Bull Volcanol (2004) 66:392–416 DOI 10.1007/s00445-003-0321-y

RESEARCH ARTICLE

Richard J. Brown · Michael J. Branney

Event-stratigraphy of a caldera-forming ignimbrite eruption on Tenerife: the 273 ka Poris Formation Received: 20 August 2002 / Accepted: 9 September 2003 / Published online: 25 November 2003
Springer-Verlag 2003

Abstract The 273 ka Poris Formation in the Bandas del Sur Group records a complex, compositionally zoned explosive eruption at Las Caadas caldera on Tenerife, Canary Islands. The eruption produced widespread pyroclastic density currents that devastated much of the SE of Tenerife, and deposited one of the most extensive ignimbrite sheets on the island. The sheet reaches ~ 40-m thick, and includes Plinian pumice fall layers, massive and diffuse-stratified pumiceous ignimbrite, widespread lithic breccias, and co-ignimbrite ashfall deposits. Several facies are fines-rich, and contain ash pellets and accretionary lapilli. Eight brief eruptive phases are represented within its lithostratigraphy. Phase 1 comprised a fluctuating Plinian eruption, in which column height increased and then stabilized with time and dispersed tephra over much of the southeastern part of the island. Phase 2 emplaced three geographically restricted ignimbrite flowunits and associated extensive thin co-ignimbrite ashfall layers, which contain abundant accretionary lapilli from moist co-ignimbrite ash plumes. A brief Plinian phase (Phase 3), again dispersing pumice lapilli over southeastern Tenerife, marked the onset of a large sustained pyroclastic density current (Phase 4), which then waxed (Phase 5), covering increasingly larger areas of the island, as vents widened and/or migrated along opening caldera faults. The climax of the Poris eruption (Phase 6) was marked by widespread emplacement of coarse lithic breccias, thought to record caldera subsidence. This is inferred to have disturbed the magma chamber, causing mingling and eruption of tephriphonolite magma, and it Editorial Responsibility: J. Stix R. J. Brown · M. J. Branney ()) Geology Department, University of Leicester, University Road, LE1 7RH Leicester, England e-mail: [email protected] Tel.: +44-116-2523647 Fax: +44-116-2523918 Present address: R. J. Brown, Osservatorio Vesuviano, Via Diocleziano 328, 80124 Napoli, Italy

changed the proximal topography diverting the pyroclastic density current(s) down the Gimar valley (Phase 7). Phase 8 involved post-eruption erosion and sedimentary reworking, accompanied by minor down-slope sliding of ignimbrite. This was followed by slope stabilization and pedogenesis. The fines-rich lithofacies with abundant ash pellets and accretionary lapilli record agglomeration of ash in moist ash plumes. They resemble phreatomagmatic deposits, but a phreatomagmatic origin is difficult to establish because shards are of bubble-wall type, and the moisture may have arisen by condensation within ascending thermal co-ignimbrite ash plumes that contained atmospheric moisture enhanced by that derived from the evaporation of seawater where the hot pyroclastic currents crossed the coast. Ash pellets formed in co-ignimbrite ash-clouds and then fell through turbulent pyroclastic density currents where they accreted rims and evolved into accretionary lapilli. Keywords Pyroclastic density currents · Ignimbrite · Tenerife · Caldera collapse · Accretionary lapilli · Phreatomagmatic · Stratigraphy

Introduction The 273 ka Poris eruption, recorded by the Poris Formation (Brown 2001) was one of over seven major Quaternary ignimbrite eruptions in southern Tenerife (Brown et al. 2003). It devastated at least 600 km2 of Tenerife, deforested and eroded the upper slopes, and deposited a large but unknown quantity of pyroclastic material in the ocean. It deposited a cream-white, phonolitic, compound ignimbrite sheet, up to 40-m thick, that comprises massive and variously stratified lapillituffs and lithic breccias intercalated with thin but widespread ashfall and pumice fall layers that show systematic regional thickness changes. The eruption was compositionally zoned, and culminated in a phase of widespread emplacement of lithic breccias, inferred to record caldera collapse. Transverse-to-current variations

393 Fig. 1 Map showing occurrences of the Poris Formation in the Bandas del Sur. Shaded area is inferred minimum distribution for the whole formation. Stars indicate location of known Poris exposures. The Poris Formation was erupted from within the Las Caadas caldera. The numbered grid comprises 10 km intervals of the UTM Grid, zone 28. Locality names referred to in the text are shown

within the ignimbrite sheet are superbly constrained in a 5 km broad, >50 km long strip of coastal desert, between the villages of Aldea Blanca in the southwest and El Baul in the northeast (Fig. 1). This area lies ~ 15–20 km from the inferred source. Proximal to distal variations are less well understood because the pyroclastic currents deposited little on the steeper more proximal flanks of the volcano and entered the sea ~ 20 km from source. This paper describes the lithostratigraphy of the Poris Formation, interprets the eruption history and considers the origin of fines-rich lithofacies that contain abundant ash pellets and accretionary lapilli. Geological background The Las Caadas shield volcano on Tenerife is composed of basalt to phonolite lavas and pyroclastic rocks, and buries older basalt shield volcanoes. A nested caldera complex, 169 km, at its top (Fig. 1) subsided in stages during successive phonolitic explosive eruptions (Mart et al. 1994; Bryan et al. 1998b; Brown et al. 2003) and was modified by northerly-directed sector-collapses (Ablay and Hurlimann 2000). Source vents of the large phonolitic

explosive eruptions and intracaldera deposits are thought to be located somewhere within this caldera complex, but are buried by the younger Teide—Pico Viejo stratovolcano (Fig. 1). Outflow sheets (extracaldera ignimbrites) are well exposed in the caldera walls (Ancochea et al. 1990) and in an extensive pyroclastic apron on the volcano’s southeast coastal flanks, known as the Bandas del Sur (Bryan et al. 1998b; Brown et al. 2003; Fig. 1). The Poris Formation is part of this coastal pyroclastic apron. Submarine fallout tephra from Las Caadas is recorded as far as 150 km ENE of the caldera (Rodehorst et al. 1998). More than seven extensive ignimbrite sheets have so far been recognized within the Bandas del Sur succession, intercalated with numerous pumice fall deposits, other small ignimbrites of unknown extent, alluvial sediments, paleosols, lavas and scoria cones (Bryan et al. 1998b; Brown et al. 2003 and references therein). The ignimbrite sheets share many characteristics, including widespread basal Plinian layers, thin ignimbrite veneer deposits with abundant diffuse bedding and complex lateral and vertical grading patterns, internal erosion surfaces, and widespread layers of lithic breccia that probably record successive caldera-forming eruptions (see Brown et al. 2003). Despite excellent exposure, little has been pub-

