Early Miocene calc-alkaline felsic tuffs within deep-marine turbidites in the Kyrenia Range, north Cyprus, with a possible post-collisional eruptive centre in western Anatolia

Abstract

Felsic tuff as a direct fallout deposit is known from one small area in the Kyrenia Range, north Cyprus, within deep-sea terrigenous turbidites. Nearby tuffaceous siltstones contain compositionally similar felsic volcanic rocks (c. 5–10%), mixed with terrigenous material. Sedimentary evidence indicates that the fallout tuff was variable reworked locally, whereas the tuffaceous siltstones are interpreted as turbidites mixed with terrigenous material derived from Anatolia. U–Pb dating of zircons that were extracted from a sample of relatively homogeneous tuff yielded a dominant age of 16.64 ± 0.12 Ma (Burdigalian). Zircon trace-element analysis indicates predominant derivation from within-plate-type felsic magma. Whole-rock chemical analysis of the tuffaceous sediments as a whole is compatible with a felsic arc source, similar to the post-collisional magmatism within Anatolia. Regional comparisons suggest that the nearest volcanism of similar age and composition is located c. 500 km away, within the Kırka area (Eskişehir region) of the Western Anatolia Volcanic Province. Evidence of tephra dispersal in the western Mediterranean region and climatic modelling suggests E-wards prevailing winds and therefore tephra transport over southern Anatolia and adjacent areas during early Miocene time. The north Cyprus tuffs could represent powerful Minoan (Plinian)-type eruptions in western Anatolia, coupled with SE-wards tephra transport during and soon after the onset of post-collisional magmatism.

1. Introduction

There is increasing interest in subduction and collision-related magmatism, especially concerning its recent and ancient societal impacts (e.g. Sparks, 2003; Loughlin et al. 2015). To characterize a modern volcano, it is necessary to understand its geometry, volume, historical development, petrological and chemical characteristics, and also its ejected fragmental material including local gravity flows and further-travelled tephra (e.g. Mount St Helens, Cascade; Evarts et al. 1987). Tephra layers, our present subject, provide age markers (tephrochronology) and event horizons (tephrostratigraphy) (Lane et al. 2017) and can provide regional to global correlations of volcanism (Shane, 2000; Harangi et al. 2005; Lowe et al. 2017; Petrelli et al. 2017; Chen & Robertson, 2019).

For ancient volcanoes, complete characterization can be difficult because of erosion or burial, such that the link between the volcanic centre and dispersed tephra may be lost. In well-studied areas, such as the NW Pacific or Central America, extensive geochemical data on subaerial volcanoes, combined with integrated studies of volcanic ash from deep-sea cores, allows volcanoes to be linked to far-travelled tephra in space and time (Scudder et al. 2016). Major- and trace-element analysis of volcanic glasses are particularly useful to pinpoint specific volcanic centres (Kutterolf et al. 2018; Schindlbeck et al. 2018). Whole-rock geochemical data of tephra successions are also useful, especially for identifying long-term trends and the relative contributions of different source materials (e.g. volcanogenic versus terrigenous) (Scudder et al. 2016; Robertson et al. 2018). Similarly, in the Mediterranean Sea, geochemical data from tephra in deep-sea cores has been linked to specific volcanic centres and eruptive events (Clift & Blusztajn, 1999). Study is more complicated where both the source volcanoes and the associated tephra are on land, especially for older lithologies that may be eroded, diagenetically altered or metamorphosed (Cerling et al. 1985).

In the circum-Mediterranean region, our present study area, there is extensive but far from complete documentation of the geochemistry of major volcanic centres (e.g. Schleiffarth et al. 2018). However, until now little compositional data were available for volcanic glasses to compare dispersed tephra with eruptive sources (Pearce et al. 2002, 2007). Comparisons of tephra with potential source volcanoes in this region therefore have to rely mainly on whole-rock geochemical data.

For any dispersed tephra, it is useful to characterize the physical characteristics, sedimentology and chronostratigraphy of the deposit, which can be achieved by petrography, mineralogy and both palaeontological and radiometric dating (Fisher & Schmincke, 1984; Carey & Schneider, 2011; Lowe, 2011). The alteration-resistant accessory mineral (e.g. zircon) allows accurate U–Pb dating, particularly of felsic volcanic products. The chemistry of zircon also aids correlation of dispersed tephra with potential eruptive centres (e.g. Aydar et al. 2012; Baresel et al. 2017).

Miocene tuffs and tuffaceous sediments are locally exposed in the Kyrenia Range, northern Cyprus (Baroz, 1979) and provide an opportunity to investigate a link between dispersed tephra and a possible volcanic source. The northern part of Cyprus was located in a deep-water setting during the Miocene Epoch, close to its present position adjacent to the southern margin of Anatolia (Fig. 1a, b). The tuffs are associated with siliciclastic turbidites that were derived from southern Anatolia and accumulated to the south of this landmass (McCay & Robertson, 2012; Chen et al. 2019; Shaanan et al. 2020).

Fig. 1. (a) Simplified tectonic map of the Eastern Mediterranean and (b) Geological map of the Kyrenia Range, northern Cyprus after Robertson et al. (2012). The two tuffaceous sampling sites are marked by small red boxes.

