Lithos 314–315 (2018) 307–322

Contents lists available at ScienceDirect

Lithos

j ourna l homepage: www.e lsev ie r .com/ locate / l i thos

Permian to quaternary magmatism beneath the Mt Carmel area, Israel:Zircons from volcanic rocks and associated alluvial deposits

W.L. Griffin a,⁎, S.E.M. Gain a, J.-X. Huang a, E.A. Belousova a, V. Toledo b, S.Y. O'Reilly a

a ARC Centre of Excellence for Core to Crust Fluid Systems, Dept. Earth and Planetary Sciences, Macquarie University, NSW 2109, Australiab Shefa Yamim (A.T.M.) Ltd., Netanya 4210602, Israel

⁎ Corresponding author.E-mail address: bill.griffin@mq.edu.au (W.L. Griffin).

https://doi.org/10.1016/j.lithos.2018.06.0070024-4937/© 2018 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 30 March 2018Accepted 4 June 2018Available online 18 June 2018

Xenocrystic zircons from Cretaceous pyroclastic vents on Mt. Carmel, N. Israel, document two major periods ofearlier mafic magmatism: Permo-Triassic (285–220 Ma) and Jurassic (200–160 Ma). Related alluvial depositsalso contain these zircon populations. However, most alluvial zircons are Cretaceous (118–80 Ma) or younger,derived from Miocene to Pliocene volcanic episodes. The Permo-Triassic-Jurassic zircons are typically large andglassy; they have irregular shapes and a wide variety of internal zoning patterns. They appear to have grownin the interstitial spaces of coarse-grained rocks; many show evidence of recrystallization, including brecciationand rehealing by chemically similar zircon. Grainswith relict igneous zoning havemantle-like δ18O (5.5± 1.0‰),but brecciation leads to lower values (mean 4.8‰, down to 3.1‰). Hf-isotope compositions lie midway betweenthe Chondritic Uniform Reservoir (CHUR) and DepletedMantle (DM) reservoirs; Hf model ages suggest that thesource region separated from DM in Neoproterozoic time (1500–1000 Ma). Most Cretaceous zircons have176Hf/177Hf similar to those of the older zircons, suggesting recrystallization and/or Pb loss from older zirconsin the Cretaceous thermal event. The Permo-Jurassic zircons show trace-element characteristics similar tothose crystallized from plume-related magmas (Iceland, Hawaii). Calculated melts in equilibrium with themare characterized by strong depletion in LREE and P, large positive Ce anomalies, variable Ti anomalies, andhigh and variable Nb, Ta, Th and U, consistent with the fractionation of monazite, zircon, apatite and Ti-bearingphases. We suggest that these coarse-grained zircons crystallized from late differentiates of mafic magmas,ponded near the crust-mantle boundary (ca 30 km depth), and were reworked repeatedly by successively youn-ger igneous/metasomatic fluids.The zircon data support a published model that locates a fossil Neoproterozoic plume head beneath much of theArabia-Levant region,which has been intermittentlymelted to generate the volcanic rocks of the region. The Cre-taceous magmas carry mantle xenoliths derived from depths up to 90 km, providing a minimum depth for thepossible plume head. Post-Cretaceous magmatism, as recorded in detrital zircons, shows distinct peaks at30 Ma, 13 Ma, 11.4 ± 0.1 Ma (a major peak; n=15), 9–10 Ma and 4 Ma, representing the Lower and Cover Ba-salts in the area. Some of these younger magmas tapped the same mantle source as the Permian-Jurassicmagmatism, but many young zircons have Hf-isotope compositions extending up to DM values, suggesting der-ivation of magmas from deeper, more juvenile sources.

© 2018 Elsevier B.V. All rights reserved.

1. Introduction

Zircon is an excellent tracer of the timing andnature ofmagmatic ac-tivity. It also can be used when magmas remain in the upper mantle ordeep crust, where they can be sampled by later eruptive magmas. Thisapproach is directly analogous to detrital-zircon analysis in crustal set-tings, where zircons are used to investigate the magmatic and meta-morphic history of rocks in the drainage area. As with detrital-zircon

studies, this tool becomes especially effective when U\\Pb dating iscombined with the analysis of Hf\\O isotopes and trace elements.

Several studies of kimberlites (Belousova et al., 2001; Kinny et al.,1989; Tretiakova et al., 2017) have identified multiple populations ofmantle-derived zircons, demonstrating that zircon can retain crystalli-zation ages despite high temperatures; it thus can record magmatic ep-isodes for which surface expression has been lost, or not recognized.This is less commonly possible with more mafic types of volcanicrocks, perhaps because xenocrystic zircons have been dissolved or per-haps because the presence of magmatic zircons is not expected. In gen-eral, zircon does not crystallize directly frommafic magmas, because ofthe high solubility of Zr in such melts. However, if mafic magmas stall

308 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

and fractionate during long slow cooling in the upper mantle or lowercrust, zircon can crystallize from late residual melts or fluids. In areasof continuing magmatic activity, such zircons can be entrained in thenext erupting magma, and their ages may be indistinguishable (withinerrors) from the age of the host basalt. Alternatively, the deep-seatedzircons may react with later melts and fluids to give complex age pat-terns, when they are sampled by still later magmas. The analysis of zir-con xenocrysts in mafic rocks thus can be used to examine themagmatic history of the uppermost mantle and lower crust. In thispaper we use this analytical tool to investigate the history of basalticmagmatism, as sampled by late Cretaceous volcanoes on Mt. Carmel,northern Israel, with special emphasis on events of crustal underplatingfrom the breakup of Gondwana to the Pliocene. A secondary aim hasbeen to constrain the ages of some of the eruptive centers and to under-stand differences in the sampling patterns of individual centers, as aguide to exploration.

Late Cretaceous magmatism in theMt. Carmel area (Fig. 1) involvedrepeated pyroclastic eruptions in a shallow marine-shelf setting, build-ing several km-scale seamounts over the period from ca 97–80 Ma(Segev and Sass, 2009). Although no older volcanic rocks outcropwithinthe region, the Cretaceous vents and tuffs described here contain severalpopulations of older zircons with distinctive characteristics, revealing ahistory of mantle-derived magmatism stretching back to Lower Perm-ian time (ca 280 Ma). These occurrences offer new insights intomagmatism related to the breakup of Gondwana, and raise questionssurrounding the behaviour of the lithospheric and asthenosphericman-tle during rifting, and the conditions that would focus repeated maficmagmatism into this small area (ca 150 km2) since Permian time.

Herewe present U\\Pb dating, O\\Hf isotopic analyses and trace-el-ement data from N400 zircons, collected both from the Cretaceous pyro-clastic rocks and from associated alluvial deposits. We use these data toconstrain the ages of some of the Cretaceous eruptive centers, to mapout themantle distribution of intrusivemagmas related to the older vol-canic episodes, and to discuss the questions flagged above.

Fig. 1. Setting of the Mount Carmel area withinAfter Segev (2009).

1.1. Regional setting

Thematerial describedhere is derived fromCretaceous volcanic cen-ters and related alluvial deposits in the Mt. Carmel-Yizre'el Valley areaof northern Israel (Fig. 2). This area lies within a complex system ofminor rifts and other faults, related to the Africa-Arabia plate boundarythat later developed into the Dead Sea Transform, with N100 kmmove-ment since the initiation of offset in Miocene time. The NW-SE CarmelFault that bounds Mount Carmel on the east (Fig. 1) is part of the Car-mel-Gilboa system, which forms a 10–20 km wide belt of faulting run-ning through the Yisre'el Valley, and may extend across thecontinental margin (Segev and Rybakov, 2011). The SW side of MountCarmel is dissected by multiple N-striking vertical faults with offsetsup to a fewhundredmeters. Thebasement rocks,which are not exposedin this area, are considered to have formed in Pan-African (Cadomian)time (N620 Ma; Stein and Goldstein, 1996). The Galilee area representsa zone of thin continental crust, with aMoho depth of 23–32 km (Segevand Rybakov, 2011). The geophysical data, and estimates of the Creta-ceous geotherm from mantle-derived xenoliths and xenocrysts (Apter,2014; our unpublished data) suggest a thin (b100 km), hot lithosphericmantle. It is not clear whether this is a remnant of a previously thickerlithosphere that was thinned during rifting and drifting, or has devel-oped later through magmatic processes (Stein and Hofmann, 1992).

This region experienced considerable mafic and felsic volcanismduring and following the Permo-Triassic Palmyra rifting, related to thebreakup of Gondwana and the opening of the Neo-Tethys ocean refs.The northern end ofMt. Carmel is the centre of amajorNW-SEmagneticanomaly,which extendswell offshore.Modelling by Segev and Rybakov(2011) indicates that the magnetic source is buried at least 5 km belowthe surface, and the authors suggest that it reflects magmatism relatedto the Palmyra rifting. This was followed by a major Early Jurassic(197–193 Ma) event which erupted up to 2.4 km3 of alkali-olivine ba-salts and pyroclastic rocks with OIB-intraplate geochemical signatures(Dvorkin and Kohn, 1989). A high positive gravity anomaly centered

the tectonic framework of the Middle East.