394

Fig. 2 Composite vertical section through the Poris Formation as seen at the reference section (Montaa Magua; Fig. 1) with summary description and interpretation of Members, flow-units and eruptive phases. Graphic log illustrates lateral variations from

valley-fill facies into topography draping veneer facies (see text for full description and interpretation of members). For description of lithofacies codes see Table 1. Lithofacies nomenclature adopted from Branney and Kokelaar (2002)

lished on the individual deposits (see Bryan et al. 1998a, b; Bryan et al. 2000). Deposits of the Poris eruption were named by Bryan et al. (1998b) after the type area near the village of Poris de Abona (Grid Reference 5915 1615; see Fig. 1). They were divided into two: “POR-A”, a lower, thin-bedded unit of alternating pumice and fine ash layers with accretionary lapilli; and “POR-B”, an upper, generally massive, crystal and lithic-poor ignimbrite, ~ 8-m thick, locally with ash layers containing accretionary lapilli (Bryan et al. 1998b). We describe the Poris Formation with reference to a thicker and more complete section, 800 m NW of a scoria cone named Montaa Magua (2.6 km NNW of Poris de Abona; Fig. 1). This section, henceforth referred to as the Magua section, benefits from the following: (1) It is stratigraphically the most complete section known, with a cumulative thickness in excess of 70 m (Fig. 2); in contrast, the former type locality is condensed, lacks six of the ten members and its top. (2) It exhibits both flow-parallel and flow-transverse laterally continuous exposures, which

illustrate the relationships between contrasting facies within both valley-fills and topography-blanketing veneers. Deposits of the Poris eruption have recently been recognized more extensively across Tenerife (Brown 2001; Brown et al. 2003), including correlative deposits within the Diego Hernandez Formation on the caldera wall and on northern flanks (Wolff et al. 2000; Edgar et al. 2002), and the geochemistry has been described (Edgar et al. 2002). The present paper focuses on the main Poris depocentre, the Bandas del Sur. The Poris ignimbrite is not welded. Post-depositional induration of much of the lapilli-tuff precludes systematic granulometric analysis. The pre-Poris landscape Ignimbrite facies and thickness variations and the presence of fossils in the pre-Poris paleosol allow some reconstruction of the landscape of SE Tenerife at the time of the Poris eruption. Although climate varied during the

395

Quaternary, this evidence, which includes the nature and spacing of plant fossils at the base of the Poris Formation (Brown 2001), suggests that the environments just prior to the eruption were broadly similar to those at the present time; the Bandas del Sur was a semi-arid gentle flank of Las Caadas volcano, cut by numerous steep- to shallowsided radial valleys (‘barrancos’), typically as much as 50-m deep and up to several hundred meters wide, separated by broad, flat interfluves or ridges, whereas the upper flanks of the volcano were covered in pine forests and had a more temperate climate. Prior to the Poris eruption, a nested caldera at the top of Las Caadas existed as a result of earlier explosive eruptions (e.g. Granadilla and Fasnia eruptions; Mart et al. 1994; Brown et al. 2003), but details of its morphology at that time are not resolved. Ignimbrite thickness variations within the Poris Formation near the coast indicate that pre-eruption fluvial incision was most marked between El Mdano and Fasnia where Bandas del Sur Group pyroclastic deposits overlie 3.3–2 Ma Lower Group lavas. Further west and east around Los Abrigos, and between Fasnia and the Gimar valley, the pyroclastic rocks overlie much younger (~ 600 ka) and less-eroded widespread flanksourced basalt lavas (Fig. 1). A broad concave valley with little near-coastal fluvial incision existed at Gimar, as a result of earlier lateral collapse event (>287€7 ka; Brown et al. 2003). Re-establishment of the drainage systems in the Bandas del Sur after the Poris eruption was by ephemeral flash floods initiated by intense rain showers dominantly on the upper flanks of the volcano. This resulted in preferential erosion of the valley-filling ignimbrite compared to the thinner deposits perched on interfluves, which were better preserved. Terminology In this work, the term ‘caldera’ is used exclusively for a topographic depression caused by subsidence into a subvolcanic magma chamber in response to magma withdrawal. We use the term valley-fill to describe parts of an ignimbrite sheet that were deposited within paleovalleys; no particular emplacement mechanism is implied. We use the term ignimbrite veneer, similarly in a non-genetic sense, to describe relatively thin, topographydraping ignimbrite deposited on paleo-topographic highs; no inference of deposition from the ‘tail’ of a pyroclastic density current is implied (e.g. as was by Walker et al. 1981). In the Bandas del Sur, the irregular gently sloping nature of the topography means that at many locations ignimbrites are intermediate in character between thick valley fills and thin topography-draping veneers. An example of this is the deposits at paleovalley margins. Lithofacies (e.g. massive lapilli-tuff; mLT) are defined using a non-genetic terminology based upon internal sedimentary structure, grainsize and lithology (see Table 1 for descriptions and interpretations of the lithofacies, and an explanation of lithofacies abbreviations used in the text). Where possible, we avoid using terminology that

may carry genetic implications, such as ‘Layer 1 and 2a’ of Sparks et al. (1973). Many lithofacies contain ash aggregates, which we subdivide into (1) ash pellets, which we define as subspherical coarse sand to lapilli-sized ash aggregates that lack concentric laminations, and (2) accretionary lapilli, in which one or more concentric laminations of fine ash enclose a core of non-laminated aggregated ash or a central pumice or lithic lapillus. The identification of flow-units and flow-unit boundaries within ignimbrite sheets can be an important step towards understanding an ignimbrite sheet’s internal architecture and depositional history. In this study, the terms ‘flow-unit’ and ‘flow-unit boundary’ follow the definition of Ross and Smith (1961) and Branney and Kokelaar (2002), and are invoked where it can be demonstrated that the passage of the current stopped, however briefly, at the location under consideration. Criteria that indicate a cessation of a pyroclastic current includes the presence of a pumice fall or ashfall layer, a reworked horizon, and/or a water-rilled scour surface. We diverge with some recent practice in which flow-units have been inferred solely on the basis of bedding surfaces, scours, grading patterns and pumice concentrations (e.g. Sparks 1976), because, although such interpretations may be correct in some cases, bedding surfaces, scours and inverse grading also can form as a result of fluctuations in flow competence, velocity and clast-concentration during the sustained passage of a single current (see Branney and Kokelaar 2002) so the criteria do not alone provide unequivocal evidence for a cessation of flow. The Poris ignimbrites locally exhibit impersistent diffuse thin bedding, internal scours, and a wide variety of pumice and lithic grading patterns. All gradations between diffuse grading and bedding surfaces occur, and bedding typically passes laterally into homogeneous and entirely massive ignimbrite that can be traced extensively (see splay-and-fade stratification, below). Where a sharp or diffuse bedding contact can be traced into a locally preserved ashfall layer, pumice fall layer or other evidence for a hiatus in flow, we invoke a flow-unit boundary. In the absence of positive evidence for abundant discrete pyroclastic density currents, we take the simpler explanation that much of the internal diffuse bedding, grading, and impersistent scour surfaces developed during progressive aggradation from largely sustained currents, the lower flow-boundary zones of which varied spatially and temporally, producing many of the observed grainsize variations (Branney and Kokelaar 2002). On this basis, and mindful that some flow-units may have been obscured, we conservatively divide the Poris ignimbrite into a minimum of four flow-units (flow units 1–4; Fig. 2), with each flow-unit boundary marked at least locally by a thin ashfall or pumice fall layer. The lithostratigraphy of the Poris Formation The Poris Formation (Brown 2001; Brown et al. 2003) comprises ten members (Fig. 2). These are interpreted as