Within Anatolia, intra-continental magmatism followed suturing of Neotethyan ocean basins during latest Cretaceous – Palaeogene time (e.g. Schleiffarth et al. 2018). Eruptions occurred repeatedly during early Miocene – Quaternary time, becoming generally younger eastwards (Dilek & Altunkaynak, 2010; Schleiffarth et al. 2018).

Here, we provide new sedimentological, petrographic and/or geochemical, and radiometric age evidence for felsic tuffs and tuffaceous siltstones in the eastern Kyrenia Range. We use the combined evidence to infer the timing, chemical composition and possible source of the tuffs and tuffaceous sediments. We identify a possible post-collisional eruptive source in western Anatolia, contributing to knowledge of post-collisional volcanism in the region.

2. Geological background

Anatolia is situated in the western segment of the Arabia–Eurasia collision zone (Fig. 1a). Three collision-related volcanic provinces are recognized within Anatolia, mainly based on geographic location, chemical composition and age (Innocenti et al. 1982; Pearce et al. 1990; Le Pennec et al. 1994; Schleiffarth et al. 2018): (1) the Western Anatolian Volcanic Province, in the İzmir-Afyon-Isparta area (western Turkey), is characterized by high-K, calc-alkaline andesites and dacites that are associated with felsic ignimbrites and dated at 21–10 Ma (Aquitanian–Tortonian) (Innocenti et al. 1975; 1982; Keller, 1983); (2) the Central Anatolian Volcanic Province, in the Ankara-Karaman-Kırşehir region (central Turkey), includes typical calc-alkaline volcanics, namely, ignimbrites, volcanogenic sediments and subordinate lavas, beginning at c. 10 Ma (Late Miocene) (Innocenti et al. 1975; Pasquare et al. 1988); and (3) the Eastern Anatolian Volcanic Province, in eastern Turkey, Armenia and NE Iran, is represented by volcanism dated as from c. 11 Ma to 17th century AD (middle Miocene to Holocene) of both calc-alkaline and alkaline affinities (Pearce et al. 1990; Şengör et al. 2008).

Collision-related tuffaceous products are likely to have been deposited around the periphery of Anatolia in the Mediterranean and Black seas, but remain largely unknown. Exceptionally, felsic tuffs and tuffaceous sediments are exposed in the Kyrenia Range as a result of the Pleistocene uplift (Kinnaird & Robertson, 2013; Palamakumbura et al. 2016). The tuffs and tuffaceous sediments occur within the Panagra (Geçitköy) Formation (Fig. 2), which is exposed throughout the Kyrenia Range (Baroz, 1979; Robertson & Woodcock, 1986; Hakyemez et al. 2000; McCay & Robertson, 2012). (We use English stratigraphical names where possible; in cases where Greek and Turkish names are synonymous, the latter is stated in parentheses at first occurrence.) The Panagra Formation, c. 50–100 m thick, begins with green to grey, fine-grained hemipelagic limestone, rich in planktic foraminifera. The succession grades upwards into a distinctive interval of red to brown calcareous mudrock (marl), together with thin interbeds of siltstone/sandstone turbidites containing siliciclastic and biogenic detritus (Robertson & Woodcock, 1986; McCay & Robertson, 2012). Strontium analysis and planktic foraminiferal dating indicate a Burdigalian–Langhian (middle Miocene) age (Baroz, 1979; McCay et al. 2013) for the Panagra Formation. The tuffs and tuffaceous sediments are only recorded within the Panagra Formation in the eastern part of the Kyrenia Range (Fig. 1b).

Fig. 2. Summary log showing the age and stratigraphy of the succession that includes tuffaceous sediments within the Panagra (Geçitköy) Formation, highlighted with red arrow (data from Robertson & Woodcock, 1986; McCay & Robertson, 2012; McCay et al. 2013).

3. Methods

3.a. Analytical objectives

Assuming the age of zircon crystallization is synchronous with volcanic eruption and ash-bed deposition (e.g. Bowring et al. 1998), U–Pb zircon geochronology can be used to determine the age of the eruption from its dispersed tephra and help to identify volcanic centres of appropriate age. It is assumed that far-travelled tephra layers retain a similar mineral composition to their source eruption within an entire proximal, to distal ash horizon (e.g. Ovtcharova et al. 2015).

In addition, whole-rock geochemical analyses of tuff, tuffaceous and non-tuffaceous sediments were carried out with the objective of evaluating the contribution of tephra compared with terrigenous and biogenic constituents. The whole-rock chemical analysis also aimed to identify possible volcanic centres in the region, potentially including western Anatolia where violently explosive volcanic products (e.g. felsic ignimbrites) are relatively well-documented (e.g. Seghedi & Helvacı, 2016).

