Abstract
Tephra layers are a unique tool for stratigraphy. Their geologically instantaneous deposition across wide areas makes them a powerful instrument for dating and correlating distant localities. In this paper, we apply tephra fingerprinting for high-resolution stratigraphic purposes in the upper Miocene portion of the Pisco Formation, which crops out along the southern Peruvian coast. The Pisco Formation diatomaceous strata host an important marine vertebrate Fossil-Lagerstätte, whose palaeontological relevance has entailed the necessity of reconstructing a detailed chronostratigraphic framework using 39Ar–40Ar dating and tephra correlations. Distal ashes from the Central Andes volcanoes are frequent in the Pisco Formation, but their similar glass composition and mineral assemblage, together with the shallow marine environment limiting tephra preservation, could be unpromising for tephra-based correlations. In this study, tephra layers from measured stratigraphic sections were fingerprinted with a combined petrographic and microanalytical approach, including glass shard morphology and granulometric analyses. Based on the obtained results, we correlated sections several kilometres apart, greatly increasing the chronostratigraphic detail. Major element composition of biotite proved a valuable tool to discriminate the Central Andes metaluminous to peraluminous tephra deposited in the Peruvian forearc basins. This study highlights the applicability of tephra fingerprinting in unfavourable shallow marine environments.
Supplementary material: A compilation of tephra data (field observations, granulometric parameters, petrographic data and EPMA analyses on glass), EPMA single analyses on glass and biotite, 39Ar–40Ar step heating results, measured stratigraphic logs and biotite chemical diagrams are available at https://doi.org/10.6084/m9.figshare.c.4201484
Because of their regional dispersion, geologically instantaneous deposition and the common presence of minerals suitable for radiometric age determination, the value of volcanic ashes as regional chronostratigraphic markers has long been recognized (Lowe et al. 2017, and references therein). Detailed chronostratigraphic reconstructions are particularly important for the sedimentary fill of the East Pisco Basin exposed along the southern Peruvian coast (Fig. 1a). The East Pisco Basin hosts one of the most important cetacean-bearing Fossil-Lagerstätten worldwide, with exceptionally well-preserved fossils of marine vertebrates (Bianucci et al. 2016a, b). In particular, along the western side of the Ica River Valley, strata of the Mio-Pliocene Pisco Formation are characterized by fossil remains of cetaceans, both odontocetes and mysticetes, but also of pinnipeds, seabirds, sea-turtles, crocodiles, sharks, rays and bony fish (Lambert et al. 2009, 2017a, b; Collareta et al. 2015; Bianucci et al. 2016a, b, c, and references therein; Di Celma et al. 2017; Landini et al. 2017a, b; Marx et al. 2017). Correspondence between stratigraphy and vertebrate findings has been a long-term neglected issue and fossil-bearing localities have not been correlated with the adjacent ones. To better understand, both qualitatively and quantitatively, the evolution of the rich marine vertebrate fauna during a pivotal time frame for the evolutionary history of marine vertebrates, such as the late Miocene, the reconstruction of an accurate, high-resolution absolute and relative age framework for the Pisco Formation is needed. This age interval is characterized by a significant cetacean turnover outlined by the diversification of several lineages of crown neocetes (e.g. delphinidans, beaked whales and rorquals; Fordyce & Muizon 2001; Bianucci & Landini 2007; Marx et al. 2016). Accurate dating of the origin of some key taxa can also allow a better definition of several calibration points for molecular divergence among crown neocetes (e.g. Lambert et al. 2017b).
(a) Map of the major sedimentary basins along the Peruvian coast. Sedimentary basins are indicated with blue ellipses. Major structural highs, the Coastal Batholith, the Outer Shelf High and the Upper Slope Ridge, are represented in red, orange and yellow, respectively. The green square highlighting the East Pisco Basin indicates the study area, shown in (c). Modified from Di Celma et al. (2016a). (b) Volcanism and magmatic arcs during the Cenozoic Era in southern Peru. Lines in different colours highlight the extension of Cenozoic magmatic arcs. Modified from Mamani et al. (2010). (c) Ica Desert localities mentioned in the paper. The study area includes eight localities in the western side of the Ica River Valley, to the west and south of Ocucaje, in the Ica Region. Modified from Lambert et al. (2009). CC, Cerro Colorado; CB, Cerro Blanco; CTC, Cerro Toro Chico; CHZ, Cerro Hueco la Zorra; CLQ, Cerro los Quesos; CLQw, Cerro los Quesos west; CLB, Cerro la Bruja; CLZ, Cerros Cadenas de los Zanjones. The green square highlights the study area of Cerro los Quesos and Cerro la Bruja, shown in (d). (d) Satellite image of the localities of Cerro los Quesos and Cerro la Bruja in the western side of the Ica River Valley. Based on image © 2018 DigitalGlobe.
The Pisco Formation is a diatomite-rich marine sedimentary unit, which features numerous layers of volcanic ash, representing distal air-fall events from the Peruvian Central Andes volcanoes, that are a very important tool for defining its stratigraphic framework (Di Celma et al. 2017; Gariboldi et al. 2017). So far, their use has been limited to field correlations within a single locality and to obtain 39Ar–40Ar ages (Di Celma et al. 2016a,b). The high number of macroscopically similar ash layers interbedded within the monotonous diatom-rich succession hampers the unequivocal identification of these ashes directly in the field. With the aim of fully exploiting the potential of these volcanic ash layers for tephrostratigraphy, in this paper we fingerprint tephra layers by a chemical and petrographic characterization and correlate them between different localities.
