 0.214 0.100 0.201 0.170 0.190 0.107 0.107 0.081

(continued on next page)

feldspar, and to identify opaque grains. All samples examined are oliv- Ol phenocrysts have rims that are more Fe-rich than their cores based ine, plagioclase-phyric basalts with porphyritic textures (Fig. 5A–B). on backscatter electron imagery. Average compositions (from EDS Units were differentiated in the ﬁeld by the presence/absence of analyses) for Lp1 are Fo81 (n = 48, range is Fo87–59). Most pillows abundant olivine phenocrysts, whereas optical microscopy revealed are vesiculated to some degree, and some show ﬁne-grained alteration that olivine microphenocrysts were also present in some units (Fig. 6). along vesicle walls. Thin sections of the Lp1 and Lp2 pillow units reveal intersertal textures Olivine-bearing units (Ld1–2, LpW) show textures and mineral with plagioclase laths (b0.5 mm) set in a glassy matrix of plagioclase, assemblages similar to the pillow units, with the addition of olivine augite, and Fe–Ti oxides (Fig. 6Aiii). Plagioclase crystals in the ground- and clinopyroxene phenocrysts (Fig. 6 Aiv–vi). The phenocrysts are mass commonly show skeletal and swallow tail quench textures. Aver- largest in the dikes, with olivine grains up to 1 cm long and plagioclase age compositions (from EDS analyses) for Lp1 are An70 (n = 51, range grains N2 cm, and show disequilibrium textures. Plagioclase grains have is An81–48). Olivine is absent in nearly all Lp1 and Lp2 samples, complex zoning, rounded edges, and contain inclusions of Fe–Ti oxides. although it occurs as a phenocryst phase in Lp1 sample S-10-24. Locally Average compositions (from EDS analyses) for Lp1 are An75 (n = 47, 326 M. Pollock et al. / Lithos 200–201 (2014) 317–333

Table 1 (continued)

ID S-10-01 S-10-02 S-10-03 S-10-12 S-10-06 S-10-07 S-10-11 S-10-19 S-10-20 S-10-21 S-08-02-gl S-08-02 S-08-03 12BRE628 12BRE630 Materiala wr wr wr wr wr wr wr wr wr wr g wr wr wr wr Unitb Ld1 Ld1 Ld1 Ld1 Ld2 Ld2 Ld2 Ld3 Ld3 Ld3 Ld3 Ld3 Ld3 TB TB

Major elements (wt.%)

SiO2 49.34 48.25 48.80 47.21 48.26 47.96 47.54 48.52 49.66 48.34 51.47 47.94 50.32 48.10 48.20

TiO2 1.51 1.57 1.52 1.37 1.50 1.53 1.48 1.72 1.59 1.66 1.56 1.52 1.69 1.63 1.71

Al2O3 15.24 15.51 15.26 16.12 15.24 15.33 15.42 15.63 14.98 15.44 14.19 17.58 15.55 16.31 15.72 FeO*c 11.16 11.44 11.20 10.68 11.12 11.43 11.17 12.17 11.55 12.00 11.17 11.32 12.23 11.53 11.93 MnO 0.19 0.19 0.18 0.18 0.18 0.19 0.19 0.19 0.19 0.20 0.19 0.19 0.20 0.19 0.20 MgO 9.22 9.27 9.39 9.89 9.25 10.03 9.71 8.40 8.22 8.39 7.29 8.14 8.48 8.04 8.08 CaO 11.58 11.93 11.94 11.06 11.60 11.76 11.54 11.50 11.24 11.85 11.48 11.61 11.95 11.73 11.65

Na2O 2.20 2.10 2.11 2.08 2.04 2.02 2.04 2.18 2.35 2.20 2.43 2.14 2.30 2.18 2.17

K2O 0.19 0.20 0.22 0.29 0.18 0.18 0.17 0.54 0.18 0.20 0.49 0.10 0.11 0.16 0.17

P2O5 0.15 0.17 0.18 0.17 0.14 0.20 0.17 0.08 0.15 0.13 0.14 0.16 0.13 0.14 0.15 Total 100.77 100.62 100.80 99.04 99.52 100.64 99.43 100.95 100.12 100.42 100.42 100.70 102.96 100.01 99.99 LOI (wt.%) 2.3 0.2 0.1 0.0 1.4 0.8 0.4 1.2 0.0 0.7 –––0.2 0.6

Trace elements (ppm) Sc 36.2 36.4 37.8 33.2 37.5 36.3 35.7 39.1 39.2 42.2 42.2 40.0 41.0 V 322 317 317 292 322 309 305 353 358 373 337 311 347 347 357 Cr 323 372 341 318 365 410 364 268 278 282 212 215 273 321 302 Ni 166 156 165 212 176 193 190 129 139 136 109 119 138 136 127 Cu 140 141 121 129 138 131 131 152 146 163 167 167152 152 175 177 Zn 94 89 103 88 100 91 90 98 99 105 109 10991 97 98 100 Rb 2.01 1.92 1.91 1.80 1.96 1.91 1.90 2.77 2.69 2.81 4.26 2.79 Sr 198 196 204 198 202 196 191 195 200 210 203 184 202 216 212 Y 2222222022222124252627 22 25 24 26 Zr 77 77 76 70 78 75 73 79 80 84 83 83 75 80 Nb 948 9.57 9.45 8.72 9.46 9.26 9.10 10.01 10.01 10.42 11.10 10.46 11 12 Mo 0.47 0.32 0.33 0.30 0.37 0.36 0.31 0.39 0.37 0.39 0.47 0.48 Cs 0.020 0.021 0.020 0.018 0.019 0.019 0.018 0.029 0.028 0.030 0.040 0.024 Ba 37.42 38.72 37.91 35.51 38.34 37.25 36.93 46.04 45.88 47.10 83 60.85 48.65 45 47 La 6.09 6.23 6.06 5.64 6.03 5.96 5.81 6.50 6.39 6.53 6.87 6.33 Ce 15.26 15.64 15.22 14.27 15.23 15.16 14.54 16.21 15.89 16.30 17.22 16.55 Pr 2.33 2.39 2.33 2.17 2.32 2.33 2.23 2.48 2.44 2.51 2.50 2.51 Nd 11.07 11.26 11.14 10.22 10.93 10.99 10.53 11.53 11.60 12.09 11.47 11.90 Sm 3.15 3.19 3.25 2.90 3.10 3.05 2.95 3.26 3.30 3.41 3.13 3.39 Eu 1.14 1.16 1.16 1.07 1.15 1.11 1.08 1.18 1.21 1.23 1.11 1.24 Tb 0.64 0.65 0.64 0.60 0.63 0.63 0.60 0.68 0.68 0.69 0.63 0.72 Gd 3.78 3.82 3.84 3.50 3.73 3.68 3.57 3.95 3.96 4.11 3.76 4.18 Dy 3.89 4.02 4.03 3.63 3.95 3.92 3.72 4.25 4.26 4.37 3.93 4.42 Ho 0.80 0.82 0.82 0.75 0.80 0.80 0.77 0.86 0.86 0.89 0.83 0.93 Er 2.21 2.25 2.20 2.05 2.26 2.18 2.10 2.43 2.44 2.44 2.25 2.50 Yb 2.14 2.15 2.14 1.96 2.13 2.08 2.08 2.25 2.37 2.33 2.11 2.40 Lu 0.31 0.34 0.33 0.31 0.32 0.32 0.31 0.37 0.36 0.36 0.32 0.38 Hf 1.99 2.05 2.08 1.85 2.05 1.99 1.90 2.09 2.11 2.16 2.19 2.29 Ta 0.54 0.55 0.55 0.51 0.53 0.52 0.50 0.57 0.61 0.58 0.73 4.72 Pb 0.359 0.365 0.360 0.334 0.372 0.342 0.337 0.416 0.429 0.421 0.688 0.459 Th 0.257 0.266 0.251 0.240 0.266 0.254 0.260 0.371 0.361 0.384 0.472 0.369 U 0.067 0.071 0.074 0.065 0.069 0.063 0.068 0.105 0.099 0.107 0.145 0.110