3.b. Zircon geochronology and geochemistry

Zircon grains were separated from the tuffs of the Panagra Formation (sample no. 14–37) using standard gravitational separation techniques. Zircon grains were randomly picked under a binocular microscope. The grains, together with zircon U–Pb standard 91500 (c. 1062.5 Ma; Wiedenbeck et al. 1995) were then cast in epoxy mounts and polished 1/3–1/2 to expose the grain interior. The morphology and internal microstructure were observed and imaged by cathodoluminescence (CL) prior to analyses. U–Pb analysis (grain no. 1–10) was performed on a Cameca ims-1270 secondary ion mass spectrometer (SIMS) at the School of GeoSciences, University of Edinburgh, using the methods detailed by Kelly et al. (2008) and Ustaömer et al. (2012). Each analysis was c. 27 minutes in duration (including a preliminary 2 min, 15 μm raster across the analysis site) and employed a 4 nA primary O²⁻ ion beam current and Köhler illumination to produce a spot c. 20 μm in diameter on the sample. Oxygen flooding increased the Pb ion yield by a factor of c. 2. Isotope ratios were measured in 20 cycles; the first five cycles were excluded in order reduce possible near-surface contamination of common lead. Additional zircon U–Pb analyses (grain no. 11–34) were undertaken on a laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS) at the Beijing GeoAnalysis Co. Ltd. Laser sampling was performed using a Resolution SE model laser ablation system, coupled to an Agilent 7900 ICP-MS to increase the quantitative abundance. Pre-ablation was conducted for each spot analysis using 5 laser shots (c. 0.3 μm in depth) to remove potential surface contamination. The analysis was performed using a 30 μm diameter spot at 5 Hz and a fluence of 2 J cm^–2. Iolite software package was used for data reduction (Paton et al. 2010). Zircon 91500 and GJ-1 (c. 604 Ma; Jackson et al. 2004) were used as the primary and the secondary age reference materials. Zircon 91500 and GJ-1 were analysed twice and once every 10–12 analyses of the sample, respectively. Analytical uncertainties in the calculated ages are quoted as ±1σ. The results were processed using Isoplot/Ex version 3.75 (Ludwig, 2012). Zircon trace elements were acquired simultaneously with the U–Pb isotopic data. The National Institute of Standards and Technology Standard Reference Material (NIST-SRM) 610 glass and ⁹¹Zr were used to calibrate the trace-element concentrations as external reference material and internal standard element, respectively. All of the analytical results are listed in online Supplementary Tables S1–S3 (available available at http://journals.cambridge.org/geo).

3.c. Whole-rock geochemistry of tuff and tuffaceous/non-tuffaceous sediments

Three samples of relatively homogeneous tuff were selected for whole-rock X-ray fluorescence (XRF) at the School of GeoSciences, University of Edinburgh, using the methods of Fitton et al. (1998) and Fitton & Godard (2004). Accuracy and precision were typically c. 5%. Additional trace and rare earth elements (REEs) were analysed at the ACME Laboratories, Vancouver by ICP-MS. Trace-element contents were determined from a LiBO₂ fusion by ICP-MS by using 5 g of sample pulp. Detection limits were c. 0.01–0.04 wt% for major oxides, and 0.01 and 0.1 ppm for trace and rare earth elements. The relative standard deviation for the REEs is c. 5% and up to 10% for all other trace elements with quality control using international geostandards (see http://acmelab.com). In addition, whole-rock major- and trace-element (including REEs) analyses of two samples of tuffaceous siltstone–sandstone were conducted by XRF and also with an Agilent 7700e ICP-MS at the Wuhan Sample Solution Analytical Technology Co. Ltd., Wuhan, China. Standards BHVO-2, AGV-2, BCR-2 and RGM-2 were used to ensure analytical precision. The uncertainties are 1–5% for elemental abundances of > 1 ppm and 5–10% for abundances of < 1 ppm. The analytical data for the major, trace and rare earth elements are listed in online Supplementary Table S4 (available at http://journals.cambridge.org/geo).

4. Results

4.a. Field occurrences

E–W-striking red mudstones of the Panagra Formation, c. 50 m thick (Figs 1b, 2) are exposed near Çınarlı (Platani) in the eastern Kyrenia Range (Fig. 3a, b). These sediments grade upwards into a pale-coloured interval of tuff, c. 8 m thick. This is followed by red calcareous marl and/or mudstone. The Panagra Formation is, in turn, overlain by medium- to thick-bedded sandstone turbidites of the Davlos (Kaplıca) Formation (Fig. 3b).

Fig. 3. Field occurrences of the tuffaceous deposits in northern Cyprus. (a) Exposure of felsic tuff near Çınarlı (Platani); (b) Sketch of section of the tuff (see (a) for field location); (c) Field photographs of the tuffs (white), including thin mudstone interbeds (yellow-brown); inset: repeated tuffaceous interval with sharp sandy based and tops; and (d) Measured log of tuffaceous siltstone–sandstone turbidites, near Tirmen (Trypimeni).

The relatively homogeneous tuff interval near Çınarlı (Platani) consists of repeated depositional units (Fig. 3c): (1) well-bedded normal-graded units (up to 10 cm thick) with sharp, scoured sandy bases, grading upwards into well-sorted, parallel, planar or wavy-laminated intervals, and then into silty or fine-grained pale tuff; (2) thin-bedded (2–5 cm thick), poorly sorted mixtures of fine-grained tuff and sand-sized materials, commonly with sharp sandy or silty bases (< 5 mm) and, in places, relatively sharp tops; and (3) fine-grained, generally massive or weakly parallel-laminated, homogeneous tuff (5–10 cm).