Tephra fingerprinting is widely recognized as a useful method for stratigraphic reconstructions. This method is largely employed in the Quaternary (e.g. Paterne et al. 1988; Shane et al. 2003; Lowe 2011; Smith et al. 2011; Zanchetta et al. 2011; D'Antonio et al. 2016) and has been used for correlating deposits of large eruptions of the Central Andes (de Silva & Francis 1989; Lebti et al. 2006; Breitkreuz et al. 2013). We focus on the P2 depositional sequence, which is the youngest unconformity-bounded unit documented so far in the upper Miocene portion of the Pisco Formation with a high palaeontological content (Di Celma et al. 2017, in press). We fingerprint tephra layers of the P2 sequence by using major element composition of biotite and glass combined with petrographic and granulometric features. Despite the expected predominance of tephra with similar silicic composition, and the discontinuity of the record owing to the shallow marine environment, we correlate ash layers in different localities, greatly improving stratigraphic resolution, and highlight the feasibility of an approach based on major element biotite and glass composition.
Setting
Geological and volcanological setting
According to Thornburg & Kulm (1981), the Peruvian margin is part of the long-lived convergent margin of South America and shows two extended structural ridges, the Outer Shelf High and the Upper Slope Ridge, dissecting the Peruvian coast into a series of Cenozoic forearc basins (Fig. 1a). The East Pisco Basin, the onshore part of one of these basins at 14°30'S latitude, is confined by the igneous rocks of the Mesozoic Coastal Batholith to the east (Cobbing 1999) and by Precambrian to Jurassic rocks of the Coastal Cordillera to the west (Romero et al. 2013). As described by previous researchers (Dunbar et al. 1990; DeVries 1998), this basin was the site of discontinuous sediment deposition between the Eocene and the Pliocene. Starting from the late Pliocene, the subduction of the aseismic Nazca Ridge beneath the South American plate has caused the uplift of the East Pisco Basin (Macharé & Ortlieb 1992; Hampel 2002), resulting in the onland exposure of its Eocene to Pliocene sedimentary fill.
The numerous volcanic ashes interbedded with the basin-filling sedimentary succession provide a snapshot of the long-lived activity of the Central Andes volcanoes, frequently punctuated by large explosive eruptions. The dispersion of Andean tephra, as demonstrated by historical eruptions (Adams et al. 2001) as well as by offshore drilling programme logs (Hart & Miller 2006), largely affected the margin, with offshore deposition of Plinian and of co-ignimbrite ashes. As reviewed by Mamani et al. (2010), the Central Andean margin has a very complex subduction-related history. The late Paleozoic to middle Cretaceous history was first characterized by tectonic stretching, allowing the formation of a marine back-arc basin. From the late Cretaceous to the middle Oligocene, a continuous relief formed, with a continental back-arc environment. The major crustal thickening typical of the Andean orogeny was developed only since the middle Oligocene, as described by Mamani et al. (2010). Those researchers subdivided the continuous magmatic activity into four discrete phases: the Tacaza arc, from 30 to 24 Ma, the Huaylillas arc, between 24 and 10 Ma, the Lower Barroso arc, from 10 to 3 Ma, and the Upper Barroso arc, between 3 and 1 Ma (Fig. 1b). Voluminous Neogene and Quaternary ignimbrite deposits occur in southern Peru, testifying to the main periods of intense and voluminous explosive activity of the arcs (Lebti et al. 2006; Thouret et al. 2007, 2016). According to Thouret et al. (2016), repeated pulses of ignimbrites occurred every 2–3·8 Ma between 24·5 and 9·0 Ma in southern Peru, whereas a quasi-continuous volcanism produced four smaller ignimbrite sheets and related pyroclastic density current deposits from 5 Ma onward. The present-day active volcanic arc, or Western Cordillera, lies about 230 km east of the trench (Fig. 1b). In southern Peru, the active volcanic arc ends at about 14°S, corresponding to the subduction of the buoyant Nazca Ridge beneath Peru and to the change from ‘normal’ to ‘flat’ subduction (Hampel 2002; Bishop et al. 2017).
Stratigraphic and geochronological setting
The lithostratigraphic units recognized by Dunbar et al. (1990) and DeVries (1998) in the East Pisco Basin are, from the oldest to the youngest, the Eocene Paracas Group, the upper Eocene to lower Oligocene Otuma Formation, the upper Oligocene to lower Miocene Chilcatay Formation and the upper Miocene to Pliocene Pisco Formation. These formations are separated by regional unconformities marked by pebble- and boulder-grade conglomerates, which represent long periods of subaerial exposure.
Adams (1908) provided a first description of the fossil-bearing deposits around Ocucaje, among which he pointed out the presence of volcanic materials. Muizon & Bellon (1980, 1986) and Dunbar et al. (1990) noticed the presence of ‘tuffaceous siltstones’ in the Pisco Formation without describing the petrology and the composition of tephra layers. Some researchers have provided ages of the Pisco Formation by applying K–Ar and 39Ar–40Ar radiometric dating on these tephra beds (Muizon & Bellon 1980, 1986; Brand et al. 2011; Esperante et al. 2015), integrating the biostratigraphic framework obtained by diatoms, radiolarians and planktonic foraminifera (Macharé & Fourtanier 1987; Schrader & Ronning 1988; Tsuchi et al. 1988; Dunbar et al. 1990). These radiometric ages, however, are not placed in a stratigraphic context and, therefore, provide age indications for only individual localities without correlations among them.