range is An87–63). These units also show zoning toward the edge of Ol we describe our methods and the results of our major and trace element grains, and also contain large chromite grains (Fig. 6B). Average compo- analyses. The Undirhlíðar samples are similar in composition to histori- sitions for olivines (from EDS analyses) for LpW are Fo81 (n = 48, range cal lavas from the Reykjanes Peninsula, whose petrogenesis has been is Fo87–67). Some dike samples show sub-ophitic to ophitic textures, in described in detail by previous workers (Gee et al., 1998; Jakobsson which plagioclase laths are partially to completely enveloped by augite. et al., 2000; Kokfelt et al., 2003; Koornneef et al., 2012; Peate et al., Two units that were not distinguished as olivine-bearing in the 2001, 2009; Thirlwall et al., 2004, 2006). Here, we focus on comparing field were found to contain olivine when examined by thin section compositional patterns to spatial and temporal relationships to better (Fig. 5 A–B). Dike 3 (Ld3) and the pillow unit that it feeds (Lp3) have understand the construction of the pillow ridge. sparse small (~0.25 mm) olivine microphenocrysts that were not observed in hand sample. 3.3.1. Analytical methods A total of 47 geochemical analyses were acquired, of which, 2 are 3.3. Geochemistry from the fragmental units, 13 are from the three dikes, and 32 from the pillow lavas (Table 1). Analyses were performed on fresh glass The goal of the geochemical analysis is to determine whether com- (6) and whole-rock (41) samples. Glasses were hand-picked for pieces positional variations correlate with stratigraphic relationships. Below, with unaltered surfaces. Whole-rock powders were prepared from the M. Pollock et al. / Lithos 200–201 (2014) 317–333 327 fresh crystalline interiors of dikes and lavas and crushed in alumina 2003; Koornneef et al., 2012; Peate et al., 2001, 2009; Thirlwall et al., ceramic grinding containers. Loss on ignition (LOI) was determined by 2004, 2006). The Undirhlíðar data span most of the regional range in heating the samples at 950 °C for 1 h following the methods of Boyd MgO but are generally constrained to the higher sides of the Na Oand 2 and Mertzman (1987). SiO2 fields and the lower side of the FeO* field. In the Western Volcanic Major and minor elements were measured for 36 samples at Duke Zone and on the Reykjanes Peninsula, major and trace element variations University by direct current plasma emission spectrometry (DCP; are correlated; depleted basalts have been characterized by lower FeO*

Fisons SpecterSpan 7) following the methods described in Klein et al. and higher SiO2 than enriched basalts (Shorttle and Maclennan, 2011), (1991). The remaining 11 samples were measured at The College of suggesting that the higher SiO2 and lower FeO* abundances in the Wooster by X-ray fluorescence spectrometry (XRF; 4 kW Rh-target Undirhlíðar data are related to mantle source composition. Rigaku ZSX Primus II) following methods modified after Boyd and Trace element variations also suggest the importance of mantle pro- Mertzman (1987) and Mertzman (2000). XRF analyses of major cesses in the generation of the Undirhlíðar samples. The Undirhlíðar element oxides were carried out on fused glass beads prepared by data define two trace element populations: (1) an incompatible combining 1.2500 ± 0.0001 g of sample with 8.7500 ± 0.0002 g of element-enriched (LaN/SmN ~1.6; Zr/Y ~3.5) group and (2) a more flux (1.2 Li2B4O7/2.2 LiBO2). Beads were fused in a Katanax K1 fluxer depleted (LaN/SmN ~1.3; Zr/Y ~3.2) group (Fig. 9A). The difference in that continuously agitated the sample and increased the temperature trace element ratios between these groups cannot be produced by to 1015 °C over 11.5 min. Minor elements (V, Cr, Ni, Cu, Zn, Sr, Y, Zr, shallow-level crystal fractionation and accumulation processes. Instead, Nb, Ba) were measured by XRF on pressed pellets that were prepared the range of trace element ratios can be generated by relatively simple by thoroughly mixing 8.8000 ± 0.0002 g of sample with 1.2000 ± models involving variations in mantle source compositions or mixing 0.0001 g of SpectroBlend binder and pressing the pellet on a Carver of variable degree melts. 3853 manual press at 20 tons for 1 min. Low-abundance trace elements It is well-established that the sub-Icelandic mantle is chemically het- for 34 of the 36 DCP samples were measured by inductively coupled erogeneous (Rudge et al., 2013; Shorttle and Maclennan, 2011; Thirlwall plasma mass spectroscopy (ICP-MS; VG PlasmaQuad 3) at Duke Univer- et al., 2004). Incompatible trace element ratios in Icelandic basalts are sity following a procedure described in Cheatham et al. (1993). positively correlated with isotopic ratios (e.g., McKenzie et al., 2004; Typical analytical runs consisted of unknowns, calibration standards, Shorttle and Maclennan, 2011; Stracke et al., 2003), suggesting that pat- and drift correcting solutions for the DCP and ICP-MS. The XRF was terns in trace element ratios may reflect differences in mantle source corrected for drift prior to each run. Standards for all techniques includ- composition. However, without isotopic data, we cannot evaluate the ed NIST-688 and USGS standards AGV-2, BIR-1, DNC-1a, and SDC-1. role of mantle heterogeneity in producing the Undirhlíðar sample suite. ICP-MS and DCP runs also included AII92-29-1 (a Mid-Atlantic A heterogeneous mantle source may not be necessary for Ridge standard) and USGS G-2 while XRF runs included USGS BCR-2, explaining the relationship between the Undirhlíðar trace element BHVO-1, GSP-2, and W-2. Replicate analyses of standards and selected groups (e.g., Koornneef et al., 2012; Slater et al., 2001). Sub-Icelandic samples as well as errors (1σ) are available in the electronic supplemen- melting processes are dynamic (Koornneef et al., 2012; Shorttle and tary material. A comparison of major element oxides measured by XRF Maclennan, 2011; Sinton et al., 2005) and incomplete mixing of partial and DCP reveals excellent agreement with the exception of P2O5 at melts can generate a wide range of trace element compositions (Slater low concentrations (Fig. 7). Fig. 7 also compares XRF, DCP, and ICP-MS et al., 2001). We find that the Undirhlíðar trace element data can be analyses of trace elements used in later figures. In general, trace generated by mixing of variable degree melts from a homogeneous elements show good agreement with the following exceptions: (1) Nb mantle. The melt trajectories shown in Fig. 9A were generated using measured by XRF at low concentrations; (2) Zr measured by XRF in the equation for instantaneous melts produced by non-modal dynamic SDC-1; and (3) Zr measured by DCP in most standards. Because of the melting (Langmuir et al., 1977; McKenzie, 1985; Wood, 1979; Zou, discrepancy in Zr and Nb concentrations, only values measured by 1998). Dynamic melting is different from perfect fractional melting ICP-MS are used for these elements in the following discussion. in that a small amount of melt is retained in the residue until it reaches a critical value for separation (Φ). The initial mantle composition 3.3.2. Major and trace elements is Workman and Hart's (2005) E-DMM and the critical value of Φ is Overall, the Undirhlíðar samples are tholeiitic basalts similar in 1.0%. Melts were generated in the spinel (Kinzler, 1997) and garnet major element composition to historical lavas from the Reykjanes (Walter, 1998) stability fields. Details of the melting model are provided Peninsula (Fig. 8; Gee et al., 1998; Jakobsson et al., 2000; Kokfelt et al., in the electronic supplementary material.