Fig. 2. Geological map of the Yizre'el Valley-Mt Carmel area, showing sampling localities on Mt. Carmel. BO, Beit Oren; HA, Har Alon; MAK, Makhura; MUH, Muhraka, RMC, RakefetMagmatic Complex; KM, KeremMaharal; BS, Bat Shelomo; T, Tavassim; S, Shefaya; O, Ofer.

309W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

under the northern end ofMount Carmel has been interpreted to reflectthe existence of a buried shield volcano beneath Mt. Carmel (Ben-Avraham and Hall, 1977; Gvirtzman et al., 1990). However, a more re-cent study (Segev and Rybakov, 2011) suggests that the anomaly isonly high relative to deep basins on either side, which are filled withthick clastic sediments.

The early development of the late Jurassic-Early Cretaceous Levantmagmatic province, extending ca 800 km from central Syria to the

Gulf of Suez, coincided with major rifting across Africa, but no majorrifting events appear to correlate with the relatively sparse distributionof Early Cretaceous (C1, 137–139 Ma; Segev, 2005; references therein)continental magmatism within Israel. This magmatism comprisessubalkaline to alkaline basalts, nephelinites, basanites andmicrogabbros with hotspot/mantle plume geochemical and isotopicsignatures (Garfunkel, 1989; Laws and Wilson, 1997; Stein andHofmann, 1992; Stein and Hofmann, 1994). This type of magmatism

310 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

continued until ca 125 Ma in Lebanon and possibly into northern Israel(C2; Segev, 2005).

The Late Cretaceous (98–94Ma, Turonian-Cenomanian) volcanic ac-tivity of northern Israel took place in the Mount Carmel-Umm El Fahmarea (Fig. 2), and has been described by Sass (1980). The dominant vol-canic rocks are pyroclastics of mafic to ultramafic composition; lavaflows played a minor role in this volcanic activity. The volcanism tookplace in a shallowmarine environment, in intimate contact withmarinesediments containing rich assemblages of fossils. Several outcrops ofblack and variegated pyroclastics in isolated areas west of KeremMaharal (Fig. 2) overlie the Yagur Formation of Albian (Bein, 1974;Rosenfeld and Raab, 1984) to earliest Cenomanian age (Lewy, 1991)and represent the first volcanic eruptions in the area. The black pyro-clastics appear to represent eruptive vents, while the variegated pyro-clastics occur as layers of various thickness interbedded with thecarbonates, indicating repeated explosive eruptions and the construc-tion and levelling of small seamounts. The eruptions continued up toca 94 Ma, represented by the Shefaya volcano (Fig. 2).

There is little published information on the composition of the LateCretaceous eruptive products, which are variably altered by contactwith sea water. Area-scan EDS analyses of finely-crystalline lapillifrom the Rakefet Magmatic Complex have compositions correspondingto tholeiitic picrites, but even these apparently fresh samples may havelost alkali elements during diagenesis. The pyroclastic deposits carrymantle-derived xenoliths (spinel peridotites, garnet ± spinel pyroxe-nites, rare garnet websterites), crustal xenoliths ranging from garnet-il-menite granulites to Triassic limestones, megacrystic Mg-ilmenite,amphibole and clinopyroxene, and xenocrysts (olivine, orthopyroxene,diopside, pyrope garnet) derived from mantle peridotites, pyroxenitesand eclogites. P-T estimates on rare garnetwebsterite xenoliths indicatethey were derived from depths up to ca 90 km (Esperança andGarfunkel, 1986; Mittlefehldt, 1986; Apter, 2014; Kaminchik et al.,2014; our unpublished data), giving a minimum depth for generationof the host magmas.

Abundant Neogene volcanic rocks extend throughout the regionboth on the surface and in the subsurface, and are divided into twomain volcanic episodes. The earlier episode (Lower Basalt Fm.) isassigned to Mid-Late Miocene and was characterized by extension andfaulting across the entire region, coinciding with the initiation of left-lateralmovement on theDead Sea Transform. This extensional stress re-gime caused the break-up of the crust, including reactivation of NW-SEfault systems and graben development, opening paths for volcaniceruptions focused on the valley margins, adjacent to normal faults.The younger volcanic period (Cover Basalt Fm.) is assigned to the Plio-cene and characterized by further faulting of the valleys. Large-scale vol-canism prevailed across Saudi Arabia, Jordan, Syria and northern Israel(Segev, 2005), and large areas of central and eastern Galilee as well asthe Golan Heights were covered by basalt flows (Weinstein et al.,2006), originally totaling ca 300,000 km3 (Shaliv, 1991). Later faultingand accumulations of the Cover Basalt in the range of 0–150 m, andminor eruptions between these two episodes, are expressed in out-crops, mostly in eastern Galilee. In the lower Miocene, between theeruptive periods of the Lower and Cover basalts, the areawas inundatedby a marine transgression, which had important implications for thedistribution of the heavy minerals eroded from the Cretaceous volca-nics. For example, beach placers (quartzites with carbonate cement)containing abundant moissanite, garnet and spinel have been samplednear Migdal-el-Haq during the Shefa Yamim exploration program, farfrom known Cretaceous volcanic centres.

1.2. Local setting

The late Cretaceous volcanism onMount Carmel has been describedin elegant detail by Sass (1980). Despitewidespread normal faulting, hewas able to identify 16 volcanic events (Table 1); in 9 cases the corre-sponding vent was identified, while in three other cases the vents are

buried under younger sediments. Lavas play a minor role in this volca-nism; four sources were identified but none is correlated with theknown explosion vents. The typical Mount Carmel volcano was a pyro-clastic cone, erupted in shallowwater (b60mdeep), and later eroded tobelow the water level. Vents are up to several hundred meters across,and are characterized by dense black pyroclastics, while the flanks con-sist of variably altered air-fall and water-laid pyroclastic debris, dippingaway from the cone. Estimated heights of the uneroded cones are from100 to 350 m, and the ash-fall deposits extend for several kilometersaround some of the vents, where not disrupted by faulting (e.g. Makura,Fig. 2).

The most complicated volcanoe described by Sass (1980) is theRakefet Magmatic Complex, the outcrops of which extend for ca 2.5km and are bounded by faults on either side. The Complex comprisesthree separate eruptive events. The vent of the first Rakefet volcano ismarked by black pyroclastics exposed over at least 60 m vertically.Our own field investigations suggest that at least three eruptions oc-curred within the area of the main vent. The bedded pyroclastics ofthis vent are overlain by those of the second volcano, from a ventsouth of the exposed area. The third volcanic event comprises ca 25 mof olivine basalt and andesine olivine basalt, some vesicular. Theirsource is not located but is presumed to be nearby. The Rakefet VolcanicComplex is the only example in the area of repeatedmagmatismwithina small area. The Muhraka volcano to the north, known mainly frombedded pyroclastics, may be related to this complex (Fig. 2).

Mount Carmel is bounded on the south by the sediments of theRamot Menashe syncline (Fig. 1). Drilling (Ein Ha-Shofet #1 deepmon-itoring drill, Mekorot Ltd) on a magnetic high near the axis of the syn-cline intersected volcanic rocks at 788–792 m depth, near the base ofthe Deir Hanna formation. The rocks at this level are tuffaceous, andsome horizons are marked by abundant large euhedral, zoned crystalsof high-Mg calcite. Several other basalt/tuff horizons were intersectedhigher in the Deir Hanna formation (550–750 m). The Uhm-al-Fahmarea, still further south of Mt. Carmel (Fig. 2), exposes a thick sectionof basalts, which have not been sampled in this study.

The Shefa Yamim exploration program has sampled many of thesmaller streams and rivers in the drainage basin of the Kishon river,lying to the east and southeast of Mount Carmel (Fig. 2). Bulk samplinghas been focused on the mid-reach area, where the Kishon River flowsthrough a narrow gap between Mount Carmel and the adjoining up-lands, creating a natural trap for transient placers. Most of the sampleslabelled “alluvial” in this paper are from this area.