Over the last few years, new studies focusing on the stratigraphy of the Pisco Formation have been carried out (Di Celma et al. 2016a, b, 2017). In particular, to place its fossil vertebrates in a proper chronostratigraphic framework, new stratigraphically constrained sediment samples have been collected and analysed for diatom biostratigraphy and new 39Ar–40Ar ages have been provided (Gariboldi et al. 2017). The fossiliferous strata of the Pisco Formation exposed along the western side of the Ica River Valley have been divided into three fining-upward depositional sequences named, from the oldest to the youngest, P0, P1 and P2 and bounded by three regionally extensive unconformities, PE0.0, PE0.1, and PE0.2 of possible eustatic origin (Di Celma et al. 2017, in press). Combined diatom biostratigraphy and tephrochronological 39Ar–40Ar results show that in the study area the interval of the P1 sequence characterized by the co-occurrence of the diatoms Denticulopsis praekatayamae and Lithodesmium reynholdsii was deposited between 9·5 and 8·9 Ma, whereas the P2 sequence is younger than 8·5 Ma (Di Celma et al. 2017; Gariboldi et al. 2017). With this approach, a detailed chronostratigraphic framework is being reconstructed, improving previous scattered radiometric data. The stratigraphic framework of the Pisco Formation has been much improved, and a nearly continuous record is now available. For this study, we decided to focus on two fossil-rich localities for demonstrating the feasibility of applying tephrostratigraphy on ash layers in the Pisco Formation. It is important to consider that sediments of the P2 sequence have been dated at the localities of Cerro los Quesos (14°30'59·0"S, 75°42'59·1"W) and Cerro la Bruja (14°31'44·10"S, 75°40'0·10"W), two fossiliferous areas located about 6 km apart (Fig. 1c and d), where most of the tephra layers studied in this paper were collected and high-resolution stratigraphic sections have been measured (Di Celma et al. 2016a, 2017). From the oldest to the youngest, the 39Ar–40Ar ages obtained on biotite phenocrysts from samples collected at Cerro los Quesos are 7·55 ± 0·05 Ma (CLQ-T49 tephra layer, sampled 31·5 m above the base of the Cerro los Quesos measured section; abbreviated: abs), 6·93 ± 0·09 Ma (CLQ-T1a tephra layer, sampled at 142 m abs) and ≥6·71 ± 0·02 Ma (CLQ-T9b tephra layer, sampled at 193·5 m abs) (Gariboldi et al. 2017). To confirm the field-based correlation between the two stratigraphic sections measured at these localities, an additional 39Ar–40Ar age of 7·45 ± 0·01 Ma was obtained from biotite phenocrysts of the tephra layer LB-T11 at Cerro la Bruja (sampled at 174 m abs) as introduced by Di Celma et al. (2017) and described in the present Supplementary material (Fig. S1). Thanks to these radiometric ages and the previous stratigraphic reconstructions, in this study we were able to apply tephra fingerprinting and tephrostratigraphy to make correlations and increase the chronostratigraphic resolution of the exposed succession. This study allows us to increase the detail in chronostratigraphy compared with using the radiometric data alone.
Methods
An area of more than 240 km2 in the western side of the Ica River Valley was explored during several field campaigns between 2014 and 2016, to reconstruct the stratigraphy of this area (Di Celma et al. 2016a, b, 2017). In the course of this fieldwork, more than 200 samples of volcanic ash layers were collected at eight localities (see Fig. 1c): Cerro los Quesos (14°30'59"S, 75°42'59"W), Cerro los Quesos west (14°31'13"S, 75°44'34"W), Cerro la Bruja (14°31'44"S, 75°40'00"W), Cerro Colorado (14°21'04"S, 75°53'43"W), Cerro Hueco la Zorra (14°26'58"S, 75°41'01"W), Cerro Blanco (14°24'57"S, 75°41'17"W), Cerro Toro Chico (14°27'36"S, 75°48'01"W) and Cerros Cadenas de los Zanjones (14°34'21"S, 75°42'60"W). Stratigraphic sections were measured in each locality (see Supplementary material, Fig. S2), with the exception of Cerro Hueco la Zorra, where an ash layer was stratigraphically recognized and sampled for correlation analyses with an ash layer in Cerro Blanco, 3 km away. The collected tephra were named with initials related to the locality, a ‘T’ for tephra, and a number concerning the stratigraphic position along the measured section (e.g. CLQ-T1); lower-case letters are used for sublayers recognizable in the same tephra (e.g. CLQ-T1a).
In the field, volcanic ash layers were described through observations on their lateral continuity, thickness, vertical grading, colour, sedimentary structures and presence of crystals. Samples were collected close to the base of the ash layer, avoiding the upper portion so as to exclude bioturbation and reworking. To avoid the weathered particles, the superficial parts of the tephra layers were eliminated before sampling. Among the collected tephra, 45 samples were selected for petrographic and chemical investigations, focusing on the fossil-rich localities of Cerro los Quesos and Cerro la Bruja because of their palaeontological relevance. They were examined under a stereomicroscope and prepared as smear slides for petrographic analysis and for a semi-quantitative component analysis, to check the presence of non-volcanic particles and evidence of alteration of volcanic components. Based on the results of this first screening, samples with more than 5 vol% of non-primary particles were interpreted as reworked or remobilized, and therefore were discarded (Griggs et al. 2014; Tada et al. 2015).
Grain-size analyses were performed with the Malvern Mastersizer 2000E™ Laser Granulometer at the Università di Milano-Bicocca. For each sample, three subsamples were made, and each of them was measured three times. Grain-size data were processed with the Grain Size Analysis Program GRADISTAT (Blott & Pye 2001). Samples were also wet sieved with meshes of 500, 250, 125 and 63 µm. The ashes from the 250–500 µm and 125–250 µm fractions were mounted in resin, polished with silicon carbide and alumina, and carbon-coated.
Glass shard morphology and mineral assemblages of the samples representing the product of primary deposition of ashes were observed by optical microscopy on smear slides and analysed by scanning electron microscopy (SEM) back-scattered electron (BSE) imaging on the carbon-coated polished mounts, to fingerprint the petrographic and textural features of the juvenile volcanic components. For the chemical composition, 32 samples were selected for biotite analyses and 35 samples were analysed for the glass composition. Electron probe micro-analysis (EPMA) and energy-dispersive X-ray spectroscopy (EDS) analyses were performed on at least 10–15 representative glass shards and 10 biotite phenocrysts (core and rim) per sample, to obtain a chemical characterization of the tephra layers and an estimate of the alteration degree of biotite phenocrysts. SEM-EDS analyses were carried out at the Università di Pisa with a Philips™ XL30 scanning electron microscope equipped with an Edax Genesis microanalytical device, with 20 kV filament voltage, 5 nA beam current and ZAF correction, in raster mode and 10 × 10 or 5 × 5 µm windows. EPMA was performed with a JEOL JXA-8600™ at Consiglio Nazionale delle Ricerche in Firenze and with a JEOL 8200 Superprobe™ at the Università di Milano-Bicocca. Analytical conditions were 15 kV accelerating voltage, 5 nA beam current, 3 µm beam focus for biotite and 10 µm for glass shards.