100 1000 A B

10 100

1 10

XRF (wt%) SiO2 TiO2 NIST-688 Al2O3 0.1 Fe2O3 XRF or DCP (ppm) 1 Y (XRF) AGV-2 MnO Zr (XRF) BIR-1 MgO CaO Nb (XRF) DNC-1a Na2O Y (DCP) SDC-1 K2O P2O5 Zr (DCP) 0.01 0.1 0.01 0.1 1 10 100 0.1 1 10 100 1000 DCP (wt%) ICP-MS (ppm)

Fig. 7. Comparison of standards analyzed as unknowns by XRF, DCP, and ICP-MS. Standards are distinguished by symbol; colors represent elements as detailed in the legends. Error bars are shown for symbols that are smaller than 2σ uncertainty in the analytical data. Gray lines show perfect correlations. Major element oxides (wt.%) are shown in A. With the exception of P2O5 for AGV-2 (certified value 0.48 ± 0.02 wt.%), there is excellent agreement between XRF and DCP standards. Trace elements (ppm) that are used in later figures are shown in B. The most notable discrepancies are Zr measured by DCP (gray symbols), Zr measured by XRF in SDC-1 (certified value 290 ± 30 ppm), and Nb measured by XRF at low concentrations. 328 M. Pollock et al. / Lithos 200–201 (2014) 317–333

Mixing of 1% melt from the garnet ﬁeld and 2.5% melt from the spinel A ﬁ 6 kbar eld can produce the pattern in trace element ratios shown in Fig. 9A. However, the chondrite-normalized REE patterns show that this mix-

ture overestimates the heavy REE for LaN/SmN ratios that approximate 3 kbar the two Undirhlíðar trace element groups (Fig. 9B and C), indicating that the mixture underestimates the role of melts from the garnet sta- fi – FeO* bility eld. Rather, the more enriched group (Lp1 2) and the more depleted group (Lp3, LpW, Ld) can be better approximated by mixing Cpx 2% garnet-field melt with 1% and 2% melt from the spinel field, respec- tively. The simple melt model is non-unique, but it provides an alterna- Ol tive explanation that does not require multiple mantle sources for Plag explaining the relationship between the Undirhlíðar trace element groups (e.g., Koornneef et al., 2012; Slater et al., 2001). It is likely that B both mantle source composition and incomplete mixing of partial melts play a role in generating the Undirhlíðar trace element groups. The two Undirhlíðar trace element groups can also be distinguished in major element data (Fig. 8). The trace element-depleted group is gen-

CaO erally higher in FeO* and Al2O3 and lower in CaO compared to the trace element-enriched group, and the difference in CaO is larger than analytical uncertainty (Fig. 8). These major element variations are consistent with the effects of fractional crystallization at different pressures. Petrolog Plag Cpx (v 3.1.1.3; Danyushevsky and Plechov, 2011)wasusedtomodelfraction- al crystallization of a primitive parent magma composition observed Ol within each group. Variation within the trace element-depleted samples (Ld, Lp3, LpW) can be explained by higher-pressure (6 kb, QFM-1) frac- C tional crystallization of an anhydrous parent magma with a composition

of S-10-07 (Mg# 54). For this group, an olivine cumulate (Fo84)fractionates until 8.88 wt.% MgO, when clinopyroxene (Mg# 83) joins the frac- 3

O tionating assemblage. At 8.65 wt.% MgO, plagioclase (An70) begins 2 – Al crystallizing. For the trace element-enriched samples (Lp1 2), major element variations are better approximated by fractional crystallization Plag of an anhydrous parent (E-11-05, Mg# 59) at lower pressures (3 kb,

QFM-1). Olivine (Fo84) fractionates until 8.34 wt.% MgO, when plagioclase (An ) joins the fractionating assemblage. At 8.25 wt.% MgO, Ol 74 clinopyroxene (Mg# 83) begins crystallizing. Cpx Scatter within the groups is consistent with observed petrological D variations (Fig. 8). For example, olivine-bearing units (Ld1–2, LpW) are offset from the other trace element-depleted samples (Ld3, Lp3) to higher MgO values along trajectories deﬁned by olivine accumulation

(Fig. 8). Assuming an olivine composition of Fo80 (see Section 3.2), the olivine-bearing units can be generated by accumulating ≤10% olivine. O

2 Overall, the Undirhlíðar major and trace element data record two

Na geochemical groups that appear to have separate magmatic histories. Lp1&2 LpW The trace element-enriched group (Lp 1–2) crystallized at shallower Plag TB Ld1&2 depths (~3 kb), evolving from a parent magma that may have been gen- Ol glass Lp3 erated by a greater proportion of melting in the garnet stability ﬁeld. Cpx Ld3 The trace element-depleted group (Ld, Lp3, LpW) crystallized at deeper depths (~6 kb), evolving from a chemically distinct parent magma that E may have had a greater proportion of melt from the spinel stability ﬁeld.

3.4. Spatial and Temporal Variations in Composition

2 The geochemical and petrological groups align with lithostratigraphic units. The core of the quarry consists of the lowermost pillow SiO units (Lp1–2), which are plagioclase-phyric and incompatible-element

Cpx Plag Ol Fig. 8. Major element compositions of Undirhlíðar dikes and lavas. A. FeO*, B. CaO, C. Al2O3,

D. Na2O, and E. SiO2 versus MgO, all in wt.%. Enclosed ﬁelds represent historic Reykjanes Peninsula lavas of Peate et al. (2009). Open circles indicate pillow units 1 and 2 (Lp1, 57911Lp2). Black symbols represent dike 3 (Ld3; squares) and pillow unit 3 (Lp3; circles). MgO (wt. %) Gray symbols are dikes 1 & 2 (Ld1 – 2; squares) and west wall pillows (LpW; circles). Tri- angles indicate tuff-breccia units (TB). Glasses are indicated with crosses. Arrows show the effects of the accumulation of olivine (ol), plagioclase (plag), and clinopyroxene (cpx). Error bars show 2σ uncertainty in analytical data. Lines show fractional crystallization models calculated using Petrolog (Danyushevsky and Plechov, 2011). See the text for model parameters. M. Pollock et al. / Lithos 200–201 (2014) 317–333 329