Description

mLT (massive lapilli-tuff) acc, contains accretionary lapilli l = lithic-rich

Lithology: variable proportions of pumice and lithic lapilli matrix-supported in poorly sorted, fine- to coarse-grained tuff; pumice lapilli 20–30 vol%; lithics 2–10 vol%. Pumice clasts sub-rounded, recess-weathered and <30 mm in diameter; lithics <25 mm and sub-rounded to angular. Sub-facies: massive lapilli-tuff with accretionary lapilli (mLTacc); as mLT; contains scattered, multi-rimmed accretionary lapilli. Lithic-rich massive lapilli-tuff (lmLT); as mLT; 20–25 vol% lithic lapilli. Structure: massive; non-graded; localized, diffuse, discontinuous stratification 10–30 cm thick; mLTl often exhibits normally-graded lithics, c.15 cm thick inverse-graded layers (denoted by i) common at base of thick units of mLT. Geometry/occurrence: lenticular deposits infilling topography; max thickness 20 m; mLT passes laterally into mTa and bTa; mLTa passes laterally into bTa, mTa and vertically into mLT, sLT, and tLbr; lmLT passes laterally into mLT, mlBr. dbLT (diffuse bedded Lithology: dbLT—as mLT. ldbLT—as mLTl. lapilli-tuff) Structure: internally massive, decimeter-thick beds with gradational, sub-parallel l = lithic-rich or undulating boundaries; occasional discontinuous, weak internal stratification defined by subtle grainsize variation; lower boundaries often defined by diffuse, fines-rich horizons; small scale, low-angle scours common at base. ldbLT—as dbLT. Diffuse, solitary, 5–30 cm horizons; imbrication common. Geometry/occurrence: dbLT typically in multiple units. Individual beds laterally persistent over 2–10 m and often pinch out upslope. Passes laterally into pLT and vertically into sLT. ldbLT persistent over several meters; passes vertically into dbLT, mLT, bTa and mLTl. Dies out into valley-fill sections. dsLT (diffuse-stratified Lithology: as mLT. lapilli-tuff) Structure: discontinuous, diffuse, sub-parallel, cm-scale stratification, in sets ~5 cm to 3 m thick; defined by alternation of fines- and lapilli-rich horizons; very low-angle truncations and lenses common; individual strata persistent for several meters before dying out, amalgamating or bifurcating. Crude normal and inverse grading present. Geometry/occurrence: passes laterally and vertically into sLT and mLT. pLT (pumice-rich Lithology: clast-supported, sub-rounded to sub-angular coarse pumice-lapilli in a sparse, lapilli-tuff) fine-tuff matrix; plensLT—pumice 8–20 cm; concentrated towards tops of beds; fines-rich matrix. Structure: plensLT—internally massive; discontinuous, low-angle lenses; <20 cm thick; diffuse lower contacts; sharp upper contacts; pumice clasts non-graded, or very crudely inverse-graded. pLT(pod) - irregular decimetre pods; common association with branch/shrub moulds; some in situ. Geometry/occurrence: plensLT – upper parts or lateral edges of mLT and dbLT. pLT(pod) – occurs in close contact with substrate; typical association with shrub and branch moulds. Lithology: fine pumice and lithic lapilli supported in poorly sorted tuff matrix. Structure: xsT/ xsLT (cross-stratified tuff low-angle cross-stratification; strata dips 5–15 up- and downstream; defined by diffuse, lithic-, fines- and pumice-rich horizons; 2–15 cm thick. and lapilli-tuff) Geometry/occurrence: discontinuous sets between 8 cm to 90 cm. Isolated dune bedforms with wavelength of ~16 m; gradational lower contact with sLT; unconformably overlain by lsLT; passes vertically/laterally into bTa, dbLT and mLT

Lithofacies

Table 1 Summary descriptions and interpretations of the lithofacies in the Poris Formation

Tractional sedimentation from pyroclastic density currents; poor sorting indicates deposition from a flow boundary transitional between traction-sedimentation and granular flow (Branney and Kokelaar 2002).

plensLT—records accumulation of pumice at the lateral margins of the current (pumice levees). Concentration at the tops of beds and fine–grained matrix record deposition during waning conditions. pLT(pod) - represents vegetation-trapping of pumice clasts in current by in-situ, or transported, shrubs and branches.

Rapid progressive aggradation from high-concentration fluid-escape to granular-flow dominated flow 35 of 38 boundary zone of a pyroclastic density current (Branney and Kokelaar 2002). Diffuse stratification and disordered structure records unsteadiness during deposition.

Massive internal structure, composition and lack of internal traction structures indicate deposition under similar conditions as mLT; diffuse bedding and low angle truncations indicate current unsteadiness; thickness variations reflect current non-uniformity.

Rapid progressive aggradation from high-concentration fluid-escape dominated flow-boundary zone of pyroclastic density currents (Branney and Kokelaar 2002); mLTacc as mLT but accretionary lapilli fell into current from overlying upper parts of the current; lmLT as mLT, higher lithic content reflects increase in supply at source. Inverse grading records deposition during waxing flow (Branney and Kokelaar 2002).

Interpretation

396

Lithology: matrix- to clast-supported, pebble to boulder-grade lithic clasts in poorly sorted lapilli-tuff matrix; max clast size ~ 2 m, average 5–20 cm; angular to rounded clasts; exfoliation rinds; sub-ordinate rounded pumice (<3 cm) and scoria clasts (<50 cm) present (2–10%; absent in clast-supported breccia). Structure: massive; imbrication generally absent; large boulders aligned long-axis parallel to flow; normal grading common. Geometry/occurrence: highly variable—sheet form, lenticular, and isolated, irregular pods; lower contact commonly loaded; thickness varies from 10s of cm to 5 m; passes vertically into mlBr(nl) and laterally into lmLT. Lithology: as mlBr Structure: dm-thick beds with diffuse boundaries; beds defined by grainsize and/or packing density; tabular, lenticular or low-angle dune bedforms. Imbrication common. Geometry/occurrence: occurs in veneer ignimbrite; passes laterally into mlLT and mLBr. Lithology: fine-grained, poorly sorted tuff; minor pumice and lithic lapilli (<2 vol%). Structure: internally massive; to diffusely stratified; boundaries sharp to gradational; lateral thickness variations common. mTpel – clast-supported pellets. Geometry/occurrence: mantles topography; 1–300 cm thick; lenticular, valley-margin deposits that thin (to <20 cm) over paleo-ridges; passes laterally into mT, mLT and mLTacc; mTpel – drapes topography.

mlBr (massive lithic breccia)

Records efficient elutriation associated with conditions in the flow boundary. Possibly due to ingestion of air, combustion of vegetation, or gas-escape associated with compacting substrate.

Emplacement by either dilute density currents or by finegrained, vitric ash-fall reworked by surface winds.

Pumice-fall deposit indicated by sorting, composition, mantle bedding and lateral extent. Grainsize stratification reflects unsteadiness in the eruption column and shower bedding.

Fine-grained valley-margin and topographic-high equivalent of mLT see above. mTpel –co-ignimbrite ashfall layers; mTacc - fallout of pellets into pyroclastic density current.

As mlBr. Diffuse bedding indicates current unsteadiness; thickness variations reflect current non-uniformity. See also Bryan et al. (1988b).

Co-ignimbrite lithic breccia. Deposited from high-concentration flow-boundary zone of lithic-rich pyroclastic density currents (Branney and Kokelaar 2002). Diffuse bedding and low angle truncations indicate current unsteadiness; thickness variations reflect current non-uniformity. See also Bryan et al. (1998b).