Results
Field data
Along the stratigraphic sections, ash layers were more frequent in the P2 sequence than in the other sequences (about 10–30 tephra layers thicker than 2–3 cm every 100 m) and particularly abundant in the diatomitic portion. For the majority of the sampled tephra, thickness ranges between 5 and 15 cm. However, whereas their bases are commonly sharp and easy to recognize (Fig. 2a), their upper boundaries may be irregular or diffuse and not always clearly detectable, particularly in diatomite-rich sediments, owing to the similar light colour and grain size. Some thicker tephra (up to 50 cm), organized in different sub-layers, have been found. Except for very fine-grained and crystal-free ashes, the ash beds show an evident normal grading and the lower portion is typically characterized by a higher concentration of phenocrysts (Fig. 2b). Although in most cases they are unconsolidated, in some cases volcanic ash layers are cemented by secondary minerals such as gypsum and jarosite. At the base, a hard gypsum or anhydrite crust of a few millimetres thickness is commonly found. The colour in the field is usually grey to light grey, with a ‘pepper and salt’ appearance owing to biotite, when present (Fig. 2b). Thin (2–3 cm) dark grey tephra were also found, although less frequently.
Field photographs. (a) Tephra layer BL-T1 interbedded with Pisco Fm diatomites at Cerro Blanco. Tephra are often revealed by the sharp bottom discontinuity highlighted by gypsum wavy layers protruding from the outcrop. Hammer is 33 cm long. The arrow shows the tephra layer. (b) Detail of the same tephra layer before cleaning the outcrop: ‘pepper and salt’ appearance is due to the biotite concentration at the base. (c) Load casts in the ash layer CLQ-T12 (Cerro los Quesos). Marks on the stick are at 10 cm intervals. (d) Diapirs and flames recognizable in the tephra layer CLQ-T1 (Cerro los Quesos). Lens cap diameter is 6·5 cm.
Internally, a variety of soft-sediment deformation structures (including load casts, small-scale folding, convolutions, and dish and pillars structures) have been observed in some of the thicker layers (CLQ-T1 and CLQT-12). These deformation structures could have developed by rapid deposition of the thick ash layers on waterlogged diatom ooze, which may have started consolidation in the immediately underlying sediments (Pedersen & Surlyk 1977). Waters escaping from such layers may liquefy or fluidize the lower portion of the ash bed, resulting in sinking of the overlying denser portion into the liquidized lower portion (load casts; see Fig. 2c) and compensatory upward intrusion (diapirs and flames; see Fig. 2d) of the latter into the overlying sediment (Owen 2003). In one case, these deformation structures are recognizable in two exposures of the same tephra layer 6 km apart (i.e. CLQ-T12 and LB-T7).
Petrographic fingerprinting of tephra
Grain-size distribution and components of volcanic ash
The grain sizes of the sampled ash layers fall in the range of fine-grained sand to silt. Inspections under SEM of bulk samples of some particularly coarse tephra and of the coarser fractions of pilot wet sieved samples indicate, in some cases, a minor percentage of strongly agglomerated particles, cemented by secondary minerals. However, most of the samples consist of loose ash, nearly unaffected by secondary cements. The grain-size distribution is unimodal or bimodal and the classification is dominated by fine-grained ashes, very fine-grained ashes or extremely fine-grained ashes, mainly moderately to well sorted (White & Houghton 2006) (see Supplementary material, Table S1). In the case of tephra subdivided in sublayers (e.g. CLQ-T1, CC-T1), the lower sublayer is usually coarser than the upper one; only in one case (CLQ-T9), a normally graded coarser layer directly overlies a well-sorted fine-grained ash layer, indicating deposition of ashes of two distinct, but relatively close, eruptive events.
The component particles are dominated by glass shards (>85–95 vol%), without visible evidence of alteration, and variable amounts of juvenile phenocrysts (biotite, sanidine and plagioclase, in some cases amphibole, in one case muscovite) and traces of terrigenous and biogenic materials. The ash layers containing more than 5 vol% of particles extraneous to the primary volcanic material, such as diatom frustules, rock fragments, heterogeneous crystal assemblages with olivine, pyroxene, quartz and reddish aggregates probably consisting of Fe-hydroxides and clays, were interpreted as tephra modified by reworking and the samples were not further processed.
The 45 ash layers interpreted as primary tephra, sampled at eight sites focusing on Cerro los Quesos and Cerro la Bruja, were studied for textural characteristics.
Texture of ash particles
Glass shard morphology is a promising tool for distinguishing and correlating tephra layers (Lowe 2011). Optical and scanning electron microscopy allowed the study of the glass morphologies and characteristics (Fig. 3). Five main groups of glass shards were identified: bubble-wall shards, with round and concave sides; vesiculated shards, with variable vesicularity; stretched shards, being vesiculated shards with strongly elongate vesicles owing to stretching; platy shards, without any vesicles and not stretched; shards with microlites, mainly plagioclase. Tephra were fingerprinted based on the prevalent shard features in the 125–250 µm size fraction.
BSE images. (a) 250 µm size fraction of a tephra layer. (b) Biotite phenocryst showing an apatite inclusion and coated by volcanic glass. No clays along cleavages are observed. (c) Example of bubble-wall glass shard morphology. (d) Stretched and vesiculated glass shards, with strongly elongate vesicles owing to stretching. (e) Detail of a glass shard with microlites of plagioclase.