In addition, near Tirmen (Trypimeni), c. 11 km further west (Fig. 1b), the Panagra Formation includes a c. 2–5 m thick interval of well-bedded, normal-graded sandstone–siltstone turbidites. There are several well-lithified interbeds (mostly 3–5 cm thick, but up to 10 cm thick) of white to pale brown tuffaceous siltstone–sandstone (Fig. 3d).

4.b. Petrography

In thin-section, the relatively pure tuff from near Çınarlı (Platani) (Fig. 4a) is fine- to medium-grained and well-sorted, with abundant volcanic glass (ash-sized) (c. 50%) (Fig. 4b1–b4), together with quartz (c. 40%), muscovite (c. 5%), hornblende (c. 1%), feldspar (mainly plagioclase) (c. 1%) opaque grains (Fig. 4a) and scattered planktic foraminifera (Fig. 4c). In contrast, the tuffaceous sediments from near Tirmen (Trypimeni) comprise a mixture of terrigenous and tuffaceous material (c. 5–10%) (Fig. 4d), namely volcanic glass, quartz, plagioclase, muscovite and biotite. Volcanic glass shards in the tuffaceous sediments have been partly dissolved and replaced with clays. However, glass in the homogeneous tuff remains relatively fresh.

Fig. 4. (a–c) Photomicrographs of the tuffs (plane-polarized light) and (d) Tuffaceous siltstone (cross-polarized light). (a) Abundant colourless volcanic glass; (b1–b4) Enlargements of panel a showing different ash morphologies; (c) Enlargement of (a) showing Orbulina sp. (planktic foraminifera) in tuff; and (d) Tuffaceous siltstone including sub-angular quartz crystals, opaque grains, mica (altered) and rare plagioclase grains. Q – quartz; Vg – volcanic glass; Ms – muscovite; Pl – plagioclase; Bt – biotite; F – foraminifera; P – pumiceous; C – cuspate; Fr – frothy; B – blocky; T – tabular. Scale bar: 100 μm.

4.c. Zircon U–Pb geochronology

Typically needle-like, 80–200 μm-long grains of zircon were separated from a sample of relatively homogeneous fine-grained tuff (sample no. 14–37) from the section near Çınarlı (Platani). Most crystals show magmatic-type concentric zoning, as revealed by CL images (Fig. 5a). Thirty-four analyses were obtained from 34 zircon grains, of which 20 were concordant (90–110% concordance) and 14 discordant. The pre-Miocene zircons (aged 37–80 Ma) are generally affected by inclusions or cracks and are all discordant. Concordant analyses of both bright cores (sample no. 21) or dark cores (sample no. 4, 25), and also of the rims of core–rim structure (sample no. 6), all yielded Precambrian ages ranging from 678.8–2505.0 Ma. These zircons are interpreted as recycled sedimentary grains (sample no. 4, 6, 21) or the cores of older zircons (sample no. 25). Of the concordant 16 zircons, eight grains define a tight age cluster, yielding a mean ²⁰⁶Pb/²³⁸U age of 16.64 ± 0.12 Ma with a mean square weighted deviation (MSWD) of 1.13 (Fig. 5b, c). This is inferred to represent the crystallization age of the felsic tuff source magma. Other concordant zircon grains yielded slightly older ages of 17–21 Ma (Fig. 5b). These ages are interpreted as earlier volcanic events in the source area. For comparison, the palaeontologically determined age of the Panagra Formation as a whole is Burdigalian–Langhian (16.95–15.61 Ma) based on planktic foraminiferal and nannofossil evidence and also Sr isotope dating (Baroz, 1979; McCay et al. 2013). The overall radiometric age data therefore suggest that the zircons from the dated sample contain a mixture of contemporaneous and older, reworked volcanic material.

Fig. 5. (a) Cathodoluminescence images of zircon grains analysed from the tuff sample (sample no. 14–37) near Çınarlı (Platani). Locations of the measured spots and the corresponding ages (²⁰⁶Pb/²³⁸U ± 1σ) are indicated. Yellow circle (SIMS): 20 μm; red circle (LA-ICP-MS): 30 μm. (b) Wetherill Concordia diagrams for the Miocene age population. (c) ²⁰⁶Pb/²³⁸U weighted mean diagram for the tight zircon age cluster.

4.d. Zircon trace-element compositions

The Miocene zircons, together with most others, have REE patterns that increase steeply from La to Lu, with a pronounced positive Ce anomaly and a negative Eu anomaly (Fig. 6). This is consistent with the normal association of zircon with heavy REE relative to light REE (Hanchar & van Westrenen, 2007) in magma that is typically oxidizing (Smythe & Brenan, 2015), and with concurrent feldspar crystallization (Rubatto, 2002). The pre-Miocene zircons (aged 37–80 Ma) are characterized by enrichments in light REE (La, Ce) compared with the Miocene zircons (Fig. 6); this could have resulted from contamination by mineral inclusions (Zhong et al. 2018). The Precambrian zircons are variable, characterized by middle REE enrichment (no. 31) or heavy REE depletion (no. 25) (Fig. 6). The analyses of the middle REE-enriched zircon (no. 31) are likely to be affected by inclusions (i.e. titanite). The lesser heavy REE enrichment in grain no. 25 is consistent with a metamorphic origin (Th/U = 0.05; Rubatto, 2002).