Interpretation

Lithofacies abbreviations follow the scheme of Branney and Kokelaar (2002). Composition: T, tuff; L, lapilli; LT, lapilli-tuff; LBr, lithic breccia. Structure: m, massive; b, bedded; //b, parallel bedded; db, diffuse bedded; s, stratified (tractional discontinuous stratification); xs, cross-stratified; lens, lenticular bed; pod, podiform beds; F-poor, fines-poor; l, lithic-rich; p, pumice-rich; acc, contains accretionary lapilli; pel, contains ash pellets; pip, elutriation pipes; i, inverse grading; n, normal grading (il, inverse graded lithics; np, normal graded pumice)

Lithology: framework-supported, pumice and lithic lapilli; well-sorted; (sj 0.75–2.1); pumice clasts <10 cm, sub-angular to angular, often impact-fractured; varying proportions (25%) of lithic lapilli, <5 cm. Structure: generally massive with diffuse stratification defined by grading patterns; parallel boundaries. Geometry/occurrence: sheet form; mantles topography; up to 1.3-m thick sT (stratified tuff) Lithology: fine-grained tuff supporting 5–25 vol% coarse tuff-grade pumice and lithic clasts. Structure: thin, sharp-bounded tractional strata 2–5 mm thick with sight thickness variations over 10s of cm. Geometry/occurrence: laterally persistent over kilometers; intercalated with mp-bT. f-poor lLT (fines-poor, Lithology: clast-supported lithic clasts in a sparse vitric tuff matrix lithic-rich lapilli-tuff) Structure: internally massive with scours/loads at base. pip, elutriation pipes Geometry/occurrence: discontinuous; lenticular; passes upwards into mLT/bLT(i)

mL (massive lapilli)

mT (massive tuff) acc, contains accretionary lapilli pel, contains ash pellets

dblBr (diffusely bedded lithic breccia)

Description

Lithofacies

Table 1 (continued)

397

398

time slices representing different phases of the eruption. Each member is defined on the basis of distinctive features and its stratigraphic position, and each can be traced regionally. Most members are lithologically diverse and show marked lateral and longitudinal variations across an irregular paleotopography. The base of the Poris Formation is defined at the Magua section, and is the contact between the pumice fall layers of Member 1 and a subjacent orange-brown paleosol, 0–80-cm thick, developed in the top of the 289€6 ka Fasnia Formation (Fig. 2; Brown et al. 2003). At some locations however (e.g. at Poris de Abona and at El Arrecife quarry; Fig. 1), the lower members of the Poris Formation are absent, so that Members 7 or 8 of the Poris rest directly upon the Fasnia Formation. This is because of localized non-deposition or erosional removal of lower members of the Poris Formation by Poris pyroclastic density currents. The base of the Poris Formation locally oversteps the paleosol and the Fasnia Formation to rest directly upon the 600€7 ka Granadilla Formation (Brown et al. 2003), and in some places on older phonotephrite and basalt lavas (Fig. 2). These relations are the result of erosion prior to the Poris eruption. The upper surface of the Poris ignimbrite sheet is smooth and slopes 1.5–3 seaward, similar to the general slope of the substrate. A brown paleosol, <120-cm thick, is developed locally within the uppermost part of the

formation (this is variously within Member 10, at Magua, and Member 9 at Tajao, Fig. 1). Locally within the Gimar valley, a thin well-sorted layer predominantly of lapilli of mafic pumice and scoria overlies Member 9. The mafic lapilli are up to 2 cm in diameter, and some are banded and streaky. The layer also contains subordinate phonolite pumice lapilli, commonly hydrothermally altered lithic lapilli, and crystals of alkali feldspar and amphibole. It locally shows evidence of grainflow and is locally obscured by pedogenesis. It may record a final stage mafic Plinian fall deposit of the Poris eruption (Edgar et al. 2002; Brown et al. 2003). However, it does not outcrop elsewhere on Tenerife, is insufficiently preserved to determine its distribution, and possibly derives from one of the numerous local, post-Poris, strombolian cones in the Gimar valley. Edgar et al. (2002) also interpret fines-poor, lithic-rich deposits with ‘four juvenile types’, channel scours, ripple cross-lamination, and load-and-flame structures as an eroded remnant of another Poris ignimbrite (their ‘Confital ignimbrite’) overlying Member 9 at Gimar and La Mareta. We interpret these deposits as sedimentary in origin: aqueous deposited pumiceous sands and gravels at La Mareta, and the results of pedogenesis and downslope debris flow around Gimar. Impersistent reworked volcaniclastic deposits of this type are common between most eruptive packages in the Bandas del Sur.

Fig. 3 a Selected graphic logs through Member 1 pumice fall deposits. Note that individual beds have different distributions, although a lack of outcrop of the lower beds precludes isopach construction (for isopach map of Bed H see Edgar et al. 2002). Inset shows inferred dispersal axes for Beds A-G and H, and the location

of logged sections. Modified from Brown et al. (2003). b Histograms of the granulometry of Beds B, D, E, G and H (from sites where layers are non-indurated). N/A represents phi sizes that were not analyzed for pumice, crystal and lithic content

399

There is no evidence elsewhere within the formation for pauses during the eruption sufficient to allow significant sedimentary remobilization, erosion, reworking, or pedogenesis. It is inferred that the Poris Formation represents a single eruptive unit that records a single explosive eruption, sustained over a period of possibly hours to days rather than months to years. This is broadly consistent with durations of historic caldera-forming explosive eruptions elsewhere (e.g. Pinatubo, Scott 1996). The formation is variously overlain by the Sabinita, La Caleta (e.g. around Poris de Abona; Fig. 1), and Abrigo formations (see Brown et al. 2003).

Eruption history The definition of the internal lithostratigraphy within the Poris Formation has allowed the discrimination of eight eruptive phases. Each phase is different in character to the preceding and succeeding phases, although some phases may have been very short-lived. Most are recorded by a single lithostratigraphic member, although Phase 3 produced three ignimbrite flow-units, each comprising a member (see Fig. 2). We note that extracaldera successions may incompletely record eruption histories, and it is possible that some other phases are represented only by pyroclastic deposits within the caldera (e.g. see Valentine et al. 1992). Phase 1: Plinian pumice fall and ashfall The onset of the Poris eruption is recorded in the Bandas del Sur by regionally extensive (400 km2), topographydraping pumice fall and ashfall layers that together comprise Member 1 (Fig. 3). Five pumice fall layers (Beds B, D, E, G and H) occur interstratified with three thin (2 cm) layers of fine tuff (Beds A, C and F; Fig. 3 and Fig. 4a).

Fig. 4 Member 1 pumice fall deposits. a Lower beds at Tajao (Fig. 1). Beds B, D, E and F composed of massive and diffusebedded lapilli (mL and dbL). Beds A and C comprise thin laminae of massive tuff (mT). Coin is 17 mm across. b Bed H at La Hidalga in the Gimar valley (Fig. 1). It directly overlies a paleosol developed in the top of the underlying Fasnia Formation (see Brown et al. 2003; lower Beds are absent here), and is overlain by Member 2B. The upper third of Bed H is often extensively altered. Scale in 10 cm divisions

Pumice fall layers of Member 1 (lithofacies pmL and p//bT) The pumice fall layers (Beds B, D, E and G) comprise massive pumice-rich coarse ash to lapilli (lithofacies pmL, Table 1) with angular, framework-supported pumice clasts (15 mm; Fig. 4a), subordinate green obsidian clasts and lithic clasts (3–5 mm), and rare loose crystals (<2.5 wt%; alkali feldspar, biotite, plagioclase, amphibole and hayne). The layers are well sorted (sj=0.75–2.1; sj after Inman 1952; Walker 1971), slightly positively skewed, and show both inverse and normal grading (e.g. Beds D and E, Fig. 3). Beds B and E are distinctively rich in lithic clasts (>11 wt%) and Bed D is rich in conchoidal, platy to blocky, obsidian clasts (23 wt%), which show micron-scale vesicle banding and perlitic cracks. The pumice is pale green phonolite (Fig. 5) with rare phenocrysts. Pumice in Bed H contains sporadic, irreg-

Fig. 5 TAS classification graph (Le Bas et al. 1986) for whole rock pumice samples collected from the Poris Formation during this study (XRF analyses normalized to 100% to facilitate comparison with published analyses). 1Edgar et al. (2002); 2microprobe analyses of Bryan et al. (2002)