9 The stratigraphic relationships in Undirhlíðar also suggest a time 8 A sequence, where the incompatible element-enriched lavas near the base of the quarry (Lp1–2) are the oldest and the less-enriched 7 cross-cutting dikes (Ld1–3) and capping lavas (LpW, Lp3) are the 6 Garnet youngest. We cannot know for certain the timing of western dikes – 5 (Ld1 2) and the west wall pillows (LpW) with respect to the eastern dike (Ld3) and overlying pillows (Lp3) because we do not observe Zr/Y 4 them in direct contact with one another. However, the overlap in 3 Spinel trace element chemistry and presence of olivine suggests that these units are from the same magma batch and are likely to be closely 2 related in time. 1 Overall, the compositional difference between the younger and 0 older units requires processes beyond simple crystal accumulation or 0.0 0.5 1.0 1.5 2.0 2.5 3.0 fractionation, implying a shift in the magmatic system over time. Furthermore, this shift correlates with the TDP lithofacies association La /sm n n (vitric tuff-breccia cut by dikes that feed pillow lava ﬂows emplaced 100 immediately above the tuff-breccias), suggesting that changes in the B magmatic system are intimately linked to eruption dynamics. Below, we consider the spatial and temporal variations in lithofacies and composition. We propose a model for the construction of Undirhlíðar that has implications for the magmatic plumbing system and for the role of explosive volcanism in the construction of pillow-dominated glaciovolcanic ridges.

4. Discussion

4.1. Sequence of eruptive events Lp1 & 2 La /Sm ~1.6 Chondrite Normalized Values N N The glaciovolcanic deposits exposed in Undirhlíðar reflect a complex 10 sequence of events that include multiple eruptive and intrusive phases. La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lu The combined lithostratigraphic, petrographic, and geochemical relationships have been used to formulate a schematic model for the forma- 100 tion of Undirhlíðar described below (Fig. 10): C (1) An initial effusive phase represented by units Lp1 and Lp2. Pauses in the eruptive event may have occurred, as demonstrated by the stratified layers of vitric tuff (T) that separate the pillow units (Fig. 10A). Locally advancing fronts of pillow mounds oversteepened, producing pillows with higher plunges interstratified with TB; during more rapid advance, entire mound fronts collapsed to produce massive TB units with only fragments of pillow lavas. The bulk of the core of the subglacial ridge was built during this phase. These lavas are incompatible element- enriched and generally olivine-free (Fig. 10B). Lp3, LpW, Ld 1-3

Chondrite Normalized Values (2) Intrusions either of more H O-rich magma or of lavas into a LaN/SmN ~1.3 2 shallower englacial lake resulted in subaqueous Strombolian 10 eruptions that generated the lenses of tuff-breccia (TB2/LT) on La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Yb Lu the eastern side of the ridge (Fig. 10C). (3) A return to effusive volcanic activity (represented by Ld1-2 and Fig. 9. A. Zr/Y versus LaN/SmN for Undirhlíðar samples measured by ICP-MS. Colors and symbols as in Fig. 8. Solid lines show instantaneous melts from dynamic melting LpW) on the west side of the quarry that intruded the initial (Langmuir et al., 1977; McKenzie, 1985; Wood, 1979; Zou, 1998) of EDMM (Workman pillows and erupted pillow lavas that draped over the edge of and Hart, 2005) in the spinel (thick line; Kinzler, 1997) and garnet (thin line; Walter, the existing pillow edifice. This effusive phase is distinguished fi 1998)stability elds. Thick dashed line shows mixing between 2.5% spinel melt and from the first as a different batch of less-enriched magma bearing 1% garnet melt. Thin dashed lines show mixing of 2% garnet melt with 1% spinel melt, shown in B, and 2% spinel melt, shown in C. See the text for discussion and electronic large olivine phenocrysts. supplementary material for model parameters. B–C. Chondrite normalized rare earth (4) A final effusive phase on the east side of the quarry that intruded elements for Undirhlíðar enriched (B; Lp1–2) and depleted (C; Lp3, LpW, Ld) groups. Chon- (Ld3) into TB1 and produced a capping layer of pillow lavas drite values from Sun and McDonough (1989). Dashed lines show mixing between melts (Lp3). This effusive phase is genetically related to the second fi from the garnet and spinel stability eldsshowninA. olivine-bearing effusive phase but does not contain large olivine phenocrysts (Fig. 10D). enriched. The dikes (Ld1–3), west wall pillows (LpW), and the upper- Our model for the formation of Undirhlíðar adds to the observations most pillow unit (Lp3) comprise the olivine-phyric, less-enriched of previous workers (e.g., Edwards et al., 2009; Höskuldsson et al., 2006; group. This less-enriched group can be further differentiated based on Jones, 1969, 1970; Moore et al., 1995) by documenting three important mineralogy (Fig. 2). The distinctly olivine-bearing units (LpW, Ld1–3) relationships: (1) repeated eruptive episodes ranging in style from effu- with cm-scale olivine phenocrysts are on the western side of Undirhlíðar sive to more explosive; (2) a specific stratigraphic sequence of vitric while the units on the eastern side contain olivine as microphenocrysts tuff-breccia cut by dikes that feed pillow lava flows emplaced immedi- (Ld3, Lp3). ately above the tuff-breccias (TDP lithofacies association), which occurs 330 M. Pollock et al. / Lithos 200–201 (2014) 317–333

dominated eruption of pillow lava mounds under relatively thick ice [no evidence for explosivity]

Phase 2 (Lp2/TB): intrusions feed new pillows; some areas collapse to form deposits of

topography

Phase 3 (Ld/LT): new magma

fountaining) producing LT with vitric bombs.

Phase 4 (Lp3/LpW): eruptions switches to dominantly

Fig. 10. Schematic diagrams illustrating possible sequential emplacement of lithofacies found at Undirhlíðar. A. Phase 1 is dominated by effusive eruption of pillow lava ﬂows that formed adjacent mounds. B. In Phase 2, tuff-breccia becomes more prominent as steepening mounds result in gravity-driven collapse. C. Phase 3 is delineated by explosive onset and marks shift to Ol-phyric, incompatible element-depleted lava. D. Phase 4 returns to production of effusion-dominated pillow lavas.