Volcanic glass composition
The chemical composition of volcanic glass was analysed on 35 samples (Table 1; Supplementary material, Table S2), avoiding the regions very close to vesicles and cracks, which could be more strongly affected by alteration, and microlites. However, BSE imaging of glass shards did not reveal, except in some cases, the formation of visible hydration cracks or of secondary minerals at the expense of the glass. As regards microlites, these are absent in most samples but are common in a few tephra having a dark grey colour and an andesitic composition. For each tephra, glass shards were analysed and the mean and standard deviation of microprobe analyses was calculated to check for homogeneity. Highly heterogeneous glass analyses can indicate compositionally zoned volcanic products, but, if joined to the presence of non-volcanic clasts, could reveal a suspected non-primary ash layer. In most analysed tephra, the standard deviation of the SiO2 concentration in glasses is <1%. A few samples have a larger variability of the SiO2 concentration (e.g. LB-T9), which can correspond to the eruption of a zoned reservoir, or to two glass populations (e.g. LB-T8). As in the latter tephra non-volcanic clasts are absent, the two glass populations can be interpreted as two eruptions close in time, or as magma mixing/mingling. Because some hydration affected at variable degree all the analysed ashes, resulting in microprobe totals of 89–95% (Table 1; Supplementary material, Table S2), the analyses were recalculated to 100 to allow comparison of tephra with different degrees of hydration in the same diagrams (Fig. 4a–d).
Harker diagrams. (a) Volcanic glass K2O v. SiO2 diagram showing the glass composition of P2 sequence tephra layers. Fields for arc-related volcanic rocks follow Peccerillo & Taylor (1976). (b) Volcanic glass Al2O3 v. SiO2 diagram. (c) Volcanic glass Na2O v. SiO2 diagram. (d) Volcanic glass CaO v. SiO2 diagram. Major elements are shown as weight per cent and recalculated to 100 to compare glasses with variable hydration. Data are shown in Table 1 and Supplementary material, Table S2.
Electron microprobe glass analyses of some P2 tephra samples, selected to show the andesite-to-rhyolite compositional variability of tephra in the Pisco Fm
The results are shown in a K2O v. SiO2 diagram in Figure 4a, with fields for arc-related volcanic rocks following Peccerillo & Taylor (1976). The silica content ranges from 57 to 78 wt% and K2O content ranges from 2 to 6%, with a general increase with silica. As regards the low-silica glass compositions, these belong to the few dark grey tephra (see Supplementary material, Table S1). These show scattered microlites in the glass shards (Fig. 3e). Glass compositions indicate an andesitic to dacitic affinity for these ashes when the major elements are recalculated to 100, but the silica content of residual glass can be higher than the magma composition owing to the effect of microlite crystallization and secondary hydration. Andesitic magmas with markedly different K2O content have been found, showing also different CaO content (e.g. CLQ-T44, ZANJ-T29, ZANJ-T31 in Fig. 4a and d). As regards the silicic tephra samples, the negative trends of K2O with silica in some samples (Fig. 4a) could correspond to fractionation of K-bearing minerals such as sanidine and biotite, which have been observed in most silicic samples. However, because the K2O decrease is not always accompanied by Al2O3 decrease (Fig. 4b), and because of the scatter of the Na2O content of glass (Fig. 4c), a variable alkali loss is likely. For the silicic compositions, recalculation to 100 results in a higher silica content compared with the original erupted magma, as high as 78 wt%. However, considering that the glass shards of a same tephra have a similar hydration degree, we conclude that the Pisco Formation tephra cover a large chemical variation, from andesite to rhyolite. This broad compositional range is confirmed by the petrographic observation that silicic glass shards are often accompanied by biotite and sanidine, whereas tephra with less silicic and intermediate glass shards lack biotite. Moreover, for the purpose of fingerprinting, it is worth noting that, even if the composition of glass of most ash layers plots in the rhyolite field, it is possible to distinguish different, non-overlapping groups of tephra, all along the andesite to rhyolite compositional range (Fig. 4a).
Biotite composition
BSE imaging of biotite phenocrysts did not show obvious secondary alteration minerals such as clays along cleavages (Fig. 3b). The composition was analysed on 32 samples, avoiding cleavages, accessory minerals (apatite and zircon are common) and glass inclusions.
In terms of mica classification (Rieder et al. 1999), most of the analyses plot in the biotite field, close to the phlogopite boundary, with a trend pointing to the siderophyllite end-member (Fig. 5a). Biotite phenocrysts are homogeneous or slightly zoned in terms of the core-to-rim major elements compositional range (Table 2). The intra-sample compositional variability is moderate in most cases (Fig. 5), and a few samples present some outliers (biotite phenocrysts with very different composition, which can be interpreted as foreign crystals entrained within the ash layer or xenocrysts in the erupted magma). A few samples show bimodal (e.g. CLQ-T9b) or very heterogeneous biotite composition (e.g. LB-T2). As a whole, the collected data reveal a large variation in Mgv (Mg/(Mg + FeT + Mn)), Al, Ti (Fig. 5a and b) and Mn (see Supplementary material, Fig. S4). The loss of alkali with respect to stoichiometry is in most cases very low. The high Al content of the siderophyllite-rich biotite phenoscrysts suggest the possibility of peraluminous compositions. When the chemical composition of biotite is plotted in the ternary discrimination diagram of Abdel-Rahman (1994), as shown in Figure 5c, two clusters are evident: one in the calcalkaline field and the other in the peraluminous one.
(a) Classification diagram of biotite phenocrysts from P2 tephra layers. (b) Biotite Mg/(Mg + FeT + Mn) v. Ti diagram of P2 ash layers. Values shown as atoms per formula unit. (c) Chemical ternary diagram of biotite phenocrysts from the correlated tephra. Biotite discrimination diagram; fields are from Abdel-Rahman (1994). C, calcalkaline orogenic suites; P, peraluminous suites; A, anorogenic alkaline suites. Data are shown in Table 2 and Supplementary material, Table S3.