1.3. Sampling

The Shefa Yamim exploration project is aimed at the discovery ofeconomically viable placer deposits of gemstones (mainly sapphire,ruby, hibonite, moissanite) and other commodities within the drainagebasin of the Kishon River. The project has sampled the Cretaceous pyro-clastic centers on Mt. Carmel (Fig. 2; Rakefet, Har Alon, Bat Shelomo,Muhraka and Beit Oren complexes) and adjacent areas (Ein Ha-Shofet),and minor and major drainages in the Yaz'rael Valley. Samples range insize from several kg to N1000 t. All samples were screened through astatic grizzly screen to remove pieces larger than 100 mm in diameter.Rock samples from the vents were coarsely crushed and then treatedin the same way as alluvial samples. The b100mm fraction waswashedin a scrubber that breaks up any clods. The b0.5 mm component issuspended in the wash water and pumped to settling ponds; fractionslarger than 25 mm are used to backfill exploration pits. Samples in the+8 mm–16 mm and + 16 mm–24 mm size fractions are sorted byhand on a picking belt. The +0.5 -8 mm component of the sample iswashed and classified into 5 fractions: 0.5–0.7 mm, 0.7-1 mm, 1-2 mm, 2-4 mm, 4-6 mm, 6-8 mm. These fractions are transferred to apulsating jig plant for gravity separation. Samples in the 2 mm–8 mmsize fractions are visually inspected after the jigging process and sortedin the recovery laboratory. The three smallest size fractions are jigged

Table 1Late Cretaceous Volcanic Events in Mt. Carmel, and generalized lithostratigraphic column.After Sass (1980).

Event no. Rock typesa Vents with proximal deposits or lavas Isolated pyroclastics (distal flanks) Remarks

1 P,X Kerem Maharal (KM) N. Carmel Lower Tuffs Time equivalent2 P,X Beit Oren (BO)3 P,X N. Carmel Intermediate Tuffs Possibly same volcano4 P,X Tira N. Carmel Upper Tuffs Possible time equivalents5 P Fureidis Tuffs6 P,X Tavassim (TA) Possible time equivalents7 P,X Rakefet 1 (RAK)8 P,X Rakefet 2 (RAK)9 P Shefaya (SH) Possible time equivalents10 P Makura (MA)11 P Muhraka Tuffs (MU)12 L Shefaya lavas Probable time equivalents13 L Ofer lavas (Of)14 L Rakefet lavas Local or fissure sources15 L Muhraka lavas16 P,X Bat Shelomo (BS)

a P = pyroclastics, L = lavas, X = xenoliths and xenocrysts.

311W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

separately; most zircons in this study come from these fractions. Theheavy concentrate in the center of the jig pan is collected and dried;ma-terial on the outer part of the jig pan is discarded. The sorters in the lab-oratory have demonstrated their efficiency in identifying andrecovering a wide range of mineral species, including garnet (pyrope),ilmenite, spinel, chrome-diopside, diamond, moissanite, sapphire,ruby, Carmel Sapphire™ and hibonite, as well as rutile and zircon. Thesample set described here is dominated by zircons from the RakefetMagmatic Complex, and the alluvial deposits of the Kishon River mid-reach, because these localities have been bulk-sampled, providingabundant material.

The unprocessed heavy mineral concentrates of several sampleswere hand-picked under a binocular microscope in the CCFS laborato-ries; a few rock samples also have been processed by SelFrag (electro-static disaggregation) techniques at CCFS, sieved and hand-pickedafter magnetic and heavy-liquid separation.

1.4. Sample descriptions

Two general groups of zircons in this study can be distinguished bycombinations of external morphology and internal structure, here des-ignated “Mount Carmel zircon” (MCZ) and “common zircon” (CZ).

MCZ is themost distinctive group. It comprises generally large grains(0.5–2 mm) with blocky to rounded to irregular shapes, including pro-nounced rounded knobs and depressions (Fig. 3). Prism or pyramidfaces are visible on some grains but usually show rounding and resorp-tion. These grains are typically colourless and clear, with a brilliant

luster. Most MCZ have low Cl response, but images of some grainsshow patterns that suggest relict, often blurred, igneous or metamor-phic zoning, including core-rim structures with irregular cores. Manygrains appear to be fractured or even brecciated (see below). Similar ir-regular zircons have been separated from kimberlite-borne peridotitexenoliths (A. Guliani, pers. comm. 2016), suggesting that this morphol-ogy was imposed when the zircons crystallized frommelts in the inter-stitial spaces between other minerals, either in cumulates or in theperidotitic wall-rock. The outer surfaces of some grains show an un-usual ‘bubble” texture, suggesting partial melting of the grains, perhapsduring magmatic transport.

For the purposes of this paper, the MCZ zircons can be subdividedinto three classes: igneous, igneous/metamorphic and breccia grains.“Igneous” grains show oscillatory zoning, or broad lamellar zoning, re-gardless of external morphology; however, most have more regularshapes than other MCZ grains. Some grains have distinct cores (usuallymetamict, some with contorted zoning) surrounded by thick rims withor without oscillatory zoning. These are found especially in the BatShelomo vent and the core from the Ein Ha-Shofet locality (Fig. 3a).

“Igneous/metamorphic” grains are mainly homogeneous, but manycontain traces of zoning, or small areas with clear zoning within largegrains that otherwise are homogeneous in CL images. Some of thesegrains have relatively homogeneous cores, partially to completelysurrounded by thick homogeneous rims with lighter or darker CL re-sponse (Fig. 3a). While the grains with homogeneous CL could beregarded as metamorphic, the continuous transition toward the “igne-ous” grains, in terms of internal structure and external morphology,

Fig. 3. Images of Mount Carmel zircons. Scale bars are 200 μm. a). Cathodoluminescence (CL) and optical-microscope images of zircons showing corroded-prismatic shapes and core-rimzoning; b). Typical Mt. Carmel-type zircons illustrating irregular shapes and internal structures; c). Optical and CL images of large Mt. Carmel-type zircon showing irregular form andinternal brecciation; d). CL images of zoned Mt. Carmel-type zircons with brecciation.

312 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

persuades us that most had an igneous heritage. None of these grainshas the low Th/U (b0.1) commonly regarded as indicative of metamor-phic zircon, though this criterion may not be applicable to zircons inmafic or ultramafic rocks. These issues are discussed further below inthe light of the isotopic data.

“Breccia” grains generally have low CL response and no traces ofzoning, and are typically irregular in shape. They appear to have beenfractured and rehealed. The differences in CL suggest that the internalfragments are slightly misaligned, but the fractures are sealed with zir-con, in optical continuity with the host grain. In some grains brecciatedcores are overgrown by featureless rims (Fig. 3d). Zircons with similarbrecciated/rehealed internal structures have been found in heavy-min-eral separates from Siberian kimberlites (Tretiakova et al., 2017).

CZ from the vents and the alluvial deposits display a wide range ofmorphology, ranging from prismatic to rounded. Oscillatory to lamellarzoning in CL images indicates that most of the CZ grains are igneous inorigin, but some show evidence of metamorphism. The alluvial depositscontain a higher proportion of broken and/or rounded grains, as mightbe expected. The MCZ population is relatively abundant in the venttuffs, but much less common in the alluvial deposits, suggesting thatthe brecciated grains, in particular, may not withstand transport aswell as the CZ population.

1.5. Analytical methods

The zircons have been characterized using optical microscopy, scan-ning electron microscopy (SEM) for CL imaging, electron microprobe(EMP) for major-element analysis, laser-ablation (LA)-ICPMS forU\\Pb and trace-element analyses, secondary ion mass spectrometry(SIMS) for U\\Pb and O-isotope analysis, and LA-multicollector (MC)-ICPMS for analysis of Hf isotopes. Brief details of these methods aregiven in the Appendix. The zircons have been classified in terms of

their primary host rock using their trace-element chemistry and theclassification schemes of Belousova et al. (2001).

2. Results

2.1. U\\Pb ages

The U\\Pb dataset comprises 240 analyses of zircons from the ventsand tuffs, and 190 analyses of detrital zircons. Essentially all analyzedzircons are concordant (Table A1; Fig. 4). In this paper, ages b1 Ga are206Pb/238U ages; older ones refer to 207Pb/206Pb ages.

The zircons from the vents show age peaks in the Permian (ca285–260 Ma), Triassic (ca 240–220 Ma) and lower Jurassic (ca210–185 Ma; correlative with the “Asher volcanics” known from deepwells), which tails off with time to ca 175 Ma (Fig. 5a). There is thenan apparent 40-Myr near hiatus in magmatic activity up to ca 118 Ma,and then increasing numbers of zircons with decreasing age to a peakat ca 100 Ma; this tails off to 80 Ma. Bulk sample #479 of the RakefetMagmatic Complex yielded five older grains with ages of 3124, 2726,538 and 420 Ma. The Ein Ha-Shofet sample contains more older grains,including two Neoarchean grains (2575, 2635 Ma), two mid-Protero-zoic grains (1829, 1852 Ma) and several Neoproterozoic ones (614,757, 796, 950, 975Ma). Most of these are clearly igneous, with well-de-veloped CL zoning, but many are well-rounded or fragmental. Somemay be detrital zircons, sampled from sediments intercepted at depthby the basalt during eruption. However, the 2767-Ma grain fromRakefet is overgrown by zircon dated at 206 Ma, and presumably wascaught up in the Jurassic magmatism.