when effusive eruptions are initiated by subaqueous Strombolian 4.2. Implications for the role of explosive volcanism in pillow-dominated activity; and (3) the chemical transition from olivine-free, incompatible subglacial eruptions element-enriched rocks to olivine-bearing, less-enriched rocks that coincides with the TDP lithofacies association. Many of the original models for glaciovolcanic (subglacial) eruptions The results of this study differ from those of Jones (1970) and reported an ordered stratigraphy of basal pillow lavas, followed by tephra Höskuldsson et al. (2006) in that we see evidence for the emplacement (hyaloclastite + pillow lava breccia), and capped by subaerial lava flows of multiple episodes of pillow lavas, as was found at Pillow Ridge, in (e.g. Jones, 1969, 1970; Mathews, 1947; Smellie and Skilling, 1994). These British Columbia (Edwards et al., 2009). While difficult to assess with lithofacies were interpreted as representing an effusive eruption onset, certainty, our ability to document more complexity at Undirhlíðar followed by slowly building explosive eruptions as water depths de- may simply be a result of the exceptional exposures resulting from creased, and ending with subaerial lava once the tephra pile was above quarrying. Presumably the Kalfstindar area studied by Jones (1969, the level of the enclosing ice-confined lake. In particular, the origins of 1970) as well as the Kverkfjöll area studied by Höskuldsson et al. the fragmental deposits (tuff, lapilli tuff and pillow breccia) were very (2006) formed in the late Pleistocene, so that erosion has not produced speculative or nebulous; many were viewed mainly as products of dom- the same dissection seen in the quarries, or at Pillow Ridge, which has inantly quench fragmentation and the onset and importance of magmatic an estimated age of 0.9 Ma (Souther, 1992) and has been heavily fragmentation appear to have been underappreciated (e.g., Edwards and dissected by multiple periods of glaciation (Edwards et al., 2009). Russell, 2011; Owen et al., 2013; Russell et al., 2013). While Kalfstindar In addition, the formation of Undirhlíðar may be more akin to the appears to represent one end-member style of eruption, with a steady shorter duration (e.g., weeks to months), smaller volume (b1km3) fis- progression from effusion of pillow lavas transitioning to production of sure eruptions that produce small tindars, as opposed to longer duration fragmental units, the ridges in the Kverkfjöll area appear to represent (e.g., years to decades), larger volume (N1km3)eruptionsfromfissures the opposite extreme, being made exclusively of effusive pillow lavas and central craters that produce large tindars or tuyas (Jakobsson and (Höskuldsson et al., 2006). The deposits at Undirhlíðar appear to show a Johnson, 2012). middle-ground, where pillow effusion is dominant, but where production M. Pollock et al. / Lithos 200–201 (2014) 317–333 331 of explosive units also occurs periodically. It is also important to note that Geochemical and geophysical studies of lava flows and lower crustal further south of Undirhlíðar along the ridge, much more extensive de- rocks in Iceland (Eason and Sinton, 2009; Maclennan et al., 2003)sug- posits dominated by tuff and tuff-breccia have been documented gest that melt may reside at various depths in the crust in an intercon- (Saemundsson, pers. comm., 2013; Mercurio, 2011). Thus, Undirhlíðar nected network of smaller melt bodies. The morphology of Undirhlíðar may also represent smaller volume, more distal magmatism toward the suggests that it may have been constructed during a short-duration, end of the active fissure segment. small-volume event (Jakobsson and Johnson, 2012). Furthermore, In comparison to Antarctic subglacial volcanics, the closest structure geochemical and morphological studies of eruptive units along the that Undirhlíðar resembles is a volcanic structure represented by Mount length of rift segments in Iceland suggest that magma supply is Pinafore. Smellie and Skilling (1994) identified multiple volcanic most robust at segment centers, weakening toward segment ends (lava and hyaloclastite) and volcaniclastic (conglomerate and gravelly (e.g., Gudmundsson, 1995). Undirhlíðar's location at the northern sandstone) lithofacies. They interpreted the repetition of lithofacies at end of the Krisuvik fissure system, combined with the small volume, Mount Pinafore to represent multiple effusive events erupted under a suggests the lack of a robust shallow crustal melt reservoir, which valley-confined thin (100–150 m) ice cover, each preceded by deposition may have enhanced a network of smaller melt bodies. Therefore, it of volcaniclastic beds deposited from flowing meltwater. Unlike the may be reasonable to assume that Undirhlíðar may not have been un- Mount Pinafore exposure, however, Undirhlíðar does not show any evi- derlain by a magmatically robust and sustained shallow magma cham- dence for being valley-confined (ponded lava flows) or for large meltwa- ber, but by a network of smaller magma bodies instead (e.g., Eason and ter events (thick, continuous units of bedded volcaniclastic sediments). Sinton, 2009). The specific stratigraphic sequence that we have observed in the In the proposed scenario (Fig. 11), the Undirhlíðar sequence is Undirhlíðar deposits of vitric tuff-breccia cut by dikes that feed pillow erupted from discrete melt bodies that experience separate evolutionary lava flows emplaced immediately above the tuff-breccias (TDP lithofacies histories. The initial effusive event taps olivine-free, incompatible- association) indicates an important role of magmatic fragmentation and element enriched magma from one melt lens at shallow depth (~3 kb). an explosive phase that occurs at the onset of effusive phases. Head and This is followed by a second effusive event originating from a separate Wilson (2003) argued that even submarine eruptions in relatively deep melt lens at greater depth (~6 kb), containing less enriched, olivine- (e.g., 2 km) water could produce Strombolian activity. Our TB2/LT units bearing magma. The difference in composition between the two melt are consistent with such styles of activity, albeit in water to have been bodies could have resulted from differences in the extent of mantle melt- much shallower. While we at present have no direct data to constrain ing or differences in mantle source composition. water depths during the glaciovolcanic eruptions, previous work in the This scenario does not explain the cause of the explosive event that Reykjanes Peninsula glacial ice thicknesses are consistent with ice thick- produced the TDP lithofacies association, but it does link eruptive style nesses of ≥0.5 km (Mercurio, 2011; Schopka et al., 2006). to the magmatic system. The explosive phase may have been initiated by a new pulse of olivine-rich mantle-derived melt or, more likely, by 4.3. Implications for magmatic plumbing variables related to ice conditions, such as water drainage or thinning of the ice (Bennett et al., 2009), which led to enhanced vesiculation in Any model for the emplacement of the Undirhlíðar sequence must the deeper melt reservoirs and the eruption of olivine-rich lavas. explain the spatial distribution of olivine-free, incompatible-element enriched lavas under olivine-bearing lavas that are less enriched in 5. Conclusions incompatible elements. The model should also address the correlation of the geochemical transition with the TDP lithofacies association. Lithostratigraphic, mineralogical, and geochemical variations in Although we cannot constrain the cause of the explosive phase, nor Undirhlíðar quarry on the Reykjanes Peninsula in southwest Iceland the timing between eruptive events, the model should account for mul- provide a better understanding of the formation of pillow-dominated tiple eruptive phases during which magmatic and ice cover conditions glaciovolcanic tindars and the factors that control glaciovolcanic lithol- may have varied (Bennett et al., 2009; Smellie and Skilling, 1994). ogies. The quarry walls primarily consist of pillow lavas, but also include Given these constraints and previously documented chemical varia- dikes and fragmental lithofacies (tuff, tuff-breccia), indicating repeated tions in subglacial pillow ridges (e.g., Jakobsson and Johnson, 2012; eruptive events varying in eruptive style and location. In particular, the Moore and Calk, 1991; Moore et al., 1995) and Icelandic fissure and TDP lithofacies association shows that Strombolian activity can precede shield eruptions (e.g., Eason and Sinton, 2009; Maclennan et al., 2003), effusive events, suggesting that explosive volcanism may play an impor- we propose a model for the emplacement of two chemically and miner- tant role in the construction of pillow-dominated edifices. Composition- alogically distinct lavas in Undirhlíðar: the sequential eruption of ally, the Undirhlíðar units can be divided into two groups: (1) older, magmas from a network of smaller, chemically distinct, mid-crustal plagioclase-phyric, incompatible element-enriched lavas that comprise magma bodies (Fig. 11). the core of the quarry, and (2) younger, olivine-phyric, less-enriched