Fig. 6. REE concentrations normalized to chondrite (Nakamura, 1974) for the zircons in tuff sample (sample no. 14–37).

In addition, the zircon crystals are characterized by a relatively wide range of U (71.5–21820 ppm) and Th (17.3–3678 ppm) concentrations (see online Supplementary Tables S1–S3). Smaller grains (e.g. sample no. 15, 33) and also the single grain with a high uranium concentration (> 10 000 ppm; no. 17) are likely to have lost lead preferentially, such that no concordant ages can be calculated.

4.e. Whole-rock geochemistry

The analysed values of some key elements are as follows (see online Supplementary Table S4):

Relatively pure tuff from Çınarlı (Platani): SiO₂, 69.4–70.9 wt%; Al₂O₃, 12.7–12.9 wt%; Fe₂O₃, 1.4–1.7 wt%; CaO, 1.0–1.3 wt%; TiO₂, 0.08 wt%; Ba, 63–86 ppm; Ce, c. 46 ppm; U, c. 25 ppm; Th, 47–51 ppm; Nb, 31–33 ppm; Sr, 58–85 ppm; and Rb, 244–256 ppm.

Tuffaceous siltstone from Tirmen (Trypimeni): SiO₂, 32.2–38.3 wt%; Al₂O₃, 10.2–12.4 wt%; Fe₂O₃, 0.5–1.3 wt%; CaO, 10.7–17.1 wt%; TiO₂ c. 0.16 wt%; Ba, 71,554–78,898 ppm; Ce, 35–63 ppm; U, 4–5 ppm; Th, 23–28 ppm; Nb, 11–12 ppm; Sr, 223–649 ppm; and Rb, 15–23 ppm.

Non-tuffaceous sandstone (from both locations): SiO₂, 28.8–30.3 wt%; Al₂O₃, 1.5–1.9 wt%; Fe₂O₃, 0.9–1.3 wt%; CaO, 32.0–34.8 wt%; TiO₂, 0.1 wt%; Ba, 126 ppm; Ce, 21.0 ppm; U, 1.9 ppm; Th, 2.0 ppm; Nb c. 2.3 ppm; Sr c. 324–516 ppm; Rb, 12–17 ppm.

The relatively high loss-on-ignition (LOI) values (average 6.08 wt%) of the pure tuff are likely to represent secondary alteration processes, for example, partial devitrification of glass. Higher LOI values (16–19 wt%) occur in the tuffaceous siltstone possibly due to fluid alteration, which may also have led to the observed concentration in Ba (up to 78,898 ppm). The most calcareous non-tuffaceous sandstones are rich in carbonate grains and calcite cement (McCay & Robertson, 2012; Chen & Robertson, 2020) and, as expected, have the highest LOI values (up to 29 wt%) as a result of loss of CO₂.

The chemical data are plotted on several tried and tested geochemical plots that are indicative of source composition, provenance, sorting and/or diagenesis. On a chondrite-normalized REE plot (Fig. 7a), the non-tuffaceous sandstone turbidites have a typical terrigenous composition. The relatively pure tuff samples are marked by Eu depletion that is attributed to source-melt plagioclase fractionation (Rollinson, 1993). In contrast, the tuffaceous siltstones show Eu enrichment, probably due to plagioclase enrichment (Rollinson, 1993). The Eu enrichment could have resulted from diagenetic mobilization (MacRae et al. 1992) or from hydrothermal alteration (e.g. Michard et al. 1983), given that unaltered felspars are rarely visible in thin section. Compared with Post-Archean Australian Shale (PAAS) (Fig. 7b), the tuff is relative depleted in most elements but enriched in U, Th and heavy REE. The tuffaceous siltstones are very strongly enriched in Ba (up to 78 898) and also enriched in Th, U and Sr. The relatively high values of Ba, Th, U, Sr and Ti in the Panagra Formation tuffs are similar to the compositions of certain rhyolitic volcanics, notably the Miocene Kırka–Phrigian tuff of western Anatolia (Fig. 7c) (Seghedi & Helvacı, 2016), which is a possible source (see Section 5.c below).

Fig. 7. (a) Chondrite-normalized REE diagram for whole-rock samples. Normalizing values of chondrite from Nakamura (1974). (b) PAAS-mormalized multi-element diagram for whole-rock samples; normalizing values of PAAS from McLennan et al. (1993). (c) Panagra Formation tuffs versus Kırka–Phrigian tuff, west Anatolia (Seghedi & Helvacı, 2016).