The individual vents vary widely in their zircon populations (Table1), although some of this may reflect the volumes sampled from differ-ent sites. The zircon suite (n = 100) from the Rakefet Magmatic Com-plex, which was bulk-sampled, covers the entire range of Permian to

Fig. 4. Concordia plots of representative U\\Pb data of Mount Carmel zircons. a). 0–30Ma (inverse Concordia); b-1). 30–98Ma (inverse Concordia); b-2). 98–120Ma (inverse Concordia);c-1). 100–200 Ma (Concordia); c-2). 200–300 Ma (Concordia).

313W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Jurassic ages, but has only one grain relevant to the age of eruption (93± 3 Ma). The Bat Shelomo locality is dominated by Permo-Triassic zir-cons (mean age 253 ± 15 Ma, n = 18). It also includes one old grain(283 ± 3 Ma) and one grain (99 ± 4 Ma) that may be close to the ageof the vent. However, the sample also contains two Miocene zircons(13–12 Ma) whose origin is unknown; they may be related to the Mio-cenemarine transgression described above. The Beit Oren complex alsocontains zircons with ages ranging from 286 to 190 Ma, but no youngerones. The Makura complex yielded only 2 grains, with ages of 282 and

281 Ma. The Ein Ha-Shofet sample yielded 76 zircons, of which 50were dated. Most fall into two clearly separated populations: 190 ±3 Ma (n = 12) and 273 ± 6 Ma (n = 22). One grain at 99.9 ± 1.4 Maconstrains the age of the eruption.

The Tavassim vent yielded one Permian grain; the remainder belongto the Cretaceous suite, ranging from 128 to 84 Ma (n = 4). Similarly,the Karem Maharal tuffs yielded two Permian zircons (276 and248 Ma) and three Cretaceous ones (107–90 Ma), providing some con-straint on the eruption age. Four of the five zircons from Har Alon are

Fig. 5.Age histograms and cumulative probability curves for U\\Pb data ofMount Carmel zircons. a). All (0–350Ma); b). Vent samples (0–350Ma); c). Alluvial samples (0–350Ma); d). Allsamples (0–20 Ma).

314 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Jurassic (199–193 Ma) but a younger grain (98 Ma) may date theeruption.

The alluvial deposits (Fig. 5c) contain surprisingly few of the pre-Cretaceous zircons (MCZ) that dominate the populations from the sam-pled vents (Fig. 5b); nearly all of the zircon grains fall into the “commonzircon” group described above. However, the zircons from the streamsediments show a major complex age peak extending from 120 to75 Ma, indicating that Cretaceous vents have in fact been sampled; thescarcity of Jurassic-Permian zircons thus may reflect a lower survivalrate of these complex zircons during alluvial transport. The presenceof late Cretaceous zircons with ages (75–70Ma) younger than the sam-pled vents suggests the (former?) existence of undiscovered vents, andthe continuation of the volcanism into Campanian time. The alluvial

samples also contain zircons with a scattering of ages (Fig. 5d) extend-ing from the Eocene up into the Pliocene, with peaks at ca 30Ma, 13Ma,11–12 Ma (a major peak; n = 15, 11.4 ± 0.1 Ma, MSWD = 0.79), 9–10 Ma, and 4 Ma. These appear to represent zircons from the Lowerand Cover Basalts.

Brecciated zircons from the Permian-Jurassic populations can showspot-to-spot internal variations ranging from a few tens of Myr, up to190 Myr. In general, the more homogeneous blocks retain the oldestages, while boundaries between domains are younger, but several ex-ceptions were noted. 17 core-rim pairs were analyzed, mostly withinthe 250–200 Ma population. However, one Archean (2767 Ma) corehad a distinct 206Ma rim, suggesting that the Archean grainwas pickedup by the hostmagma emplaced in Triassic time. One Jurassic core (188

315W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Ma) had a Cretaceous rim (104 Ma), and one late Cretaceous (74 Ma)core had an Oligocene rim (31 Ma). These pairs are discussed furtherbelow, with the Hf-isotope data.

2.2. Trace-element compositions

Extended chondrite-normalized REE patterns for 90 zircons from thedifferent vents are gathered in Fig. 6; the analytical data are compiled inTable A2.

All but one of the analyzed zircons fromBat Shelomo are classified asigneous. Most of the Permo-Triassic zircons have very similar patterns,with an overall gently positive slope from Sm to Lu, and strong negativeEu anomalies coupledwith large positive Ce anomalies. One grainwith ametamict core shows disturbance in the form of enrichment in LREE.One Triassic grain (233 Ma) differs significantly in having lower LREEand no Eu anomaly, and this pattern is similar to those in both the Cre-taceous and the Miocene zircons. The Miocene grain has much higherREE than the other grains.

A similar pattern is seen at Karem Maharal, where the two Permiangrains have high HREE and negative Eu anomalies, while the Cretaceousgrains have flatter patterns with no Eu anomalies. At Tavassim and BeitOren, this pattern is repeated; Permian grains show an Eu anomalywhile the Cretaceous grains do not. The metamict cores of two oldergrains showanenrichment in LREE. The zircons fromHar Alon are Juras-sic aside from one Cretaceous one. The patterns of the Jurassic zirconsresemble those from the older populations, but with shallower Euanomalies; the one Cretaceous grain has a flatter pattern and no Euanomaly.

The relatively few obviously igneous zircons from the Rakefet Mag-matic Complex all have ages in the range 253–175 Ma and have verysimilar patterns, with modest Eu anomalies in the oldest grains andsmaller ones in the younger ones. They show a wide range in Th andU contents, and Th/U b 1. Themuch larger number of igneous/metamor-phic grains shows awider range in absolute REE contents, and especiallya larger spread in the HREE than in the other suites discussed above.Several of the patterns have flat to slightly negative slopes from Dy toLu, combined with very minor Eu anomalies. The spread in Th and Uconcentrations covers more than two orders of magnitude. In strikingcontrast, the brecciated zircons are more homogeneous, aside from arange of Eu anomalies.

The variations in REE patterns, U\\Th relationships and the Euanomalies through time are illustrated in Fig. 7. The plot of La/Sm vsage (Fig. 7a) shows that despite general similarities, especially in theslope of the HREE, the Permian to Jurassic zircons tend to have moreLREE-enriched patterns than the Cretaceous and younger zircons. Themore strongly negative Eu anomalies, and lower Ce anomalies, of thePermian to Jurassic zircons, are obvious in Fig. 7b,c. There is no clear re-lationship between Eu/Eu* and the morphological classifications. Thereappears to be an overall decrease in the depth of the Eu anomaly fromthe Permian to the Jurassic, but it is not clear if this is significant.

The abundances of Ce, U and Ti have been converted to estimates ofoxygen fugacity (fO2) using the formulation of Loucks and Fiorentini(2018). The estimates (Table A2; Fig. 7d) cover N4 log units, centeredaround the fO2 of the QFM buffer; there is no clear age trend in thedata, but the three 9–12 Ma zircons analyzed record higher fO2 thanmost of the older grains.

The Th/U ratios of the zircons show some change with time; Th/UCN

values N0.3 aremuchmore common in theMiocene to Pliocene popula-tions than in the older ones (Fig. 7e). The Triassic zircons tend to havelower Th/U than either the Permian or the Jurassic populations. Thelowest U and Th contents are found in brecciated and igneous/meta-morphic zircons. We interpret this as one effect of the metamorphismand reworking of these complex zircons.

The analyzed zircons have been classified in terms of their parentalmagmas using the trace-element criteria of Belousova et al. (2001).Nearly all classify as derived from either carbonatite (51/90) or low-Si

granitoids (36/90); three grains classify as kimberlitic zircons. This pat-tern is typical, in our experience, of mantle-derived zircons in continen-tal settings, and probably reflects crystallization of zircon from latedifferentiates of mafic magmas. In the plot of Y vs U/Yb suggested byGrimes et al. (2007, 2015)most of the grains fall into the field of “conti-nental zircon” as distinct from “ocean-crust zircon”, although a signifi-cant proportion fall outside the suggested fields due to their low U atlow Y contents (Fig. 7f). These anomalous grains largely comprise thebrecciated and igneous/ metamorphic MCZ grains. In the U/Yb-Nb/Ybplot of Grimes et al. (2015) (Fig. 7g) nearly all of the analyzed zircons,regardless of age, fall in the field of Ocean-Island zircons, within themantle array.

2.3. Hafnium-isotope compositions

Hafnium-isotope data for N350 grains, mainly from the vents, aregiven in Table A3.