Stage 1 ICE Stage 2 ICE

CRUST CRUST

Crystallization at ~3-6 kbar

Fig. 11. Schematic model of possible plumbing systems beneath Undirhlíðar. Stage 1 represents the initial construction phase in which the incompatible element-enriched, olivine-free basal pillow units (gray) are erupted. Stage 2 represents the intrusion and eruption of olivine-bearing, less incompatible element-enriched magma. An interconnected network of smaller, physically and chemically distinct magma bodies supplies melts to Undirhlíðar, erupting from a shallower magma body (~3 kb) in Stage 1 and a separate, deeper magma body (~6 kb) in Stage 2. 332 M. Pollock et al. / Lithos 200–201 (2014) 317–333 dikes and capping lavas. The two groups require differences in mantle Farrell, R.E., Andrews, G.D.M., Russell, J.K., Anderson, R.G., 2007. Chasm and Dog Creek lithofacies, Chilcotin Group basalt, Bonaparte Lake map area. British Columbia, processes; this shift in the magmatic system coincides with the TDP Geological Survey Canada Current Research A5, 1–11. lithofacies association (tuff–dike–pillow), suggesting that the magmatic Ferrini, V.L., Tivey, M.K., Carbotte, S.M., Martinez, F., Roman, C., 2008. Variable morphologic system is intimately linked with eruptive dynamics. We propose that expression of volcanic, tectonic, and hydrothermal processes at six hydrothermal vent fields in the Lau back-arc basin. Geochemistry, Geophysics, Geosystems 9 (7), Q07022. the magmas emplaced in Undirhlíðar were tapped from small, separate Furnes, H., de Wit, M., Staudigel, H., Rosing, M., Muehlenbachs, K., 2007. A vestige of crustal melt reservoirs and that the explosive event was likely caused by Earth's oldest ophiolite. Science 315, 1704–1707. http://dx.doi.org/10.1126/science. a complex interaction between variations in ice and magmatic condi- 1139170. tions. Our observed evidence for multiple eruptive events, explosive Gee, M.A.M., Thirlwall, M.F., Taylor, R.N., Lowry, D., Murton, B.J., Bohrson, W.A., 1998. Crustal processes: major controls on Reykjanes Peninsula lava chemistry, SW volcanism, and transitions in the magmatic system differ from previous Iceland. Journal of Petrology 39, 819–839. studies of tindars and tuyas, yielding new insights into the dynamics of Gregg, T.K.P., Fink, J.H., 1995. Quantification of submarine lava-flow morphology through – pillow-dominated eruptions in glaciovolcanic and other subaqueous analog experiments. Geology 23, 73 76. Gregg, T.K.P., Fink, J.H., 2000. A laboratory investigation into the effects of slope on lava environments. flow morphology. Journal of Volcanology and Geothermal Research 96, 145–159. Gudmundsson, A., 1995. Infrastructure and mechanics of volcanic systems in Iceland. Journal of Volcanology and Geothermal Research 64, 1–22. Acknowledgments Head, J.W., Wilson, L., 2003. Deep submarine pyroclastic eruptions: theory and predicted landforms and deposits. Journal of Volcanology and Geothermal Research 121 (3–4), 155–193. This work was supported by The College of Wooster Luce Fund, NSF Hickson, C.J., 2000. Physical controls and resulting morphological forms of Quaternary EAR 0958928 (MP), NSF EAR 1220176 (MP), the Dickinson College ice-contact volcanoes in Western Canada. Geomorphology 32, 239–261. Research and Development Committee, and NSF EAR 1220403 (BE). We Hickson, C.J., Moore, J.G., Calk, L.C., Metcalfe, P., 1995. Intraglacial volcanism in the Wells Gray-Clearwater Volcanic Field, East-central British Columbia, Canada. Canadian thank B. and D. Tewksbury for help with high-resolution photos and G. Journal of Earth Science 32, 838–851. Dwyer for ICP-MS and DCP analyses. Thanks to GAIN, A. Hiatt, M. Reinthal, Höskuldsson, A., Sparks, R.S.J., Carroll, M.R., 2006. Constraints on the dynamics of and J. Franceschi for help with manuscript preparation. For their assis- subglacial basalt eruptions from geological and geochemical observations at – tance in the field,wethankT.Louvain,T.Spillman,R.Lydell,A.Samale, Kverkfjöll, NE-Iceland. Bulletin of Volcanology 68, 689 701. Jakobsson, S.P., Gudmundsson, M.T., 2008. Subglacial and intraglacial volcanic formations R. Rossi, J. Haklar, and E. Was. This paper benefited from inspiring discus- in Iceland. Jökull 58, 179–196. sions with K. Saemundsson and M.T. Gudmundsson and excellent sug- Jakobsson, S.P., Johnson, G.L., 2012. Intraglacial volcanism in the Western Volcanic Zone, – gestions from two anonymous reviewers. Iceland. Bulletin of Volcanology 74, 1141 1160. http://dx.doi.org/10.1007/s00445- 012-0589-x. Jakobsson, S.P., Jónsson, J., Shido, F., 1978. Petrology of the western Reykjanes Peninsula, Iceland. Journal of Petrology 19 (4), 669–705. Appendix A. Supplementary data Jakobsson, S.P., Johnson, G.L., Moore, J.G., 2000. A structural and geochemical study of the Western Volcanic Zone, Iceland: preliminary results. InterRidge News 9, – Supplementary data to this article can be found online at http://dx. 27 33. Jarosch, A., Gudmundsson, M.T., Högnadóttir, T., Axelsson, G., 2008. Progressive cooling doi.org/10.1016/j.lithos.2014.04.023. of the hyaloclastite ridge at Gjálp, Iceland, 1996–2005. Journal of Volcanology and Geothermal Research 170 (3–4), 218–229. Johns, S.M., Helmstaedt, H.H., Kyser, T.K., 2006. Paleoproterozoic submarine intrabasinal References rifting, Baffin Island, Nunavut, Canada: volcanic structure and geochemistry of the Bravo Lake Formation. Canadian Journal of Earth Sciences 43 (5), 593–616. Alcorn, R., Pollock, M., Edwards, B.R., 2010. A three dimensional exploration of subglacial Johnson, G.L., Jakobsson, S.P., 1985. Structure and petrology of the Reykjanes Ridge between pillow eruptions on the Sveifluhals Ridge in southwest Iceland. GSA Abstracts with 62°55′N and 63°48′N (Atlantic). Journal of Geophysical Research 90, 10073–10083. Programs 42 (5), 294. Jones, J.G., 1969. Intraglacial volcanoes of the Laugarvatn Region, Southwest Iceland, I. Allen, C.C., 1980. Icelandic subglacial volcanism: thermal and physical studies. Journal of Journal of the Geological Society of London 124, 197–211. Geology 88, 108–117. Jones, J.G., 1970. Intraglacial volcanoes of the Laugarvatn Region, Southwest Iceland, II. Allen, C.C., Jercinovic, M.J., Allen, J.S.B., 1982. Subglacial volcanism in North-Central British Journal of Geology 78, 127–140. Columbia and Iceland. Journal of Geology 90, 699–715. Kinzler, R.J., 1997. Melting of mantle peridotite at pressures approaching the spinel to Batiza, R., White, J.D., 2000. Submarine lavas and hyaloclastite. Encyclopedia of Volcanoes garnet transition: application to mid-ocean ridge basalt petrogenesis. Journal of 361–381. Geophysical Research: Solid Earth 102, 853–874. Bennett, M.R., Huddart, D., Gonzalez, S., 2009. Glaciovolcanic landsystems and large-scale Klein, E.M., Langmuir, C.H., Staudigel, H., 1991. Geochemistry of basalts from the south- glaciotectonic deformation along the Brekknafjöll-Jarlhettur, Iceland. Quaternary east Indian Ridge, 115 Degrees E-138 Degrees E. Journal of Geophysicl Research 96, Glaciodynamics 28, 647–676. 2089–2107. Bowman, L., Pollock, M., Edwards, B.R., Alcorn, R., 2011. Geochemical and field relation- Kokfelt, T.F., Hoernle, K., Hauff, F., 2003. Upwelling and melting of the Iceland plume from ships of pillow and dike units in a subglacial pillow ridge, Undirhlithar quarry, south- radial variation of 238U–230Th disequilibria in postglacial volcanic rocks. Earth and west Iceland. GSA Abstracts with Programs 43 (5), 102. Planetary Science Letters 213, 167–186. Boyd, F.R., Mertzman, S.A., 1987. Composition and structure of the Kaapvaal Lithosphere, Koornneef, J.M., Stracke, A., Bourdon, B., Meier, M.A., Jochum, K.P., Stoll, B., Grönvold, K., Southern Africa. Special Publications Geochemical Society 1, 13–24. 2012. Melting of a two-component source beneath Iceland. Journal of Petrology 53, Cheatham, M.M., Sangrey, W.F., White, W.M., 1993. Sources of error in external calibra- 127–157. tion ICP-MS analysis of geological samples and an improved non-linear drift correc- Lambeck, K., Purcell, A., Funder, S., Kjaer, K.H., Larsen, E., Moller, P., 2006. Constraints on tion procedure. Spectrochim Acta Part B: Atomic Spectroscopy 48, 487–506. the Late Saalian to early Middle Weichselian ice sheet of Eurasia from field data Danyushevsky, L.V., Plechov, P., 2011. Petrolog3: integrated software for modeling and rebound modeling. Boreas 35, 539–575. crystallization processes. Geochemistry, Geophysics, Geosystems 12, Q07021. Langmuir, C.H., Bender, J.F., Bence, A.E., Hanson, G.N., Taylor, S.R., 1977. Petrogenesis of http://dx.doi.org/10.1029/2011GC003516. basalts from the FAMOUS area Mid-Atlantic Ridge. Earth and Planetary Science Eason, D.E., Sinton, J.M., 2009. Lava shields and fissure eruptions of the western volcanic Letters 36, 133–156. zone, Iceland: evidence for magma chambers and crustal interaction. Journal of Loughlin, S.C., 2002. Facies analysis of proximal subglacial and proglacial volcaniclastic Volcanology and Geothermal Research 186, 331–348. successions at the Eyjafjallajökull central volcano, southern Iceland. Geological Society Edwards, B.R., Russell, J.K. (Eds.), 2011. Thermodynamic constraints on explosive vs. effu- Special Publication 202, 149–178. sive onset of glaciovolcanic eruptions, Vol 1. EOS Transactions, p. 2578 (AGU V31E- Maclennan, J., McKenzie, D., Hilton, F., Gronvöld, K., Shimizu, N., 2003. Geochemical 2578, Fall Meet. Suppl.). variability in a single flow from northern Iceland. Journal of Geophysical Research Edwards, B.R., Skilling, I.P., Cameron, B., Haynes, C., Lloyd, A., Hungerford, J.H.D., 2009. 108. http://dx.doi.org/10.1029/2000JB000142 (ECV 4-1-ECV 4–21). Evolution of an englacial volcanic ridge: Pillow Ridge tindar, Mount Edziza volcanic Mathews, W.H., 1947. ‘Tuyas’, flat-topped volcanoes in Northern British Columbia. complex, NCVP, British Columbia, Canada. Journal of Volcanology and Geothermal American Journal of Science 245, 560–570. Research 185 (4), 251–275. McClintock, M., White, J.D.L., Houghton, B.F., Skilling, I.P., 2008. Physical volcanology of a Edwards, B.R., Ryane, C., Russell, J.K., Lasher, G.K., Dunnington, G., 2010. Glaciovolcanism large crater-complex formed during the initial stages of Karoo flood basalt volcanism, and episodic ice sheets: evidence for paleoclimate proxies and insights into eruption Sterkspruit, Eastern Cape, South Africa. Journal of Volcanology and Geothermal dynamics from the Kawdy-Tuya area of northern British Columbia. Geophysical Re- Research 172 (1–2), 93–111. search Abstracts 12 EGU2010-8822. McKenzie, D., 1985. 230Th–238U disequilibrium and the melting processes beneath ridge Edwards, B.R., Magnusson, E., Thordarson, T., Gudmundsson, M.T., Höskduldsson, A., axes. Earth and Planetary Science Letters 72, 149–157. Oddson, B., Haklar, J., 2012. Interactions between lava and snow/ice during the McKenzie, D., Stracke, A., Blichert‐Toft, J., Albaréde, F., Grönvold, K., O’Nions, R.K., 2004. 2010 Fimmvörðuháls eruption, south-central Iceland. Journal of Geophysical Source enrichment processes responsible for isotopic anomalies in oceanic island ba- – Research 117, B04302. http://dx.doi.org/10.1029/2011JB008985. salts. Geochim. Cosmochim. Acta 68 (12), 2699 2724. M. Pollock et al. / Lithos 200–201 (2014) 317–333 333