400

ular-shaped dark grey clots and streaks of mafic pumice, which are associated with larger vesicles. The different pumice layers have different geographic distributions and grainsize characteristics (see Fig. 3). The lowest layer is the most geographically restricted, and the uppermost Bed (H) is coarsest and thickest. The lower beds (B, D, E) occur between Fasnia and Aldea Blanca, whereas Bed H has a more easterly distribution (Fig. 3) and extends from Tajao in the south to north of El Baul (~ 300 km2), where it is still 80-cm thick. Bed H thickens to 130 cm towards Gimar (Figs. 4b and 6b), although the largest lithic clasts (37 mm) occur within its basal 10 cm further south, near Fasnia, where Bed H is only 30-cm thick (Fig. 3). In the north, Bed H overlaps underlying layers to rest on the pre-Poris paleosol (Fig. 3). A thin (5 mm) layer rich in lithic clasts (8 mm) forms the basal part of Bed H north of Fasnia. Interpretation: Phase 1 of the Poris eruption was marked by widespread pumice fallout over the Bandas del Sur (Fig. 7). It started intermittently, with four pumicefall events (Beds B-E and G), but the eruption became more sustained and more vigorous with time (thick, massive Bed H). The time gaps between the successive pumice fall events are not known (see below). Unsteadiness (waxing and waning mass flux and thus column height) in the early eruption column(s) may account for the laterally persistent grading patterns within the lower layers, whereas the localized diffuse bedding seen in the Bed H may record unsteady deposition due to the effects of wind. Fragmentation of a dense juvenile obsidian source is indicated by the abundant platy to blocky obsidian clasts, e.g. in Bed D (cf. similar deposits described by Smith and Houghton 1995). Thickness and maximum lithic data are insufficient for detailed isopleth and isopach maps of Beds B, D, E and G, but provide a transverse section across the inferred dispersal axes, and they show that the dispersal axes shifted from SE to E with time (Fig. 7). The first pumice fall was restricted to a narrow (20 km wide) SE sector (Bed B), but this distribution expanded to a 30 km sector (Bed D; Fig. 3). The orientation of the dispersal axis then shifted further southeast (Beds E and G; Fig. 3), and then to the East as the eruption developed into a sustained and significantly higher Plinian eruption column (Bed H, Fig. 7; also see data in Edgar et al. 2002). This last change was accompanied by a brief vent-clearing or ventwidening episode (the lithic-rich base). During this sustained Plinian event, the dispersal axis may have continued to shift further NE, because although the largest lithic clast diameters in the base of Bed H are centered over Fasnia, the maximum thickness of Bed H occurs further north, over Gimar. The differences in dispersal are inferred to result from the rising eruption column encountering differing wind speeds and directions at different altitudes at different times (e.g., as is inferred to have happened during fallout from Laacher See volcano in Germany; Bogaard and Schmincke 1984). Changing vent positions may also have played a role, but cannot account for all the variation.

Beds A–G have been interpreted as phreatomagmatic ‘pyroclastic surge’ deposits (Edgar et al. 2002). We favor a fallout origin due to the lateral persistence of each bed (over 10 km, 15–20 km from source), the topographydraping relations, the very well-sorted nature of the beds (see Figs. 3 and 4a), the systematic regional thickness changes of each layer that are independent of local topography, the angular nature of the fresh pumice clasts (clay alteration at some clast edges gives an impression of rounding), and symmetric pumice impact structures that suggest near vertical fallout (base of Bed D). These beds also exhibit symmetric rather than asymmetric draping relationships across shrub moulds preserved along the base of the formation (e.g. at Aldea Blanca; Fig. 1), which would not be expected if deposited from a lateral pyroclastic current. The blebs and streaks of dark grey pumice associated with large vesicles in the pale green pumice of Bed H suggests that hotter, more mafic magma may have locally enhanced vesiculation of the enclosing phonolitic pumice. Edgar et al. (2002) obtained a phonotephrite composition for these blebs, and considered that the injection of this mafic component into a largely phonolitic Poris magma chamber was the trigger for the Poris eruption. Fine tuff layers of Member 1 (lithofacies mT, //bT) Intermittent fine ash deposition during Phase 1 is recorded by thin (2 cm) beds (Beds A, C and F) of poorly sorted fine vitric tuff (lithofacies mT; Fig. 3) intercalated with the pumice fall beds described above. The ash beds are generally massive, although faint lamination is seen locally. The fine vitric tuff beds mantle topography. Locally they exhibit subtle (2 mm) lateral thickness variations across decimeters to meters, independent of the underlying topography. Their distributions differ (Fig. 3); Beds A and C trace laterally for 15 km whereas Bed F traces laterally for 25 km. A thin (2 mm) lithic-rich layer occurs within Bed F at some locations. The top of Bed C exhibits symmetrical impact sags from clasts from overlying Bed D. At Aldea Blanca, layers within the base of the formation drape moulds of draped in-situ shrubs, from which calcified roots project down into the underlying paleosol. The fine tuff layers do not thicken or thin significantly across these. Interpretation: fine ash deposition alternated with the pumice fall during the first phase of the Poris eruption. The fine tuff beds lack obvious tractional structures, and the symmetrical draping on both stoss and lee sides of shrubs is indicative of an ashfall origin, possibly with moist and cohesive ash. In this case the subtle (mm-scale) local thickness variations that are independent of topography may record effects of wind and/or percolation of ash into underlying openwork pumice beds. However, it is also possible that the ash layers record deposition from very dilute but widespread pyroclastic density currents that rolled across the landscape too slowly to cause traction, and deposited a thin mantling layer of fine ash

Fig. 6

t

401

402

Fig. 6 Representative graphic logs through the Poris Formation in the Bandas del Sur (selected from a total of 45 logged sections). a Sections through the formation in the southwest Bandas del Sur (Aldea Blanca to Tajao). b Sections through the formation in the northeast Bandas del Sur (Montaa Centinela to El Baul). Note extensive distribution of thick ignimbrite members 7–9, which were

deposited by a climactic, widespread pyroclastic density current. The entrachron (sensu Branney and Kokelaar 2002) for banded tephriphonolite pumice is also marked within Member 8. Log localities are shown in inset map, and grid references are given on 6b. Lithofacies codes are described in Table 1

403

Fig. 7 Cartoons illustrating the major eruptive phases of the Poris eruption, and their distributions across the Bandas del Sur. The eruption is inferred to have lasted between hours to days, and produced high plinian columns and fountain-fed pyroclastic density

currents. The dispersal of pyroclastic density currents increases dramatically during Phase 5 along with the initiation of caldera collapse, which is marked by the generation of lithic-rich pyroclastic density currents

404 Fig. 8 Photo-log of the thin veneer ignimbrites (Members 2, 3 and 6) and intercalated pumice fall (Members 1 and 5) and ash fall beds (Bed 2B) at Poris de Abona (Fig. 1). Note diffuse contacts between ignimbrites and pumice fall deposits due to incorporation of pumice lapilli by moving pyroclastic density currents and/or synchronous deposition. Pumice clasts are predominantly recess-weathered and altered to clay. Member 4 is absent from this locality