The pre-Cretaceous zircons show a range of 176Hf/177Hf)i from ca0.2828 to 0.2829 (Fig. 8a), corresponding to εHf values between +5and + 10 for most grains. A plot of 176Hf/177Hf)i vs age (Fig. 8b)shows a broad trend of increasing 176Hf/177Hf)i from Permian throughJurassic time, with a mean slope corresponding to 176Lu/177Hf = ca0.038, essentially parallel to the DepletedMantle curve. This trend is in-dependent of the morphological classification, and is visible in the indi-vidual populations from the vents of Rakefet and Bat Shelomo, forwhichsufficient data are available. It is typical of mantle-derived zircons fromareas such as Siberia andAustralia. A similar, but steeper, trend has beenrecognized in kimberlitic zircons from southern Africa (Fig. 8b;Woodhead et al., 2017).

Most of the Cretaceous zircons have 176Hf/177Hf)i ratios within thesame narrow range observed in the Permian to Jurassic suites (Fig.9a), and thus have lower εHf values. Many of the Miocene-Pliocene zir-cons have 176Hf/177Hf)i within the range covered by the main popula-tions of Cretaceous to Permian zircons (Fig. 9b), but a significantnumber have significantly more radiogenic Hf, extending toward theDepleted Mantle reference line.

A pattern of a decrease in agewithout a change in Hf isotope compo-sitionmay reflect Pb loss due to later thermal disturbance. In thepresentcase, the situation appears to be more complex. In eight of fourteencore-rim pairs with clearly-defined rims that are significantly youngerthan the core of the grain, the two zones have identical to nearly identi-cal Hf-isotope compositions (Fig. 9c). Six pairs have rims with signifi-cantly higher 176Hf/177Hf)i than the cores, including one grain with aJurassic core and a Cretaceous rim; the tie-lines between core and rimin these cases mimic the overall very broad trend to more radiogenicHf with time (Fig. 9a). There is only one case in which the rim haslower 176Hf/177Hf)i than the core. The microstructures, combined withthe Hf-isotope evidence, suggest that the rims represent growth ofnew zircon, remobilized from the existing heterogeneous reservoir,rather than simple Pb loss.

2.4. Oxygen-isotope compositions

Oxygen-isotope analyses of 45 spots on 15 zircon grains from theRakefet volcanic vent (Table A1) cover a range in δ18O from 3.7 to5.9‰ relative to Vienna Standard Mean Ocean Water (SMOW), with amean of 5.05 ± 0.45‰ (1 s). One grain gives a very high δ18O, ~12‰,which is not considered further. Most analyses are within the accepted“mantle range” of 5.6 ± 0.6‰ (2 s), but the lowest values lie outside ofthis range. Intra-grain variation typically is within the 1 s analytical er-rors (ca 0.3‰), but one brecciated grain shows a range from 4.7 to5.4‰ (n = 6 spots), and two igneous/metamorphic grains show core-rim differences of 0.6‰ (n=2) and 0.8‰ (n = 2), with the rim lighterin each case. The isotopically heaviest oxygen is found in the grainswithclearly igneous structures (mean δ18O=5.5‰; n=3) and as individualspots in igneous/metamorphic grains. The brecciated grains have a

0.001

0.01

0.1

1

10

100

1000

10000

100000

1000000Zi

rcon

/Cho

ndrit

ea) Migdal Ha-Emeq b) Har Alon

Zirc

on/C

hond

rite

c) Rakefet100000

10000

1000

100

10

1

0.1

0.01

0.001

100000

10000

1000

100

10

1

0.1

0.01

0.001

d) Beit Oren & Makura

Beit OrenMakura

e) Bat Shelomo

Zirc

on/C

hond

rite

f) Karem Maharal

h) Tavassimg) Kishon River

0.0001

0.001

0.01

0.1

1

10

100

1000

10000

100000

Zirc

on/C

hond

rite

i) Kishon River 20-150 Ma

0.0001

0.001

0.01

0.1

1

10

100

1000

10000

100000

PTiTaNbHfLuYbUErHoYDyThGdEuSmNdPrCeLa

Zirc

on/C

hond

rite

<--- Increasing Ionic Radius <---

j) Kishon River 0-20 Ma

PTiTaNbHfLuYbUErHoYDyThGdEuSmNdPrCeLa

<--- Increasing Ionic Radius <---

- 150

- 300 - 150

- 20

- 20

- 0

Fig. 6. Extended trace-element patterns of selected zircons by location, and colour-coded by age. Migdal-Ha-Emeq and Kishon River represent detrital (alluvial) zircons; all others aresampled from vents and associated pyroclastics as indicated.

316 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Fig. 7. Plots of trace-element parameters vs age ofMount Carmel zircons. a). La/Sm vs age; b). Eu anomaly vs age; c). Ce anomaly vs age; d). fO2 vs age; e). Th/U vs age; f). siscriminant plotrelating zircon chemistry to tectonic setting (Grimes et al., 2007); g). discriminant plot relating zircon chemistry to tectonic setting (Grimes et al., 2015). OI, ocean island volcanics. Slopingbox encloses zircons from mantle-derived rocks.

317W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Fig. 8. Hf-isotope data. a). Hf-isotope data vs age for analyzed zircons, with CHUR andDepleted Mantle (DM) evolution curves. Dashed lines show estimates of depleted-mantle model ages; TDM is shown for the least radiogenic compositions; TDMC model agesfor the least and most radiogenic Hf from the Permian-Jurassic zircons are calculatedusing a mean crustal 176Lu/177Hf of 0.015 (Griffin et al., 2000). b). Hf-isotope data vs age,plotted by vent. Shaded field encloses zircons from off-craton kimberlites in Siberia andAustralia. Inset shows the trend in kimberlitic zircons from southern Africa (Woodheadet al., 2017).

318 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

mean δ18O = 4.8‰ (n = 7, range 3.7–5.1‰); similar low values arefound in the igneous/metamorphic population,where averages for indi-vidual grains range from 5.3 to 3.8‰. These patterns suggest that theprocess of metamorphic homogenization and brecciation of primary ig-neous grains involved fluids with oxygen that was isotopically some-what lighter than the “mantle average”.

Fig. 9. Hf-isotope data a). Hf-isotope data vs age, plotted by textural type as seen in CLimages. b). Hf-isotope data vs age for Neogene zircons from alluvial deposits. c). Hf-isotope data vs age for zoned zircons; lines connect cores and rims.

3. Discussion

3.1. Younger basalts

The zircon data provide information on the ages and sources of theLower Basalts and the Cover Basalts in the drainage of the Kishonriver, mainly the Yizre'el Valley. The youngest population of zircons, as-sumed to represent the Cover Basalts, has 206Pb/238U ages ranging from3.6 ± 0.2 to 5.5 ± 0.4 Ma (mean 4.2 ± 0.6 Ma (1sd, n = 11)), corre-sponding to the Lower Pliocene (Fig. 5d). Another major peak at9–13.5 Ma (mean 11.3 ± 1.2 Ma, n = 22; Mid- to Upper Miocene) isinterpreted as the age of the Lower Basalts in the Yizre'el Valley; agesback to 17.5 Ma are recognized in the Lower Galilee across the water-shed at the head of the Yizre'el Valey. These ages correspond wellwith the magmatic episodes defined by Shaliv (1991), Weinstein(2000)and Rozenbaum et al. (2016). The apparent scarcity of magmaticactivity between about 9 and 5 Ma in this area also was previously rec-ognized by Shaliv (1991). Zircons in each of these age groups show alarge spread in εHf, ranging from low values similar to the Cretaceous-Permian zircons, up to values closer to the Depleted Mantle (Fig. 9a,b).This range suggests that the late Miocene to Pliocene magmas were

derived from the asthenosphere, but also interacted with the SCLMand older underplated basalts (see below) during their ascent.

3.2. Ages of cretaceous volcanic centres

Segev and Rybakov (2011) recognized 5 phases of Cretaceous mag-matic activity, interpreted as reflecting the impingement of a mantleplume centered beneath northern Israel, and that caused a majorupdoming across Lebanon, Syria and the Sinai. The “Tayasir Volcanics”are dated between 141 and 134 Ma (Lower Cretaceous), and basalts ofthis age have been reported from deep drillings around Mt. Carmel.This age range is poorly represented in our dataset (Fig. 5b), thoughthree grains are similar (133 ± 2 Ma) or slightly older (147-146 Ma).The next episode occurs in lower Aptian time (125–123Ma; Elije Volca-nics); it also is reported from drill cores beneath Mt. Carmel; but is ab-sent in our dataset. The third episode, the late Aptian-Albian Ramon

319W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

volcanics, is represented by a group of 7 zircons with ages ranging from116 ± 6 Ma to 112 ± 10 Ma. The “Carmel Volcanics”, defined as occur-ring between ca 100–95 Ma (Table 1), are well-represented in ourdataset. The younger Campanian (ca 83–72 Ma) volcanism is repre-sented by a scatter of zircons from the alluvial samples.