On the La/Th versus Hf diagram (Fig. 8a), which is indicative of magmatic composition, the tuffaceous siltstones have relatively low La/Th ratios and intermediate Hf values, compatible with a felsic arc source. In contrast, the non-tuffaceous sandstones have higher La/Th ratios, consistent with an originally andesitic arc source of possible Late Cretaceous – Miocene age, based on dating of the zircons from the Miocene sandstones of the Kyrenia Range (Chen et al. 2019; Shaanan et al. 2020). On the Th/Sc versus Zr/Sc plot, both the relatively pure tuffs and the tuffaceous siltstones are of near-granitic composition (Fig. 8b), whereas the non-tuffaceous sediments plot between basalt and continental crust (source mixing is likely). These features are again consistent with the petrographic and whole-rock geochemical evidence from the Oligocene–Miocene sandstones from northern Cyprus (McCay & Robertson, 2012; Chen & Robertson, 2020). On the Al–Zr–Ti plot (Fig. 8c) the relatively pure tuffs and the tuffaceous siltstones are grouped, whereas the non-tuffaceous sandstones have higher Zr/Al ratios compared with PAAS, which is interpreted as the result of sedimentary sorting prior to deposition. The tuffaceous siltstones have chemical index of alteration (CIA) (Nesbitt & Young, 1982) values of 44–54, compared with CIA values of 75–76 for the non-tuffaceous sandstone turbidites (Fig. 8d). These CIA values are consistent with mild weathering conditions within the source area or during sediment transport. The relatively high CIA values of the background non-tuffaceous turbidites are suggestive of relatively humid source area and/or sorting process (Garcia et al. 1991).

Fig. 8. Chemical plots of the tuffs, tuffaceous sediments and associated non-tuffaceous sandstones, compared with various possible source compositions. (a) La/Th versus Hf diagram (after Floyd & Leveridge et al. 1987); (b) Th/Sc versus Zr/Sc diagram (after McLennan et al. 1993). Black solid circles indicate average compositions of granite, andesite and basalt (after Condie, 1993). Grey square represents average compositions of UCC (Rudnick & Gao, 2003; Hu & Gao, 2008). Purple arrow indicates compositional variations and the green arrow sedimentary recycling effects (i.e. zircon addition). (c) Al–Zr–Ti ternary diagram (after Garcia et al. 1991). The post-Archean Australia Shale (PAAS) data are from McLennan et al. (1993). Blue arrow shows possible sorting effects (Garcia et al. 1991). (d) Ternary diagram (after Nesbitt & Young, 1984) of the molecular proportions of Al₂O₃–(CaO*+Na₂O)–K₂O. Black arrow indicates general weathering trend (McLennan et al. 1993), and the brown arrow diagenetic K-metasomatism (Fedo et al. 1995). Open/solid boxes represent compositions of various rock types (McLennan et al. 1993). Three early Miocene tuffaceous units in Turkey are shown for comparison (Türkmen et al. 2013; Seghedi & Helvacı, 2016; Esenli et al. 2019); these are the Kırka–Phrigian caldera (Eskişehir area), the tuffs from Gördes Basin and the Salbaş Tuffaceous Member (Adana Basin).

5. Discussion

5.a. Sedimentological process

The combined field and laboratory evidence suggest that the relatively pure tuffaceous interval at Çınarlı (Platani) has three components. First, the homogeneous, finely laminated, fine-grained tuff resulted from direct fallout of felsic ash through the water column. Secondly, the sharp-based, normal-graded interbeds are interpreted as tuffaceous turbidites that contain a mixture of contemporaneous and reworked tuffaceous material (and some much older grains). Thirdly, where sharp bed tops as well as bed bases are present, this is suggestive of seafloor reworking by bottom currents (although this appears to be minor). In addition, the tuffaceous sandstones at Tirmen (Trypimeni) contain a variable mixture tuffaceous and terrigenous grains and are interpreted as tuffaceous turbidites. The absence of coarser-grained pyroclastic deposits in both occurrences could be explained either by long-distance aeolian transport or possibly by ponding of coarser-grained material closer to the source volcanism, for example, within local silled depocentres.

Assuming the Burdigalian microfossil and Sr-isotope ages for the Panagra Formation are correct (c. 16.6 Ma), the deposition of the relatively pure ash was broadly synchronous with eruption. However, the older early Miocene ages (17–21 Ma) are indicative of previous eruptions in the source area. The tuffaceous siltstones–sandstones, c. 11 km further west (Fig. 1b), represent compositionally similar tephra that mixed with terrigenous detritus and was finally deposited by turbidity currents.

Alkali feldspar and biotite are generally abundant within the Oligocene–Miocene terrigenous turbidites, as indicated by X-ray diffraction studies (McCay & Robertson, 2012). However, within the tuffaceous sediments these minerals appear to have been largely altered to kaolinite and carbonate minerals, as suggested by the petrographic studies.

5.b. Magma affinities inferred from zircon trace elements

The relatively high Lu (> 50 ppm), U (> 70 ppm), Ta (> 1 ppm) and Hf contents (> 11 000 ppm) of the most of the zircon grains analysed are consistent with felsic magma sources (Belousova et al. 2002). The Eu/Eu* ratios of all of the grains are consistent with plagioclase crystallization prior to, or coeval with, zircon growth (Hoskin & Schaltegger, 2003). Compared with zircons from known tectonic settings, the majority of the zircon grains fall within the anorogenic (within-plate) field on the basis of their Nb/Hf, Hf/Th and Th/Nb ratios (Fig. 9). The two Precambrian zircons (sample no. 21, 25) fall within the orogenic field, suggesting that their source rocks are arc-related.