(e.g. see Talbot et al. 1994). In this case, the widespread absence of tractional structures indicates that deposition was from a direct fallout-dominated flow-boundary zone of a fully dilute current (Branney and Kokelaar 2002 p 37). Either way, the absence of pumice lapilli within the fine tuff layers suggests that pauses occurred between the successive (Plinian or subPlinian) pumice-fall events. The fine tuff layers may simply record settling out of residual ash suspended in the atmosphere between pumice fall events, possibly enhanced by rain flushing. Alternatively, the layers formed from brief phreatomagmatic eruptions (although pellets or accretionary lapilli are not preserved) or were deposited by co-ignimbrite plumes related to undocumented pyroclastic density currents elsewhere on the volcano (by analogy with fine tuff layers in Members 2–4, below). Phase 2: localized pyroclastic density currents and widespread co-ignimbrite ash falls During Phase 2, three discrete pyroclastic density currents (flow-units 1–3) flowed over the central Bandas del Sur and generated extensive ash plumes. These are recorded by Members 2, 3 and 4, which exhibit broadly similar characteristics (Fig. 2) and so are described together (see Brown 2001 for individual details). Members 2, 3 and 4 Members 2, 3 and 4 conformably overlie Member 1. Each consists of an ignimbrite (Beds 2A, 3A, 4A) and an overlying, thin ash pellet fall layer (Beds 2B, 3B, 4B; e.g., Fig. 8). The ignimbrites are restricted to central parts of the Bandas del Sur, between Poris de Abona and Tajao (Fig. 1), whereas the pellet-fall layers are more extensive

(between El Baul to Tajao; Fig. 1) and occur stacked directly upon one another beyond the limits of the ignimbrites (e.g. at Gimar). Members 2 and 4 are a distinctive yellow color, whereas Member 3 is grey. The ignimbrites (lithofacies mLT; dbLT; dsLT) Each ignimbrite deposited during Phase 2 is between 2 and 15-m thick in paleovalleys, and passes laterally and gradationally into a much thinner (8 cm) and more extensive veneer (see logs 4–6 on Fig. 6a; logs 9 and 14 on Fig. 6b). The valley-fill parts of the ignimbrites are massive (mLT), diffuse-bedded (dbLT) and diffuse-stratified (dsLT) lapilli-tuff (Table 2; Fig. 6). The extensive veneers deposited during Phase 2 rarely contain large pumice and lithic lapilli, and comprise massive tuff or stratified tuff (lithofacies mT, sT; Table 1; Fig. 8). They drape slopes up to 45, and exhibit local thickness variations (cm—dm) according to the underlying topography. Approaching paleovalley margins, the veneers thicken and coarsen into what is termed here valley-margin ignimbrite, and stratification tends to splay-and-fade towards paleovalley axes (see Branney and Kokelaar 2002 p 109). Lenses of pumice lapilli and pumice cobbles (plens) occur at margins of some paleovalleys (e.g. within Member 3 at Tajao). Fully exposed transverse sections from veneer to valley-fill (as, for example, are preserved in Member 6) are rarely preserved intact, and must be inferred. Basal contacts of the ignimbrites are locally erosive. Fine ash at the base of the ignimbrites has locally percolated between the pumice lapilli of underlying pumice fall layers, and lapilli can be seen incorporated into the ignimbrite (e.g. base of Member 2; Fig. 8). Basal parts of the ignimbrites commonly preserve moulds of current-orientated delicate twigs and grasses that have

405

Fig. 9a–c Ash-rich beds of the Poris Formation. a Frameworksupported pellets (Bed 2B) in Member 2 at Poris de Abona (Fig. 1). Coin is 20 mm diameter. b Accretionary lapilli in fine-grained ignimbrite veneer of Member 3. Accretionary lapilli comprise a dark core of massive ash surrounded by a concentric rim of finer

grained, light coloured ash. c SEM backscatter photomicrograph of Member 2 massive ash pellets with abundant cuspate and platy bubble-wall shards. Note very weakly developed finer-grained outer edge of center pellet. Scale bar is 1000 mm

been bent over during ignimbrite emplacement. Diffuse pods of unusually coarse framework-supported pumice lapilli and pumice cobbles (pLTlens) and, less commonly of lithic lapilli, locally surround moulds of in-situ branching shrubs, 70 cm tall. Upper parts of both the valley-fill and veneer ignimbrites commonly contain accretionary lapilli 7 mm in diameter, with a massive medium-grained ash core enclosed by a single rim of finer grained ash, 1 mm thick (Fig. 9c).

comprise framework-supported ash-clusters (<2 mm) and ash pellets (<15 mm; Fig. 9a) formed of curviplanar and Y-shaped bubble-wall shards (Fig. 9c). Some pellets have impacted or loaded a few millimeters into the tops of the underlying ignimbrites. The nature of the ash clusters and pellets and their preservation differ between layers and help distinguish between the Members. For example, in Bed 2B, armored lapilli (cf. Waters and Fisher 1971) occur, comprising a 1–3 mm layer of fine ash coating pumice or lithic lapilli. Pore space is locally preserved between pellets in this bed, and its upper ~2 cm locally comprises small, less well defined pellets that are barely distinguishable from the matrix (e.g. at Gimar, Fig. 1). Impact deformation of individual pellets is apparent in Member 2. Beds 3B and 4B however, comprise poorly defined pellets with no preserved pore space and few armored lapilli. Interpretation: Phase 2 was characterized by the passage of three discrete pyroclastic density currents across the Bandas del Sur (Fig. 7). Co-ignimbrite ashfall

The pellet-bearing tuff layers (lithofacies mTpel) The pellet-bearing tuff layers (Beds 2B, 3B, 4B) deposited during Phase 2 are up to 5-cm thick. Each can be traced from Tajao northeast to Gimar (Fig. 1), despite impersistent preservation resulting from local impactdeformation by subsequent pumice fallout (e.g. absence of layer 3B on Fig. 8) and due to erosion by succeeding pyroclastic density currents. The pellet-bearing tuff layers

406

occurred during the pauses between the currents (Fig. 7). The currents were restricted to a 15-km wide central part of the Bandas del Sur (between Tajao and Montaa Magua; Fig. 1), possibly reflecting a localized breach in the ESE caldera wall. The pyroclastic currents were density-stratified and their coarser grained, lower parts preferentially channeled down valleys and deposited the massive valley-filling lapilli-tuff, whereas their upper more dilute parts spread more evenly across SE Tenerife, surmounted topographic ridges, and deposited thin finegrained veneers. The abundant small ash pellets in the veneer deposits (e.g. lithofacies mTpel) suggest that aggregation of fine ash was important during pyroclastic density current transport and deposition. The pellets probably formed in the fine-grained, lofting upper parts of the density currents (‘co-ignimbrite plumes’) and fell through to the base of the current where they were deposited (see below). Each density current was followed by widespread fallout of ash pellets (Beds 2B, 3B, 4B) probably from coignimbrite plumes blown NE from the currents. Impacted pellets and gradational contacts at the top of each ignimbrite indicate that pellet fallout occurred immediately after the passage of the density currents (see below). The loss of definition of the pellets towards the top of Bed 2B records progressively wetter deposition with time, indicating condensation caused by cooling in the residual ash clouds, and possibly rain. The absence of angular coarse pumice lapilli between and within the Phase 2 co-ignimbrite ashfall layers suggest that that the discrete pyroclastic density currents did not originate from intermittent fountaining/collapse events from a sustained Plinian column. Rather, the intermittent pyroclastic density currents most likely resulted from rapid successive low pyroclastic fountaining events, each generating a short-lived pyroclastic density current, and possibly separated by minutes to days. Their fine grained, ash-rich nature may reflect a phreatomagmatic origin (see below) and/or they may represent the flow-stripped upper and finer grained parts of largely caldera-confined density currents whose coarser-grained lower parts were restricted by the caldera wall. The fine tuff beds and the pellet beds were both interpreted as recording discrete phreatomagmatic ‘pyroclastic surges’ by Edgar et al. (2002). However, the gradational lateral transitions into massive valley-filling ignimbrite (mLT) suggests that the fine tuff beds were deposited from fine-grained parts of density-stratified pumiceous pyroclastic density currents, whereas the lateral continuity, even draping of topography with constant thickness, vertical impact structures, and absence of stratification in the ash pellet layers indicate that these were deposited by direct fallout of ash and pellets from moist ash plumes (see below).