The ages of the youngest zircons from each volcanic center can becompared with available K\\Ar and Ar\\Ar ages (Segev, 2000, 2002,2009; Segev et al., 2005; Segev and Sass, 2009) (Table 1) to provide fur-ther constraints on the age of volcanism and the development of themagmatic system. The Kerem Maharal magmatism, which representsthe earliest known eruptions in the Mount Carmel area, has beendated to 99.0 ± 1.0 Ma using amphibole megacrysts (Segev, 2009).Two of the zircon xenocrysts analyzed here give slightly older ages(107 ± 3, 104 ± 3 Ma), while one is significantly younger (90 ±2 Ma). The Rakefet basalt, which overlies the tuffs sampled for this zir-con study (Table 1) has been assigned an average age of 96.7 ±2.5 Ma (Segev, 2009). Whole-rock Ar\\Ar ages on a nearby (E.Muhraqa) basalt are similar (96.2 ± 2.5 Ma; Segev, 2002). Our largedataset (n = 100) from Rakefet contains only one Cretaceous zircon,dated at 93 ± 3 Ma. Two Ar\\Ar analyses on an amphibole megacrystfrom Tavasim (Segev, 2009) give ages of 99.3 ± 0.6 and 98.0 ±0.5 Ma; both are only slightly younger than the ages of two zircons(102 ± 2 and 101 ± 1 Ma), but as at Kerem Maharal one zircon (84± 8 Ma) is younger than the accepted eruption age. The Bat Shelomovolcano has been dated by Ar\\Ar analysis of amphibole to be 81.6 ±0.8 Ma (Segev, 2002); our dataset contains one Cretaceous zircon (99± 4 Ma) but two Miocene zircons (13.1 ± 0.3, 12.0 ± 0.3 Ma).

In several of these cases, the youngest zircons are slightly older thanthe acceptedK\\Ar andAr\\Ar data. This is consistentwith the idea thatthe zircons are xenocrysts, rather than phenocrysts, in their hostmagma, and thus must be older than the erupted basalt. However, wesee several instances of zircons that are significantly younger than theaccepted eruption ages. These may suggest the existence of unrecog-nized eruption episodes, either in the same area as the sampling, oreven re-using the same volcanic plumbing system as the earlier, maineruptions. In some cases, thismight be enough to affect K\\Ar or Ar\\Arages of amphiboles; this might explain, for example, the anomalouslyyoung age for Bat Shelomo. However, the current datasets from all ofthe vents except Rakefet are small, and further work on samples fromthe other vents is needed to clarify these questions.

3.3. Source(s) of the zircon xenocrysts

In the samples from the vents, a large proportion of the zircons haveages corresponding to the Gevim volcanics (Middle Permian to UpperTriassic, 265–220 Ma), and another large population corresponding tothe Asher volcanics (Upper Triassic to Lower Jurassic, 206–188 Ma).The Hf and O isotopes of these zircons indicate that their parentalmagmas were mantle-derived. As noted above, it is most likely thatsuch zircons have crystallized from late differentiates ofmantle-derivedmaficmagmas. Their REE patterns show little obvious evidence of a gar-net signature, while the presence of negative Eu anomalies in the mostmagmatic-looking zircons suggests co-crystallization with plagioclase.This limits the depth of crystallization to b30 km; we suggest that themost likely location for these underplated basalts, where they couldcool slowly to produce large grains of cpx, amphibole and zircon,would be at the density filter represented by the crust-mantle boundary(25–20 km in this area; Segev and Rybakov, 2011).

The microstructures of the Permian to Jurassic zircons appear to re-flect a complex history of igneous crystallization, metamorphic recrys-tallization, resetting and overgrowth. There is a general increase in theproportion of reworked zircons with time (Fig. 9a), but clearly igneousgrains are present throughout the Permian-Jurassic population. TheHf-isotope data scatter considerably, but overall define a broad trendof increasing 176Hf/177Hf)i with decreasing age. The slope of this trendis essentially parallel to that of the Depleted Mantle reference curve;

the trendmight therefore reflect internal growthwithin a volume of ba-salts+mantle rocks that had 176Lu/177Hf similar to the DM. It is perhapsmore likely that it reflects the mixing of older (Permian) materials withyounger magmas, which supplied both heat and the material requiredto rework the pre-existing rocks and their zircons.

Stein and Hofmann (1992) surveyed all available data on the Phan-erozoic basalts of Israel erupted over the last 200 Ma, and noted thatthey have remarkably consistent compositions, “chemically and isotopi-cally indistinguishable frommany ordinary plume basalts”, and incom-patible with crustal contamination. They suggested that the magmasoriginated from a fossil plume head, emplaced beneath the region inNeoproterozoic time, which periodically was remelted to produce ba-salts when extended, or when influenced by other weak plumes thatfailed to penetrate this carapace. Stein and Goldstein (1996) expandedon this idea, suggesting that such stalled plume heads may be wide-spread, and that they could be “overprinted with continent-like charac-teristics by plate convergence and associatedmagmatism”, becoming anintegral part of the continental lithospheric mantle. Weinstein et al.(2006) showed that these characteristics are shared by theNeogene ba-salts of the region.

This model can offer a reasonable explanation for the Hf-isotopeevolution recorded in the Permian-Jurassic zircon xenocrysts in theMount Carmel volcanics. Their overall Hf-isotope values are similar tothose of HIMU lavas, commonly attributed to plumes with a significantcontribution of subducted oceanic crust. Interestingly, theHf-isotope ra-tios of the Permian-Jurassic zircons also partly overlap the field definedby zircon megacrysts in kimberlites from on- and off-craton situationsin Siberia and Australia, where Archean lithosphere has undergonemodification by Proterozoic (plume-related?) magmatism and defor-mation (Fig. 8b; Figs. 2 and 3 in Griffin et al., 2000). Woodhead et al.(2017) have shown that zirconmegacrysts from Cretaceous kimberlitesin South Africa, both on- and off-craton, require a widespread sourcewith high Lu/Hf (Fig. 8b), which they attribute to plume heads stalledbeneath the lithosphere. The plume-head model is also consistentwith the xenolith suites in the Cretaceous volcanic rocks of Mount Car-mel, which contain both fertile spinel lherzolites and abundantpyroxenites.

The zircons derived from the proposed Permian-Jurassic underplatehave Hf model ages (Fig. 8a) ranging from ca 700Ma (TDM) to 1500 Ma(TDMC ). These could be interpreted as reflecting the age of the proposedfossil plume head, which would be consistent with the Neoproterozoicage suggested by Stein and Hofmann (1992). However, this may notbe a valid approach; it assumes that the plume head originally had theHf-isotope composition of the Depleted Mantle, whereas a HIMU-typeplume could have a significantly less radiogenic Hf-isotope composition.This also is a period of major crustal activity in the Arabian shield(Morag et al., 2012), which involved island-arc magmatism, post-colli-sional granitoidmagmatismand at least two volcanic cycles. The Hf-iso-tope data from the older zircons therefore may reflect the derivation ofbasalts from reworked and refertilized mantle, rather than a plumehead, although the two models are not mutually incompatible.

3.4. Compositions of parental melts

The trace-element patterns of themelts fromwhich the zircons crys-tallizedwere calculated using the empirical (their geometricmean) dis-tribution coefficients Dmelt

zircon of Nardi et al. (2013), on the assumptionthat the zircons crystallized not directly from mafic magmas, but fromtheir late, more SiO2-rich differentiates. The oxygen-isotope data(Table A1) show no evidence of crustal contamination or assimilation.The patterns are grouped by age in Fig. 10.

The patterns of the oldestmelts (270–230Ma;mid-Permian tomid-Triassic) show flat MREE-HREE, but strong depletion in the LREE; theyhave marked positive Ce anomalies, and most showmoderate negativeEu anomalies. They also show high [Nb] and [Ta], a large spread in [Th],[U] and Th/U, and a range of positive and negative Ti anomalies. Several

Rakefet 270-230 Ma1000

100

10

1

0.1

0.01

0.001

Weinstein Neogene

Rakefet 230-210 Ma Rakefet 210-175 Ma

0.01

0.1

1

10

100

1000

Th U Nb Ta La Ce Pr Nd Sm Eu Ti Gd Dy Y Ho Er Yb Lu P

Kishon R. 115-80 Ma Kishon R. 80-20 Ma

Th U Nb Ta La Ce Pr Nd Sm Eu Ti Gd Dy Y Ho Er Yb Lu P

Kishon R. 20-3 Ma

Th U Nb Ta La Ce Pr Nd Sm Eu Ti Gd Dy Y Ho Er Yb Lu P

a) b) c)

d) e) f)

Fig. 10. Parental-melt compositions for selected zircon populations, calculated using the distribution coefficients of Nardi et al. (2013). Colour scales relate to age as in Fig. 6. a). Rakefetmagmatic complex, 270–230Ma. Red line showsmean pattern of Neogene basalts of N. Israel, afterWeinstein (2000); this is indistinguishable from the patterns of the Jurassic-Cretaceousbasalts (Stein and Hofmann, 1992); b). Rakefet magmatic complex, 230–210Ma; c). Rakefet magmatic complex, 210–175 Ma; d). Kishon River Mid-reach placers, 115–80 Ma; e). KishonRiver Mid-reach placers, 80–50 Ma; f). Kishon River Mid-reach placers, 30–3 Ma.