Mercurio, E.C., 2011. Processes, Products and Depositional Environments of Ice-confined Smellie, J.L., 2008. Basaltic subglacial sheet-like sequences: evidence for two types Basaltic Fissure Eruptions: A Case Study of the Sveifluháls Volcanic Complex. University with different implications for the inferred thickness of associated ice. Earth-Science of Pittsburgh, SW Iceland (Doctoral Dissertation). Reviews 88 (1–2), 60–88. Mertzman, S.A., 2000. K–Ar results from the southern Oregon–northern California Smellie, J.L., Skilling, I.P., 1994. Products of subglacial volcanic eruptions under different Cascade Range. Oregon Geology 62, 99–122. ice thicknesses: two examples from Antarctica. Sedimentary Geology 91, 115–129. Moore, J.G., Calk, L.C., 1991. Degassing and differentiation in subglacial volcanoes, Iceland. Smellie, J.L., Johnson, J.S., McIntosh, W.C., Esser, R., Gudmundsson, M.T., Hambrey, M.J., Journal of Volcanology and Geothermal Research 46, 157–180. van Wyk de Vries, B., 2008. Six million years of glacial history recorded in volcanic Moore, J.G., Hickson, C.H., Calk, L.C., 1995. Tholeiitic–alkalic transition at subglacial lithofacies of the James Ross Island Volcanic Group, Antarctic Peninsula. volcanoes, Tuya Region, British Columbia, Canada. Journal of Geophysical Research Palaeogeography Palaeoclimatology Palaeoecology 260, 122–148. 100, 24577–24592. Soule, S.A., Fornari, D.J., Perfit, M.R., Rubin, K.H., 2007. New insights into mid-ocean ridge Owen, J., Tuffen, H., McGarvie, D.W., 2013. Explosive subglacial rhyolitic eruptions in volcanic processes from the 2005–2006 eruption of the East Pacific Rise, 9°46′N–