Phase 3: short-lived Plinian pumice fall A brief period of Plinian activity recorded by Member 5 was accompanied by a pause in the passage of pyroclastic density currents across the Bandas del Sur. Pumice fall deposit of Member 5 (lithofacies pmL) Member 5 traces 35 km between La Caleta and El Baul (Fig. 1) and thickens to 12 cm around Gimar. It comprises a non-graded topography-draping layer of framework-supported pale green pumice lapilli, <30 mm in size, and subordinate lithic lapilli, 15 mm in size (Fig. 8). It is widely cemented by zeolite, and much of the pumice lapilli and some of the lithic clasts (mostly basalt) are altered to clay. The member has an irregular lower contact with lowermost pumice lapilli impacted into the underlying loose ash of subjacent members. Southwest of Poris de Abona, the deposit is only one-clast thick, and all the clasts are matrix-supported in the fine ash of Member 3. The upper contact of Member 5 is irregular and similar in form to the top of Bed H (Member 1). Interpretation: Phase 3 records a short-lived pumice fall event of probable Plinian proportions, which dispersed tephra over the central and northeast Bandas del Sur (Fig. 7). It records the brief establishment of a new Plinian column during a pause in pyroclastic density current activity in this region. Limited preservation precludes the construction of isopach and isopleths, although the maximum thickness of 12 cm attained in the Gimar valley suggests that the dispersal axis may have trended ESE (Fig. 7), similar to that of the phase 1 Plinian fall event that deposited Bed H. The irregular lower contact indicates that initial pumice lapilli either fell into the uncompacted ash deposit of Members 3 and 4, or into the aggrading deposit of these members i.e. synchronous pumice fall and density current activity (e.g. as has been inferred elsewhere by Valentine and Giannetti 1995). Phase 4: Localized pyroclastic density current and extensive co-ignimbrite ash fall Phase 4 saw the recommencement of pyroclastic density current activity in the Bandas del Sur (Fig. 7). This was to continue through Phases 4–7. Ignimbrite lithofacies of Member 6 (mLT, mT, dbT, sT, xsT, dbTacc) Member 6 comprises a valley-fill ignimbrite (Bed 6A) that contains accretionary lapilli, and outcrops between La Caleta and Gimar (Fig. 1). It shows similar topography-induced lateral lithofacies variations within similar to those seen in Members 2–5. The valley fill ignimbrite comprises massive lapilli-tuff (mLT; Table 1), ~ 20-m

407

Fig. 10a,b Ignimbrite lithofacies in Member 7 (flow-unit 4). a Diffuse-bedded lapilli-tuff at Montaa Magua (Fig. 1). Bedding is defined by the alternation of thin fines-rich and pumice-rich horizons. Note lateral impersistence of individual layers. b Complex, polyphase scours (large scour delineated by arrows)

within veneer ignimbrite flow-unit. The pyroclastic density current experienced rapid fluctuations between erosion and deposition. Current from right and into page (looking SW; Montaa Magua; Fig. 1). Scale has 10 cm divisions

thick, that passes laterally into valley-margin ignimbrite, up to 3-m thick, formed of white-weathering, massive and diffuse-bedded vitric tuff with abundant accretionary lapilli (lithofacies mTacc – dbTacc; Log 14 on Fig. 6b). The valley margin deposits then pass into thin and impersistent, topography-draping ignimbrite veneers (cf. phase 2 deposits), which consist of stratified and diffusebedded fine tuffs (sT, xsT and dbTacc; Figs. 10 and 11) that splay-and-fade into paleovalleys. Discontinuous internal scours are common, and discontinuous lithic-rich (lithic-rich massive lapilli-tuff; lmLT) layers with loaded bases occur locally (e.g. at El Arrecife, Fig. 1) in veneer deposits, which are impersistent and often eroded out by Member 7. Accretionary lapilli occur in all lithofacies of the ignimbrite, reaching their highest concentration in the veneer deposits. However, they are absent from the lower part of the ignimbrite. They are large (<40 mm; average ~ 12 mm) and distinctive, with multiple concentric laminations of fine ash, enclosing coarser-grained massive ash cores and, more rarely, enclosing pumice or lithic lapilli. At some locations, the base of Bed 6A rests on Member 5 whereas at others, the base is erosional and cuts out Members 5 and 4 to rest directly upon Member 3 (e.g. at Tajao; Fig. 1). In central regions, the top of Bed 6A, which records the end of Phase 4, is marked by the general disappearance of accretionary lapilli. Carbonate nodules up to 8 cm in diameter are common in Bed 6A, and locally overprint lithic lapilli. In the Gimar valley, Member 6 is represented by a 6cm thick layer of framework-supported ash pellets (Bed 6B; as in Phase 2) with scattered pumice and lithic lapilli. Interpretation: density-stratified pyroclastic density currents deposited coarse massive ignimbrite along valley axes, and finer-grained bedded and stratified tuff higher on valley sides, and more extensively across topographic

highs (Bed 6A; see Phase 2; Fig. 7). Similar relationships are common in ignimbrites deposited over irregular topography (e.g., Laacher See, Schumacher and Schmincke 1990; ignimbrites of the 1912 eruption of Novarupta, Alaska, Fierstein and Hildreth 1992; the climactic ignimbrite of the 1991 eruption of Pinatubo, 1991, Scott et al. 1996). The internal grading and the localized, impersistent erosional scours above the base of the ignimbrite indicate that the pyroclastic density currents were unsteady causing fluctuations between erosion and deposition at some localities. The large accretionary lapilli indicate long residence times in turbulent, moist ash clouds, and they may have formed within vigorously turbulent upper parts of the stratified density currents, or within fully lofted co-ignimbrite plumes (see below). Their absence from the base of the ignimbrite is interpreted to reflect a lag time between initial deposition of ignimbrite and the formation and deposition of accretionary lapilli at a particular locality. The restricted easterly distribution of the Member 6 ignimbrite suggests that the Phase 4 pyroclastic density current issued from a single vent and/or through a low breach on the ESE caldera wall of Las Caadas (as during Phase 2). The occurrence of Member 6 pellet fall layer (Bed 6B) at Gimar indicates the generation and northeasterly drifting of an extensive co-ignimbrite ash-cloud (Fig. 7). Phase 5: waxing pyroclastic density current Ignimbrite lithofacies of Member 7 Phase 5 was characterized by a rapid and marked increase in the geographic distribution of a sustained pyroclastic density current as the Poris eruption waxed. It is recorded

408 Fig. 11a–d Upper members of the Poris Formation. a Sheared load structures in Member 8 lithic breccia at La Mareta (Fig. 1), indicating loading of breccia during deposition. b Sunken boulder at Montaa Magua (Fig. 1). Boulder was entrained and transported by the pyroclastic density current that deposited Member 8 (possibly locally derived). c Loaded lithic block with ‘trail’ of pebbles (at Tajao; Fig. 1). Block has sunk into Member 7 from Member 8. 50 cm exposed on ruler. d Member 9 (flow-unit 4) at Gimar (Fig. 1), overlying ashfall deposits of Members 2–6. Note imbricated tree moulds. Lower two thirds exhibits columnar jointing with joints spaced 3–5 m apart. This dies out upwards with the top of the massive lapilli-tuff (mLT). Base of Poris Formation not visible in photograph

by Member 7, which is extensive (between Fasnia and Aldea Blanca, Fig. 1) and overlaps all preceding members (1–6). Member 7, corresponds to ‘POR-B’ of Bryan et al. (1998b) and comprises variously stratified (Fig. 10), thinbedded (Fig. 11) and massive lapilli-tuff that widely varies between 2 and 5-m thick, but is locally as much as

20-m thick in paleovalleys. The lower contact of Member 7 varies from gradational to sharp (erosional), and its top contact is commonly irregular (see next section). The proportion of lithic clasts increases upwards though the member at some localities (e.g., Fig. 1 and log 8 on Fig. 6a).