Electron microprobe analyses of biotite phenocrysts (wt%) of some P2 tephra samples, selected to show the compositional variability of tephra in the Pisco Fm
Discussion
Tephra of the Pisco Formation
Tephra consisting of over 95% volcanic glass and a few per cent crystals are particularly frequent in the diatomaceous portions of the upper Miocene Pisco Formation. Probably, the diatomaceous mudstones represent a relatively deeper (60–100 m), low-energy environment in the basin, well suited for tephra preservation. These fine-grained ashes can be interpreted to represent the distal, crystal-depleted ashes of large explosive eruptions, dispersed westwards from the eruption column or from a co-ignimbrite ash-cloud (Fisher & Schmincke 1984). Following Engwell et al. (2014), the rather frequent bimodal grain-size distribution within the tephra may be interpreted to represent the coarse-grained and the fine-grained modes of the Plinian ash and the co-ignimbrite ash, respectively. Ash layers consisting of sublayers can be explained by subsequent and close in time eruptions of the same or different volcanoes (e.g. CLQ-T9a, b). Particularly thick layers of fine ash can be explained, instead, by a high input of ash deposited on land and then transported to the sea after a large explosive eruption (e.g. CLQ-T16).
The offshore deposition of tephra in forearc basins could, potentially, provide a more complete record of the explosive activity of the Central Andes compared with on-land outcrops. The upper Miocene Pisco Formation corresponds to a period characterized by intense volcanic activity in the Central Volcanic Zone (Tosdal et al. 1981; Thouret et al. 2007, 2017; Kay et al. 2010). In particular, the stratigraphic interval studied here (from 8·5 to 7 Ma), which temporally includes the activity of the Lower Barroso arc (Mamani et al. 2010), slightly precedes the 5–6 Ma peak in the presence of tephra in the marine record off the coast of Peru at 10 to 11°S, found during Ocean Drilling Program (ODP) Leg 201 by Hart & Miller (2006). Several large silicic eruptions have occurred in the Central Andes during the late Miocene–Pliocene, including some in the Altiplano–Puna region with a volcanic explosivity index (VEI; Newhall & Self 1982) of 8 or greater (Salisbury et al. 2011). Moreover, nearer the latitude of the East Pisco Basin, the Macusani region (SE Peru) produced several large ignimbrites between 10 and 7 Ma (Pichavant et al. 1988; Cheilletz et al. 1992). Macusani ignimbrites, as only few others in the Central Andes, are strongly peraluminous magmas, and thus represent a suitable source for the peraluminous tephra identified in the Pisco Formation.
Applying tephra fingerprinting
Biotite and volcanic glass have already been used for Andean tephrostratigraphy, as well as amphibole (de Silva & Francis 1989; Lebti et al. 2006). Biotite phenocrysts are rather frequent in the Central Andes eruptions. The biotite enrichment of their distal ashes is due to the fact that biotite flakes are transported in air along with glass shards much better than grains such as feldspars or amphiboles (e.g. Rose & Chesner 1987). As non-stoichiometric mica compositions reveal alteration, it follows that unaltered, stoichiometric micas can be confidently identified as juvenile in the primary ash-falls.
Both biotite and glass composition was used in the Pisco Formation for fingerprinting ashes and making correlations among them. The significant differences in the major element composition of biotite found in the Pisco Formation ashes, especially the large range in Mg/Fe ratio and Al content, cannot be ascribed to alteration (see Table 2). Alteration of biotite by weathering may result in Fe oxidation, loss of Mg resulting in low MgV, and loss of interlayer K, whereas Al and Ti are retained (Gilkes & Suddhiprakarn 1979). The physical and mineralogical evidence corresponding to these chemical transformations is lacking in the studied samples, thus indicating that biotite is pristine. Instead, the large chemical variations in MgV uncorrelated to the K content, as well as the differences in minor elements as Ti, Al, Ba and Mn, can be ascribed to the physico-chemical features (chemistry, temperature, oxygen fugacity) of the magma from which biotite crystallized and, therefore, can be used for discriminating biotites from different eruptions.
Based on the composition of biotite, different samples or groups of samples form well-defined and non-overlapping compositional clusters in one or more bivariate diagrams (e.g. Fig. 5). The possibility of applying this feature for correlating tephra is particularly evident in the diagram of Figure 6a, constructed for the fossil-rich localities of Cerro los Quesos and Cerro la Bruja, 6 km apart, showing several perfectly overlapping tephra pairs that are well distinguished from each other: CLQ-T23–LB-T19a, CLQ-T21–LB-T17, CLQ-T16–LB-T8, CLQ-T15–LB-T11 and CLQ-T13b–LB-T10. The composition of biotite for each pair is in most cases well clustered (e.g. CLQ-T15–LB-T11), whereas in some cases it covers a wider field. For instance, two biotite populations can be distinguished in both samples LB-T19a and CLQ-T23 (Fig. 6a). This feature, which is not correlated to alteration evidence (i.e. low totals, low K), could be interpreted as coeval eruption of two different volcanoes, two events close in time for the same volcano, or eruption of a mingled magma (Lebti et al. 2006; Shane et al. 2008), and is, thus, an additional characteristic useful for unequivocal identification of a tephra layer. A wide field of biotite composition is also covered by the tephra pair CLQ-T17–LB-T15, which shows a very dispersed mica composition inconsistent with a single magma batch. This can be interpreted as (primary or secondary) mixing of ashes from different eruptions. Figure 6a also demonstrates the chemical identity of samples BL-T1–LZ-T1 from Cerro Blanco and Cerro Hueco la Zorra, respectively. In this case, chemical analyses confirm the correlation between an ash layer placed in a stratigraphic log (BL-T1, see Supplementary material, Fig. S2) and another one at 3 km distance (LZ-T1), previously correlated to BL-T1 based on field evidence. Strong variations in Al content of biotite crystals embedded in similarly silicic glasses indicate remarkable chemical differences of the silicic magmas. Indeed, major element compositions identify peraluminous tephra compositions, such as the CLQ-T23–LB-T19a pair. The use of the major element composition of mica grains is more robust than that of glass owing to the high reactivity to alteration of the latter. In the case of glass, the removal of mobile cations results in an apparent increase of immobile elements (Al, Ti), which may simulate peraluminous compositions. Therefore, in many glass analyses the peraluminous character is only apparent. It is the biotite compositions that unequivocally identify the peraluminous tephra.