Fig. 9. (a) Nb/Hf versus Th/U (after Hawkesworth & Kemp, 2006); and (b) Hf/Th versus Th/Nb diagrams (after Yang et al. 2012) for the zircons analysed. Contaminated zircon compositions are removed.

5.c. Possible magmatic sources

Pinpointing of magmatic sources for tuffaceous sediments can be difficult, especially for areas such as Anatolia that have experienced subsequent contractional tectonics (mainly late Miocene) and major uplift and erosion (mainly Pleistocene). However, the source, magmatic type and eruptive characteristics of the early Miocene (Burdigalian) Kyrenia Range tuffs and tuffaceous siltstones can be assessed based on chemical composition, comparative age and sediment fabric (i.e. tephra type, shape, grain size, sorting and bed thickness).

There is little evidence of a suitable local source in or around Cyprus. However, the Kyrenia Range is a thrust belt that experienced a final phase of southward thrusting during late Miocene time, concealing part of its foreland (Baroz, 1979; McCay & Robertson, 2013; Robertson & Kinnaird, 2016).

The nearest extensive Miocene felsic tuffs are exposed throughout the Adana basin to the north (Salbaş Tuff Member of the Kuzgun Formation) (Fig. 10). However, this is palaeontologically dated as Tortonian in age (c. 11 Ma) (Yetiş, 1988), significantly younger than the Kyrenia Range tuffs, and has a contrasting calc-alkaline (arc-related) composition (see online Supplementary Fig. S1). Early Miocene volcanics occur extensively in and around the Amanos Mountains bordering the Adana basin to the east (Fig. 10); these are represented by relatively quiescent-type basaltic eruptions and minor intrusions (mainly dykes) that remain poorly dated and chemically studied. However, felsic tuffs are, at most, minimal (Duman et al. 2017; unpublished data).

Fig. 10. Topography of Anatolia (Shuttle Radar Topography Mission; Farr et al. 2007) showing the three main volcanic provinces (of different age ranges) and the main locations of early–middle Miocene (20–15 Ma) post-collisional volcanism. Data from Türkmen et al. (2013), Prelević et al. (2012), Seghedi & Helvacı (2016) and Schleiffarth et al. (2018). NAFZ – Northern Anatolia Fault Zone; CAFZ – Central Anatolia Fault Zone; EAFZ – Eastern Anatolia Fault Zone; IAESZ – İzmir–Ankara–Erzincan Suture Zone; ITS – Inner Tauride Suture; EAVP – Eastern Anatolia Volcanic Province; CAVP – Central Anatolia Volcanic Province; WAVP – Western Anatolia Volcanic Province.

The Oligocene–Miocene terrigenous turbidites of the Kyrenia Range were derived from the east based on palaeocurrent data (Weiler, 1970; McCay & Robertson, 2012). Suitable source rocks are exposed in the Cenozoic SE Anatolian thrust belt (McCay & Robertson, 2012; Chen & Robertson, 2020; Shaanan et al. 2020). However, Miocene volcanics in SE Turkey are rare and mainly restricted to basalt-andesite, for example, the Bahçelievler area of the Kahramanmaraş region (Fig. 10) (Arger et al. 2000). There are no palaeocurrent data specifically from the rare Miocene tuffaceous sediments in the Kyrenia Range. There is no requirement for them to have been derived from the east, together with the terrigenous turbidites. However, some mixing of terrigenous and volcaniclastic material took place prior to final deposition.

The relative thinness of the primary bedding (c. 5 cm) and the fine grain size of the inferred fallout tuffs in the Kyrenia Range, combined with their chemical composition, prompt comparison with similar-aged felsic igneous rocks related to post-collisional magmatism in Anatolia (see Section 2 on geological background). The Kyrenia Range tuffs are broadly similar in composition and age to the felsic ignimbrites of the Eskişehir–Afyon–Isparta volcanic zone in the Western Anatolian Volcanic Province (Fig. 10) (Dilek & Altunkaynak, 2010; Seghedi & Helvacı, 2016). Specifically, large rhyolitic ignimbrite, pumice and dacitic lava-tuff units are well-exposed in the Kırka–Kütahya–Usak area of the Eskişehir–Afyon–Isparta volcanic zone (Bingöl, 1977; Yağmurlu et al. 1997; Aydar, 1998; Dilek & Altunkaynak, 2010; Seghedi & Helvacı, 2016). These are likely to represent Minoan (Plinian)-type eruptions (Fisher & Schmincke, 1984). They are chemically similar to the Kyrenia Range tuffs and tuffaceous sediments, although the latter have slightly higher values of U, Th, Ba, Sr and Ti (Fig. 7c), which could be explained by greater magmatic fractionation of the source magma. Small tuffaceous deposits of similar composition to those of the Kyrenia Range also occur further south, in the Isparta area (Fig. 10), but these are much younger (i.e. 4.1–4.6 Ma; Yağmurlu et al. 1997).