320 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

of these patterns represent zircons that have been disturbed during re-crystallization, leading to high and irregular REE patterns, and especiallyhigh LREE/MREE.

Themid-Triassicmelts (230–210Ma) are similar tomost of the olderones, but show less scatter in the REE patterns and smaller Ti anomalies.[Th] and [U] are generally lower than in the oldermelts, but the range inthese contents is similar. The Triassic-Jurassic (210–175 Ma) melts, onthe other hand, show considerable diversity relative to the older ones.Some show depletion in HREE, leading to humped patterns; othersshow either greater or lesser depletion in LREE. [Th], [U], [Nb] and [Ta]are higher overall, and the range in Th/U is greater; Ti anomalies, bothpositive and negative, also show a greater range.

The flatMREE-HREE patternswith LREE depletion are reminiscent ofthe REE patterns seen in e.g. N-MORB, but the anomalies in Ce and Ti,and the high contents of Nb, Ta, Th and U are not typical of MORB. Itis, in fact, difficult to find primary mantle melts with similar patterns.However, such patterns might develop through extended fractionalcrystallization of magmas with less pronounced REE patterns. The ba-salts of N. Israel and adjacent areas, from Jurassic through lower Creta-ceous and Neogene time, are alkali basalts with remarkably constanttrace-element patterns and Sr\\Nd isotopes (Stein and Hofmann,1992; Weinstein et al., 2006). Their trace-element patterns (Fig. 10a)have mildly negative slopes from Yb to La, with [Th], [U], [Nb] and [Ta]at ca 100 times chondrites. They have no anomalies in Eu, Ce or Ti;CN-normalized P/Yb is ca 0.5.

Compared to these probable primary basalts, the strong depletion ofLREE in the calculated melts may reflect crystallization of late phasessuch as apatite andmonazite, and theHREE depletion could reflect crys-tallization of both zircon and xenotime. Removal of apatite also can pro-duce positive Eu anomalies, counteracting the influence of plagioclasecrystallization. The depletion of P relative to the HREE (P/Yb ca 0.1)seen in all of the melts, and the wide variation in Th and U, also maybe tied to the crystallization of monazite and xenotime, as well as apa-tite. These comparisons suggest that the calculated melts, and the zir-cons crystallized from them, represent late-stage differentiates ofbasaltic melts cooling slowly in deep-seated magma chambers.

The zircons erupted from Upper Cretaceous to Neogene time arebroadly similar to the Permo-Jurassic ones in having strong LREE deple-tion, and high Th\\Ta (Fig. 10). However, they show generallymore en-hanced Ce anomalies, smaller Eu anomalies (Fig. 7b,c), andprogressively develop mild HREE depletion relative to the MREE,

producing concave-downward REE patterns. There is also a tendencytoward deeper negative Ti anomalies, especially in the youngest zircons,and toward Th/UCN values b1.

3.5. Sampling patterns in the cretaceous volcanism

The group of zirconswith ages in theUpper Cretaceous (116–95Ma)may give an indication of the duration of this episode at depth, eventhough the range of eruption ages is shorter. As noted above, it is likelythat none of the Cretaceous zircons truly date their respective eruption,but rather represent xenocrysts derived from roughly coeval, more dif-ferentiated, magmatism at depth. The range of 176Hf/177Hf values in theCretaceous zircons corresponds closely to that of the Permian-Jurassicpopulations (Fig. 8), suggesting the Cretaceous magmas were derivedfrom the same reservoir.

The distribution pattern of the older populations in the different Cre-taceous vents gives a picture of this reservoir (Fig. 11). The Permian-Tri-assic (280–220 Ma) zircons are most abundant in the main vents(Rakefet, Bat Shelomo, Makura and the Ein Hashofet centre in theRamot Menashe syncline). The Triassic-Jurassic (Asher) zircons arefound mixed with Permian-Triassic ones in Ein Hashofet and Rakefet,but make up the bulk of the sample in the northern bodies (Beit Oren,Karem Maharal and Har Alon); the younger zircons are found mainlyin Karem Maharal, Tavassim and Migdal Ha-Emeq on the western sideofMt. Carmel. This pattern suggests that the Permo-Triassicmagmatismbuilt up a “pillow” beneath the core of theMt. Carmel-Menashe area, sothat the magmas of the younger episodes tended to intrude toward themargins of this central area (Fig. 12).

Segev and Rybakov (2011) used data from deep drillings and geo-physical surveys (primarily magnetics and gravity) to outline areasshowing ‘significant influence of magmatic bodies”. In this synthesis,the Late Triassic-Jurassic magmatism is shown extending along the cur-rent coastline both NE and SW of theMt. Carmel area, while Permo-Tri-assic magmatism is isolated farther south. However, the data presentedin this paper indicate that the Permo-Triassic magmatism also was sig-nificant beneath the Mt. Carmel area and probably contributes to thecomplex geophysical signature of the area.

The Permian-Jurassic volcanic rocks reflect the formation of the Le-vant continental margin by rifting during the breakup of Gondwana,which produced the Neotethys and Mesotethys oceans. Segev (2000)listed 17 globally synchronous short-term volcanic cycles (200 to 5

Fig. 11. Map with distribution of zircon age populations in vents in Yizre'el Valley-Mt Carmel area. Lines A-A', B-B′, C-C′ mark corresponding sections 1-3 in Fig. 12.

Fig. 12. Cartoon showing the relative abundance of three zircon age populations indifferent vents, to illustrate the general structure of the proposed gabbroic underplate atthe crust-mantle boundary beneath the Mt. Carmel area. Not to scale.

321W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

Ma) that are expressed in thiswider region, and proposed that they rep-resent pulses of plume activity. An alternative interpretation would bethat they reflect periods of tectonic re-adjustment during which thin-ning and mantle upwelling led to the production of new melts fromthe proposed fossil plume head of Stein and Hofmann (1992), while co-eval rifting along old fracture systems allowed themagmas access to theupper lithosphere and the surface.

In this framework, the concentration of the Cenomanian volcanism(97–80 Ma) into the Mount Carmel area may reflect its extensive net-work of translithospheric and crustal-scale high-angle faults, most ofwhich were reactivated in Miocene time. We cannot know the real ex-tent of the Permian-Jurassic underplating beneath this area; it may ex-tend far beyond the 150–300 km2 sampled by the known Cretaceousvents, and connect with the area identified by Segev and Rybakov(2011). However, it seems likely that the same set of translithosphericstructures, dating back to the initial rifting from Gondwana, has local-ized magmatism beneath this area for at least 250–300 million years.

4. Conclusions

Zircon xenocrysts fromCretaceous pyroclastic deposits onMt. Carmeland associated alluvial deposits document extensive magmatic activitybeneath the area during Permo-Triassic and Late Triassic-Jurassic time.Themorphology and internal structures of thegrains suggest amagmaticorigin, followed by recrystallization, overgrowth and brecciation. Trace-element data indicate that the parental magmas were alkaline tocarbonatitic, and probably represent residual differentiates of basalticmelts. Hf-isotope ratios indicate derivation of these melts from anenriched mantle (OIB-like) source, probably mid- to Neoproterozoic inage. This may represent the plume head postulated by Stein and co-workers as the source of basalts across the region. The distribution ofthe different age populations across Mt. Carmel and neighbouring areassuggests that the earliest intrusions built up an underplated “pillow”,probably near the crust-mantle boundary, and that the emplacement ofthe later intrusions occurred primarily around the edges of the pillow.

322 W.L. Griffin et al. / Lithos 314–315 (2018) 307–322

The Permian to Jurassic zircons record the repeated intrusion of magmasderived from a common source, and the (repeated?) recrystallization ofolder zircons in the presence of fluids with relatively light O-isotopes.The Cretaceous zircons have Hf-isotope ratios within the range of theolder ones, suggesting that most of the Cretaceous zircon ages representPb-loss, or recrystallization and remobilization of older zircons, duringthe Cretaceous thermal event. This may indicate that relatively little Cre-taceousmagmaponded at the crust-mantle boundary and evolved to thepoint where they could crystallize zircons. This appears to be consistentwith their more explosive nature. In contrast, many of the younger (Eo-cene to Miocene) zircons have more juvenile Hf-isotope characteristics,suggesting the derivation of magmas both from the older “plume-re-lated” source and from a more primitive convecting mantle.