(a) Ti/Al v. Mgv (Mg/(Mg + FeT + Mn)) diagram of biotite phenocrysts from correlated tephra. Six correlations using biotite composition are possible at the localities of Cerro los Quesos and Cerro la Bruja, and one correlation between Cerro Blanco and Cerro Hueco la Zorra. Continuous-line ellipses highlight the correlations; dashed line surrounds the scattered points of two correlated tephra with two different biotite populations. (b) SiO2 v. K2O diagram of volcanic glass of correlated samples from Cerro los Quesos and Cerro la Bruja, and one pair from Cerro Blanco and Cerro Hueco la Zorra. Continuous-line ellipses highlight the correlations.
Glass composition was also employed for fingerprinting ash layers and making correlations between stratigraphic sections in different localities. Indeed, notwithstanding the susceptibility of glass to alteration, an overlap in the glass composition confirms the identification of the layers correlated based on biotite composition. As shown in Figure 6b, the same tephra pairs are overlapping and can be easily distinguished from each other: CLQ-T23–LB-T19a, CLQ-T21–LB-T17, CLQ-T17–LB-T15, CLQ-T16–LB-T8, CLQ-T15–LB-T11 and CLQ-T13b–LB-T10 for the localities of Cerro los Quesos and Cerro la Bruja, and the BL-T1–LZ-T1 pair at Cerro Blanco and Cerro Hueco la Zorra.
The compositional correlation of these layers is also supported by glass clast morphology and vesicularity, which give a further characterization useful for fingerprinting tephra: pairs of correlated tephra show similarities in the vesicularity, stretching, presence of microlites and clast glass shape (see Supplementary material, Table S1). In many cases, grain size validates the correlation, showing a correspondence between the grain-size distribution curves of the correlated pairs (see Fig. 7). Compared with the chemical composition and morphology, the employment of grain-size distribution for fingerprinting and correlating tephra in the East Pisco Basin may be limited by the effect of marine currents in redistributing the component particles during deposition.
Grain-size distribution curves of correlated tephra pairs from Cerro los Quesos and Cerro la Bruja, and one pair from Cerro Blanco and Cerro Hueco la Zorra. Distribution is unimodal or bimodal, particle diameter is from extremely fine-grained to very fine-grained ashes. Particle size is shown as both micrometres and ϕ.
High-resolution stratigraphy and palaeontological implications in the Pisco Formation
The mineral chemistry data recently presented for a few tephra of the Pisco Formation by Gariboldi et al. (2017) showed compositionally different sanidine–plagioclase pairs in different rhyolitic tephra, suggesting that mineral chemistry could be used to fingerprint ash layers in the Pisco Formation for long-distance correlation purposes. In the present study, the overall silicic calcalkaline tephra in the upper Miocene portion of the Pisco Formation have revealed a variety of features useful for identification, including not only biotite and glass composition, but also grain-size distribution, mineral assemblage and shard morphology. These features, together with the frequency of tephra in the successions, allow detailed stratigraphic correlations among localities several kilometres apart, which can highly improve the current stratigraphic framework.
In Figure 8, specific parts of the measured stratigraphic sections at the localities of Cerro los Quesos (between 60 and 120 m abs) and Cerro la Bruja (between 160 and 220 m abs), dated from 7·55 to 6·71 Ma based on 39Ar–40Ar dating, are correlated using the fingerprinted ash layers. In particular, Ti and Al concentrations and the MgV of biotite phenocrysts are clearly typical of each layer and can identify six pairs of the same tephra in the two stratigraphic sections. SiO2 content in volcanic glass can also be useful for fingerprinting and recognizing one more tephra pair (CLQ-T12–LB-T7). Granulometric parameters (primary and secondary mode, median, sorting) and shard morphology (shown in Supplementary material, Table S1) are used to confirm these correlations. The identification of these correlated ash layers allows us to increase the stratigraphic detail of a previous correlation based on chronostratigraphy (Di Celma et al. 2017). Taking into account 39Ar–40Ar ages in both localities (see Fig. 8) and the sedimentation rate calculated by Gariboldi et al. (2017), tephra correlation based on tephra fingerprinting allows stratigraphic reconstruction with resolution <0·1 Ma.
Correlation between the stratigraphic sections measured at Cerro los Quesos and Cerro la Bruja. Scaling-up of the sections shows the accurate stratigraphic position of tephra layers chemically correlated (coloured and in bold). Dashed lines connect correlations based on stratigraphic position of marker beds of Di Celma et al. (2017). Tephra layers are highlighted in red. Biotite and glass composition is shown in the central diagrams: Ti and Al biotite contents are expressed as atoms per formula unit; Mgv is the ratio Mg/(Mg + FeT + Mn) measured on biotite phenocrysts; SiO2 content of glass is shown as weight per cent and recalculated to 100. Ages written in red referred to 39Ar–40Ar ages obtained from selected ash layers: the age of 7·45 ± 0·01 Ma at Cerro la Bruja agrees with ages of 7·55 ± 0·05 Ma and 6·93 ± 0·09 Ma at Cerro los Quesos, and therefore supports the correlation based on biotite and glass composition.