Iceland are fuelled by high magmatic H2O and closed-system degassing. Geology 41 9°56′N. Geology 35 (12), 1079–1082. (2), 251–254. Souther, J.G., 1992. The late Cenozoic Mount Edziza volcanic complex. Geological Society Peate, D.W., Hawkesworth, C.J., van Calsteren, P.W., Taylor, R.N., Murton, B.J., 2001. of Canada Memoir 420, 320. 238U–230Th constraints on Mantle upwelling and plume-ridge interaction along the Stracke, A., Bizimis, M., Salters, V.J.M., 2003. Recycling oceanic crust: quantitative con- Reykjanes Ridge. Earth and Planetary Science Letters 18, 259–272. straints. Geochemistry, Geophysics, Geosystems 4 (3), 8003. http://dx.doi.org/10. Peate, D.W., Baker, J.A., Jakobsson, S.P., Waight, T.E., Kent, A.J.R., Grassineau, N.V., 1029/2001GC000223. Skovgaard, A.C., 2009. Historic magmatism on the Reykjanes Peninsula, Iceland: Sun, S.-s., McDonough, W.F., 1989. Chemical and isotopic systematics of oceanic basalts: a snap-shot of melt generation at a ridge segment. Contributions to Mineralogy and implications for mantle composition and processes. Geological Society of London, Petrology 157, 359–382. Special Publication 42, 313–345. Rudge, J.F., Maclennan, J., Stracke, A., 2013. The geochemical consequences of mixing melts Thirlwall, M.F., Gee, M.A.M., Taylor, R.N., Murton, B.J., 2004. Mantle components in Iceland from a heterogeneous mantle. Geochimica et Cosmochimica Acta 114, 112–143. and adjacent ridges investigated using double-spike Pb isotope ratios. Geochimicia et Russell,J.K.,Edwards,B.R.,Porritt,L.A.,2013.Pyroclastic passage zones in glaciovolcanic se- Cosmochimica Acta 68, 361–386. quences. Nature Communications 4. http://dx.doi.org/10.1038/ncomms2829 (1788 m). Thirlwall, M.F., Gee, M.A.M., Lowry, D., Mattey, D.P., Murton, B.J., Taylor, R.N., 2006. Low Saemundsson, K., 1967. Vulkanismus und tektonik des Hengill-Gebietes in Südvwest- delta 18O in the Icelandic mantle and its origins: evidence from Reykjanes Ridge Island. Acta Naturalia Islandica 11, 1–105. and Icelandic lavas. Geochimica et Cosmochimica Acta 70, 993–1019. Saemundsson, K., Einarsson, S., 1980. Geological map of Iceland, sheet 3. , SW-Iceland. Tominaga, M., Teagle, D.A.H., Alt, J.C., Umino, S., 2009. Determination of the volcano- Museum of Natural History and the Iceland Geodetic Survey, Reykjavik. stratigraphy of oceanic crust formed at superfast spreading ridge: electrofacies Saemundsson, K., Jóhanneson, H., Hjartarson, Á., Kristinsson, S.G., 2010. Geological map of analyses of ODP/IODP Hole 1256D. Geochemistry, Geophysics, Geosystems 10 (1), southwest Iceland, 1:100,000. Iceland GeoSurvey. Q01003. Schopka, H.H., Gudmundsson, M.T., Tuffen, H., 2006. The formation of Helgafell, south- Van Bemmelen, R.W., Rutten, M.G., 1955. Table mountains of Northern Iceland west Iceland, a monogenetic subglacial hyaloclastite ridge: sedimentology, hydrology (and Related Geological Notes). , Brill Archive. and volcano–ice interaction. Journal of Volcanology and Geothermal Research 152 Walker, G.P.L., 1992. Morphometric study of pillow-size spectrum among pillow lavas. (3–4), 359–377. Bulletin of Volcanology 54 (6), 459–474. Shorttle, O., Maclennan, J., 2011. Compositional trends of Icelandic basalts: implications Walter, M.J., 1998. Melting of garnet peridotite and the origin of komatiite and depleted for short-length scale lithological heterogeneity in mantle plumes. Geochemistry, lithosphere. Journal of Petrology 39, 29–60. Geophysics, Geosystems 12 (11). http://dx.doi.org/10.1029/2011GC003748. Werner, R., Schmincke, H.-U., 1999. Englacial vs lacustrine origin of volcanic table moun- Sigvaldason, G.E., 1968. Structure and products of subaquatic volcanoes in Iceland. Contri- tains: evidence from Iceland. Bulletin of Volcanology 60, 335–354. butions to Mineralogy and Petrology 18, 1–16. Werner, R., Schmincke, H.-U., Sigvaldason, G., 1996. A new model for the evolution of Sinton, J., Grönvold, K., Sæmundsson, K., 2005. Postglacial eruptive history of the Western table mountains: volcanological and petrological evidence from Herdubreid and Volcanic Zone, Iceland. Geochemistry, Geophysics, Geosystems 6. http://dx.doi.org/ Herdubreidartögl volcanoes (Iceland). Geologische Rundschau 85 (2), 390–397. 10.1029/2005GC001021. Wood, D.A., 1979. Dynamic partial melting: its application to the petrogenesis of basalts Skilling, I.P., 1994. Evolution of an englacial volcano: Brown Bluff, Antarctica. Bulletin of erupted in Iceland, the Faeroe Islands, the Isle of Skye (Scotland) and the Troodos Volcanology 56 (6–7), 573–591. Massif (Cyprus). Geochimica et Cosmochimica Acta 43, 1031–1046. Skilling, I.P., 2009. Subglacial to emergent basaltic volcanism at Hlöðufell, south-west Workman, R.K., Hart, S.R., 2005. Major and trace element composition of the depleted Iceland: A history of ice-conﬁnement. Volcano–Ice Interactions on Earth and Mars: MORB mantle (DMM). Earth and Planetary Science Letters 231, 53–72. the state of the science 4, 276–289. Zou, H.B., 1998. Trace element fractionation during modal and non-modal dynamic melting Slater, L., McKenzie, D., Grönvold, K., Shimizu, N., 2001. Melt generation and movement and open-system melting: a mathematical treatment. Geochimica et Cosmochimica beneath Theistareykir, NE Iceland. Journal of Petrology 42, 321–354. Acta 62, 1937–1945. Smellie, J.L., 2001. Lithostratigraphy and volcanic evolution of Deception Island, South Shetland Islands. Antarctic Science 13 (2), 188–209.