Radiometric dating indicates ages of 16–21 Ma for the Eskişehir–Afyon-Isparta volcanic zone as a whole (e.g. Bingöl, 1977; Aydar, 1998), similar to the age of the Kyrenia Range tuffs (16.64 ± 0.12 Ma). High-volume eruptions from the Kırka-Phrigian Caldera (Dilek & Altunkaynak, 2010; Seghedi & Helvacı, 2016) could have yielded ejecta of similar chemical composition (Fig. 7a–d) to the Kyrenia Range tuffs. Similar-aged tuffs (16–20 Ma) also occur within the Gördes and Selendi basins (Fig. 8) in western Anatolia (Purvis et al. 2005); however, these are of contrasting calc-alkaline composition (see online Supplementary Fig. S1).

5.d. Tephra distribution and sediment accumulation

Studies of tuff thickness versus distance from source (Fisher & Schmincke, 1984) suggest that the 5–10 cm thickness of the Kyrenia Range tuff fallout events could be equivalent to c. 10–100 m thickness in the source area (Fig. 11a). This is broadly consistent with the reported up to 50 m thickness of individual pyroclastic depositional events for the Kırka felsic tuffs (Seghedi & Helvacı, 2016). In addition, for tephra, the median grain size versus the known distance from source is broadly indicative of the distance of aeolian transport (Fisher & Schmincke, 1984). The Kyrenia Range tuffs are dominated by ash sizes of c. 63–125 μm (Fig. 4b1–b4) and plot close to the field of far-travelled, powerful eruptions (Fig. 11b), consistent with Minoan (Plinian)-type eruptions (Fisher & Schmincke, 1984). The position on the diagram is also suggestive of low-fall velocity of fragments carried by a high-velocity wind.

Fig. 11. (a) Thickness and distance from source along dispersal axis for several fallout tephra layers. Tephra distribution curves are modified from Fisher & Schmincke (1984). Minoan Plinian deposit curve is based upon compacted thickness of total Minoan tephra layer using adjusted isopach contours (Watkins et al. 1978). (b) Fields for median grain size versus log distance from source (km) for undifferentiated tephra samples derived from a large number of eruptions with known strength and transport distances (Fisher & Schmincke, 1984). The data are based on knowledge of a large number of modern and some ancient (i.e. Pleistocene) volcanoes from all over the world.

E-wards or SE-wards transport of tephra is suggested by studies of early Miocene fine-grained tephra distribution in the western Mediterranean region (i.e. NE Apennines) (Montanari et al. 1994). Climatic models suggest that northeasterly winds prevailed during middle Miocene time (Serravallian), changing to westerlies during late Miocene time (Tortonian) (Cornell et al. 1983; Montanari et al. 1994; Quan et al. 2014). Tuff, originating in the Kırka–Phrigian area of NW Anatolia could therefore have been dispersed E-wards and SE-wards by prevailing winds.

It is therefore possible that the Kyrenia Range tuffs represent highly explosive Minoan (Plinian)-type eruptions, during and soon after the initiation of post-collisional magmatism in western Anatolia. The felsic ash would have been dispersed c. 500 km in a southeasterly direction to reach the north Cyprus area (Fig. 10), where it fell out over the sea, sank to the seafloor and was partly reworked by gravity flows and currents. However, problematic aspects remain. There are no reported occurrences of early Miocene tuffs elsewhere within the Kyrenia Range, or within onshore sedimentary basins to the west and NW (e.g. Manavgat, Köprü, Aksu and Kaş basins); i.e. towards the suggested source area. Felsic tuffs were possibly deposited, but then reworked and diluted by terrigenous material and therefore not easily recognizable in the field.

After the early Miocene period, the prevailing winds were seemingly no longer favourable for tephra transport to the Cyprus area (Quan et al. 2014). This could explain why tuffaceous deposits are not known to occur higher in the Miocene–Pliocene stratigraphical succession in northern Cyprus, despite the continuing explosive volcanism within Anatolia.

6. Conclusions

The only known Cenozoic–Recent tuffaceous deposits in Cyprus are represented by a relatively pure felsic ash-rich interval at one locality, and by related tuffaceous siltstones–sandstones at another nearby locality, both in the eastern Kyrenia Range.

The tuffaceous deposits accumulated by a combination of direct ash fallout and by reworking, mainly by gravitational processes (i.e. tuffaceous turbidites).

U–Pb dating of detrital zircons from relatively homogeneous ash indicates a dominant eruption age of 16.64 ± 0.12 Ma (early Miocene), together with slightly older ages of 17–21 Ma that represent preceding volcanic events.

Zircon trace-element analysis suggests that the majority of the grains are felsic magma-sourced, of within-plate affinity, whereas a few grains are arc-related.

Whole-rock chemical analysis of the tuffaceous sediments is indicative of a felsic arc source characterized by low La/Hf, intermediate Hf and high Zr/Sc ratios.

The north Cyprus tuffs are generally similar in age and composition to explosive early Miocene, post-collisional felsic volcanics in western Anatolia, specifically the Kırka–Phrigian volcanic area c. 500 km to the NW of Cyprus.

During early Miocene time, tephra could have been carried SE-wards by prevailing winds to reach the north Cyprus area, followed by variable reworking.