Acknowledgements

We thank Dr. E. Sass and Prof. Oded Navon for useful discussions onthe geology of Mt. Carmel and the volcanism of Israel in general. Prof.Xian-Hua Li (IGG-CAS, Beijing) kindly facilitated O-isotope analyses bySIMS. We are also grateful to Dr. Joseph Guttman, Chief Hydrologist ofMekorot, for providing the core from the EinHa-Shofet drilling. The pro-ject benefitted greatly from the skill and dedication of the Shefa Yamimoperational staff at Akko and especially the mineral sorters. MichelGregoire and an anonymous referee provided helpful suggestions to im-prove the MS. This study used instrumentation funded by ARC LIEF andDEST Systemic Infrastructure Grants, Macquarie University and indus-try. This is contribution 1174 from the ARC Centre of Excellence for Coreto Crust Fluid Systems (http://www.ccfs.mq.edu.au) and 1235 in theGEMOC Key Centre (http://www.gemoc.mq.edu.au).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.lithos.2018.06.007.

References

Apter, D.B., 2014. High pressure indicator minerals from the Rakefet magmatic complex(RMC). Mt. Carmel, Israel, Kimberley Diamond Symposium, Kimberley, South Africa.

Bein, A., 1974. Reef developments in the Judea Group from the Coastal Plain and the Car-mel, Israel. Geol. Surv. Isr. 74.

Belousova, E., Griffin,W., Shee, S., Jackson, S., O'Reilly, S., 2001. Two age populations of zir-cons from the Timber Creek kimberlites, northern territory, as determined by laser-ablation ICP-MS analysis. Aust. J. Earth Sci. 48 (5), 757–765.

Ben-Avraham, Z., Hall, J.K., 1977. Geophysical survey of Mount Carmel structure and itsextension into the eastern Mediterranean. J. Geophys. Res. 82 (5), 793–802.

Dvorkin, A., Kohn, B., 1989. The Asher Volcanics, northern Israel: petrography, mineral-ogy, and alteration. Isr. J. Earth Sci. 38 (2–4), 105–123.

Esperança, S., Garfunkel, Z., 1986. Ultramafic xenoliths from the Mt. Carmel area (KaremMaharal volcano), Israel. Lithos 19 (1), 43–49.

Garfunkel, Z., 1989. Tectonic setting of phaneroozoic magmatism in Israel. Isr. J. Earth Sci.38 (2–4), 51–74.

Griffin, W.L., et al., 2000. The Hf isotope composition of cratonic mantle: LAM-MC-ICPMSanalysis of zircon megacrysts in kimberlites. Geochim. Cosmochim. Acta 64, 133–147.

Grimes, C.B., et al., 2007. Trace element chemistry of zircons from oceanic crust: a methodfor distinguishing detrital zircon provenance. Geology 35 (7), 643–646.

Grimes, C., Wooden, J., Cheadle, M., John, B., 2015. “Fingerprinting” tectono-magmaticprovenance using trace elements in igneous zircon. Contrib. Mineral. Petrol. 170(5–6), 46.

Gvirtzman, G., Klang, A., Rotstein, Y., 1990. Early Jurassic shield volcano belowMount Car-mel: new interpretation of the magnetic and gravity anomalies and implication forearly Jurassic rifting. Isr. J. Earth Sci. 39 (2/4), 149–159.

Kaminchik, J., Segev, A., Katzir, Y., 2014. The Origin of Intraplate Alkaline MaficMagmatism in Continental Shelves: Lavas and Xenoliths from the Upper CretaceousVolcanos of Mt Carmel. (Unpublished MSc thesis). Beer Sheva University, Israel.

Kinny, P., Compston, W., Bristow, J., Williams, I., 1989. Archaean mantle xenocrysts in aPermian kimberlite: two generations of kimberlitic zircon in Jwaneng DK2, southernBotswana. Kimberlites Relat. Rocks 2, 833–842.

Laws, E.D., Wilson, M., 1997. Tectonics and magmatism associated with Mesozoic passivecontinental margin development in the Middle East. J. Geol. Soc. 154 (3), 459–464.

Lewy, Z., 1991. Periodicity of cretaceous epeirogenic pulses and the disappearance of thecarbonate platform facies in late cretaceous times (Israel). Isr. J. Earth Sci. 40 (1–4),51–58.

Loucks, R.R., Fiorentini, M.L., 2018. New magmatic oxybarometer using trace elements inzircon reveals oxidation states of Hadean and Eoarchean lithosphere in four cratons.Geochim. Cosmochim. Acta (in press).

Mittlefehldt, D.W., 1986. Petrology of high pressure clinopyroxenite series xenoliths,Mount Carmel, Israel. Contrib. Mineral. Petrol. 94 (2), 245–252.

Morag, N., Avigad, D., Gerdes, A., Harlavan, Y., 2012. 1000–580Ma crustal evolution in thenorthern Arabian-Nubian shield revealed by U–Pb–Hf of detrital zircons from lateNeoproterozoic sediments (Elat area, Israel). Precambrian Res. 208-211, 197–212.

Nardi, L.V.S., Formoso, M.L.L., Muller, I.F., Fontana, E., Jarvis, K., Lamarao, C., 2013. Zircon/rock parition coefficients of REEs, Y, Th, U, Nb and Ta in granitic rocks: Uses for prov-enance and mineral exploration purposes. Chem. Geol. 335, 1–7.

Rosenfeld, A., Raab, M., 1984. Lower cretaceous ostracodes from Israel and Sinai. Isr.J. Earth Sci. 33 (3), 85–134.

Rozenbaum, A., et al., 2016. 40Ar/39Ar chronostratigraphy of late Miocene–early Pliocenecontinental aquatic basins in SE Galilee, Israel. GSA Bull. 128 (9–10), 1383–1402.

Sass, E., 1980. Late cretaceous volcanism inMount Carmel, Israel. Isr. J. Earth Sci. 29 (1–2),8–24.

Segev, A., 2000. Synchronous magmatic cycles during the fragmentation of Gondwana:radiometric ages from the Levant and other provinces. Tectonophysics 325 (3–4),257–277.

Segev, A., 2002. Flood basalts, continental breakup and the dispersal of Gondwana: evi-dence for periodic migration of upwelling mantle flows (plumes). EGU StephanMueller Spec. Publ. Ser. 2, 171–191.

Segev, A., 2005. Phanerozoic magmatic activity associated with vertical motions in israeland the adjacent countries. In: Hall, J.K., Krasheninnikov, V.A., Hirsch, F., Benjamini, C.,Flexer, A. (Eds.), Geological Framework of the Levant Vol II the Levantine Basin andIsrael, pp. 553–574.

Segev, A., 2009. 40Ar/39Ar and K–Ar geochronology of Berriasian–Hauterivian andCenomanian tectonomagmatic events in northern Israel: implications for regionalstratigraphy. Cretac. Res. 30 (3), 810–828.

Segev, A., Rybakov, M., 2011. History of faulting and magmatism in the Galilee (Israel)and across the Levant continental margin inferred from potential field data.J. Geodyn. 51 (4), 264–284.

Segev, A., Sass, E., 2009. The geology of the Carmel region—Albian–Turonian volcano-sed-imentary cycles on the northwestern edge of the Arabian platform. Geological Surveyof Israel Report GSI/7/2009 (Jerusalem, Israel) (Hebrew with English abstract).

Segev, A., Weissbrod, T., Lang, B., 2005. 40Ar/39Ar dating of the Aptian–Albian igneousrocks in Makhtesh Ramon (Negev, Israel) and its stratigraphic implications. Cretac.Res. 26 (4), 633–656.

Shaliv, G., 1991. Stages in the tectonic and volcanic history of the Neogene basin in thelower galilee and the valleys. Rep. GSI/11, 91: 94.

Stein, M., Goldstein, S.L., 1996. From plume head to continental lithosphere in the Ara-bian-Nubian shield. Nature 382 (6594), 773–778.

Stein, M., Hofmann, A.W., 1992. Fossil plume head beneath the Arabian lithosphere?Earth Planet. Sci. Lett. 114 (1), 193–209.

Stein, M., Hofmann, A.W., 1994. Mantle plumes and episodic crustal growth. Nature 372(6501), 63–68.

Tretiakova, I.G., et al., 2017. Recurrent magmatic activity on a lithosphere-scale structure:crystallization and deformation in kimberlitic zircons. Gondwana Res. 42, 126–132.

Weinstein, Y., 2000. Spatial and temporal geochemical variability in basin-related volca-nism, northern Israel. J. Afr. Earth Sci. 30 (4), 865–886.

Weinstein, Y., Navon, O., Altherr, R., M., S., 2006. The role of lithospheric mantle heteroge-neity in the generation of Plio-Pleistocene alkali basaltic suites from NW Harrat ashShaam (Israel). J. Petrol. 47 (5), 1017–1050.

Woodhead, J., Hergt, J., Giuliani, A., Phillips, D., Maas, R., 2017. Tracking continental-scalemodification of the Earth's mantle using zircon megacrysts. Geochem. Perspect. Lett.4, 1–6.