The tephrostratigraphic framework provided in the present paper also helps in clarifying the relationships between two of the most prominent fossil-bearing localities of the Pisco Formation, namely, Cerro la Bruja (e.g. Muizon 1988; Lambert et al. 2014, 2017a) and Cerro los Quesos (e.g. Bianucci et al. 2016a, c, and references therein; Lambert et al. 2017a; Gioncada et al. 2018b). At Cerro la Bruja, the basal sandy beds of the P2 sequence (i.e. the ‘CLB vertebrate level’ sensu Muizon & DeVries 1985; Muizon 1988) are particularly rich in fossil vertebrates (Di Celma et al. 2017). At Cerro los Quesos, in turn, most of the fossil vertebrates are found in the fine-grained sediments that form the highest portion of the P2 beds there exposed; that is, ‘member F’ sensu Di Celma et al. (2016a) (Bianucci et al. 2016a). Based on the long-distance correlation presented in Figure 8, the diatomaceous strata of member F at Cerro los Quesos are clearly geologically younger than the basal sandy beds of the P2 sequence at Cerro la Bruja, and as such, they bear two temporally distinct fossil vertebrate assemblages. The taxonomic compositions of these assemblages are largely overlapping (e.g. Di Celma et al. 2017) and some of the observed differences probably reflect different palaeoenvironmental conditions (e.g. the shallow-water sandstones of the CLB vertebrate level feature abundant remains of seals, which in turn are much rarer in the offshore mudstones of member F); nevertheless, in the light of our results, various taxonomic novelties recorded from member F can be interpreted as evidence that some faunistic turnover occurred, involving the disappearance (i.e. local or global extinction) of the pontoporiid dolphin Brachydelphis mazeasi (an extinct species closely related to the extant La Plata river dolphin; Gioncada et al. 2018a) and the appearance (i.e. first arrival or origination) of phocoenids (porpoises) and kogiids (pygmy sperm whales).
By comparing Cerro los Quesos and Cerro la Bruja, we also observe that some tephra do not occur in both localities. The reason for that should be sought in some syn- or post-depositional disturbances in the depositional environment, such as marine currents and the overall shallow water context, which cause a laterally discontinuous deposition or preservation of tephra.
In this study, we demonstrate that even in such a complex sedimentary environment, where the overall similar chemical composition of ashes could limit tephra discrimination and the action of marine currents could affect tephra lateral continuity and preservation, a petrographic and chemical fingerprinting can be applied successfully, even when dealing with pre-Quaternary successions. Also, we increase the chronostratigraphic detail and reconstruct a high-resolution stratigraphy in the Pisco Formation, which hosts a marine vertebrate Fossil-Lagerstätte with a worldwide palaeontological relevance. Finally, it is worthy of note that forearc basins, such as the East Pisco Basin, and in particular the diatomaceous sequences, represent an archive of distal Central Andes tephra, which can be used to increase the knowledge on the Central Andes volcanic activity.
Conclusion
The regional dispersion and geologically instantaneous deposition of tephra layers are a great advantage for both tephrostratigraphy and tephrochronology. A chemical and petrographic fingerprinting of the distal ashes allows a layer-by-layer correlation even over relatively long distances. When supplemented by absolute age dating of suitable samples, the reconstruction of an integrated chronostratigraphic framework becomes possible.
More than 200 tephra samples were collected in the Ica Desert of Peru, along measured stratigraphic sections at different localities. In this study, we analysed 45 ash layers from the youngest marine depositional sequence of the Pisco Formation (P2, upper Miocene) focusing on the fossil-rich localities of Cerro los Quesos and Cerro la Bruja to verify whether tephra fingerprinting can be used for correlating localities that are kilometres apart in this complex scenario. Difficulties found in the Pisco Formation in using tephra as a correlation tool reside in the overall fairly monotonous composition and the presence of syndepositional disturbances in deposition, such as the action of marine currents in shallow settings. Despite these issues, the petrological–chemical diversity of the volcanic ashes deriving from the Andean volcanoes is sufficient for identifying single tephra layers. This encourages use of the major element composition of glass and biotite, aided by glass morphology and grain-size patterns, to fingerprint and correlate stratigraphic sections across the basin.
Tephra correlation based on petrographic and chemical fingerprinting is a very promising tool for correlating distant stratigraphic sections, increasing the chronostratigraphic resolution of lithostratigraphically correlated localities, in a laterally varying sedimentation environment, such as that of the Pisco Formation. Correlations obtained for the localities of Cerro los Quesos and Cerro la Bruja prove that tephra fingerprinting and tephrostratigraphy could be extended to the other localities of the Ica Desert. Therefore, there are still major challenges in identifying equivalent stratigraphic sections among other localities and correlating over wide distances. The results of the present study also highlight the preservation potential for ash deposits of the diatomaceous sequences of the forearc basins and provide a much-needed basis for future investigation of a high-resolution late Miocene record of distal tephra from the Central Andes.
Acknowledgements
The authors are grateful to R. Varas-Malca and W. Aguirre for their assistance in the field and at the Natural History Museum in Lima. Particular thanks go to V. Barberini for the help in 39Ar–40Ar analyses and to N. Fusi for the granulometric analyses. The authors wish to thank A. Risplendente for the help with the microprobe analyses. The authors are also grateful to an anonymous reviewer and to S. Watt, who gave valuable comments and suggestions.
Funding
This study was supported by grants from the Italian Ministero dell'Istruzione dell'Università e della Ricerca (PRIN Project 2012YJSBMK) and by a National Geographic Society Committee for Research Exploration grant (9410-13) to G. Bianucci.
Scientific editing by Sebastian Watt
- © 2018 The Author(s). Published by The Geological Society of London. All rights reserved
References
Chemical and petrographic fingerprinting of volcanic ashes as a tool for high-resolution stratigraphy of the upper Miocene Pisco Formation (Peru)