409

Seven intergradational lithofacies are represented (Table 1; Fig. 6), with complex lateral and vertical lithofacies variations (see Fig. 6), including continua between diffuse-stratified, diffuse-bedded and massive lapilli-tuff (dsLT and mLT). Diffuse-bedded and diffusestratified lithofacies pass laterally, via splay-and-fade into non-graded massive lapilli-tuff at many exposures (see Branney and Kokelaar 2002), especially at valley margins transitions. An isolated antidune-like bedform with an amplitude of ~ 40 cm and a wavelength of 9 m occurs at Montaa Magua (Fig. 1). Its foresets comprise low-angle cross-stratified lapilli-tuff (xsLT) and dip ~10 downstream. The bedform gradationally overlies diffuse-stratified lapilli-tuff and passes upwards into diffuse-bedded lapilli-tuff. A giant regressive bedform occurs at Montaa Magua (Brown 2001). Member 7 exhibits many steep-sided and polyphase (cross-cutting and bifurcating) scours up to 3-m deep (e.g. Fig. 10c), and erosion surfaces and diffuse low-angle truncations that pass laterally into diffuse-bedding and then fade out laterally into massive lapilli-tuff (this is ‘scour splay-and-fade stratification’ of Branney and Kokelaar 2002, p. 110). Scattered sub-rounded intraclasts of Member 6 tuff occur at Poris de Abona. Locally derived scoria clasts occur downstream of Montaa Fasnia scoria cone, and locally sourced palagonite tuff clasts occur at the Montaa Pelada tuff ring. Abundant moulds of allochthonous trees with trunk diameters of 50– 100 cm are common (e.g. Montaa Magua). A poorly exposed phonolite pumice fall (<20-cm thick; Fig. 4b) overlying the Member 6 pellet fall layer in the Gimar valley is attributed to this phase. Interpretation: Phase 5 marked a dramatic increase in the mass-flux of the pyroclastic density current, causing widespread erosion and depositing ignimbrite more extensively across the volcano’s lower flanks (Fig. 7) than in previous phases. The occurrences of locally eroded lithologies and intraclasts, and the abundant tree moulds, which were stripped from the volcanoes upper flanks, illustrate an increase in the erosive nature of the pyroclastic density current(s). The abundant basal and internal scours indicate that the currents were unsteady and intermittently erosive even 15 km from the caldera. These and the upward increasing lithic contents in the ignimbrite are characteristic effects of waxing flow conditions (cf. Allen et al. 1999; Branney and Kokelaar 2002). The marked increase in geographic distribution of the ignimbrite may reflect breaching of increasingly wide segments of the caldera wall by the pyroclastic current(s). This is consistent with an increase in the eruptive mass flux such as accompanies vent widening or a change from a single vent to a multiple-vent/fissure or ring-vent source (e.g., Bacon 1983). Whereas coarser-grained parts of the previous density stratified pyroclastic density currents (e.g. during Phases 2 and 4) had been largely confined to paleovalleys, coarse-grained parts during Phase 5 were more extensive and deposited several meters of ignimbrite widely across interfluves and ridges. This was in part

because the valleys had been already partially filled by the deposits of the earlier phases of the Poris eruption. However, lateral transitions from diffuse-bedded and stratified lapilli-tuff into massive lapilli-tuff from overbank into valley-fill still occur and show that topography continued to exert some control on the current’s depositional behavior. The abundance of thin, laterally impersistent diffuse bedding and stratification in the ignimbrite on paleo-topographic highs indicates relatively low aggradation rates and deposition from granular-flow dominated flow-boundary zones (sensu Branney and Kokelaar 2002) in which current unsteadiness (e.g. current velocity, fluctuations and impingement of turbulent eddies) is increasingly registered in the lower flowboundary zone of the current and thus in the deposit (see Branney and Kokelaar 2002 p 71). The greater thickness of the correlative ignimbrite in paleovalleys suggests that the valley filling lapilli-tuff aggraded more rapidly than the thin veneers. We ascribe the dominantly massive nature of the lithofacies in the paleovalleys as indicating that deposition at these locations occurred where both the lowermost levels of the pyroclastic current and uppermost levels of the forming compacting deposit were characterized by similar, high concentrations of particles, supported predominantly by interstitial fluid escaping through the depositing dispersion (‘hindered settling’). This is a ‘fluid escape-dominated flow-boundary zone’ of Branney and Kokelaar (2002 p 39). The boundary between the current and forming deposit does not have a marked rheological interface, and turbulence and tractional sorting tend to be dampened near it by the high concentrations of particles there. Minor flow fluctuations may be dampened there too, and the resultant deposit lacks stratification (Kneller and Branney 1995, Vrolijk and Southard 1998). Phase 6: caldera collapse and widespread emplacement of lithic breccia Phase 6 records the caldera collapse phase of the eruption during which coarse lithic breccias (part of Member 8) were emplaced across a large sector of the Bandas del Sur, and tephriphonolite magma was erupted for the first time. Lithic breccias of Member 8 (lithofacies mlBr to dblBr) Member 8 comprises predominantly massive to diffusebedded, coarse lithic breccias (lithofacies mlBr to) that pass vertically and laterally into lithic-rich pumiceous lapilli-tuff (lithofacies lmLT; Fig. 11a). Member 8 is even more widespread than Member 7, and locally overlaps it. The breccias are moderately to very poorly sorted, matrixsupported to framework-supported, and vary between ~30-cm thick topography-draping veneers to 15-m thick valley-fills. They commonly take the form of layers to lenses that persist laterally for over tens to hundreds of meters and thicken up to 6 m into depressions. There are

410

several such layers within Member 8 at some locations. Some exhibit normal grading (Logs 4 and 8 on Fig. 6a), and the grainsize populations vary across short distances (e.g. at Tajao). Imbrication is developed locally (e.g. La Mareta) and moulds of large, allochthonous tree trunks are common (e.g. at Magua; Fig. 1). The lower contact of Member 8 is laterally variable. At several localities, the content and clast-size of lithic clasts increases with height in upper parts of the underlying member, and the base of Member 8 is arbitrarily defined as the level where they first abundantly attain the size of blocks. Discontinuous lenses of fines-poor lithic-rich ignimbrite (lithofacies lmLTfines-poor) occur at the base of Member 8 in places (e.g. Magua, Fig. 1), and at other places meter-scale scouring of the underlying units has occurred (e.g. at Tajao, Fig. 1). Elsewhere, the base exhibits spectacular load structures (see below). Member 8 passes gradationally up into Member 9 (see Fig. 6). The accidental lithic blocks are angular to rounded, 2 m in diameter, and include basalt, phonolite and trachyte lava (75–85 vol%), lesser amounts of syenite and micro-syenite (<10 vol%), welded lapilli-tuff (<2 vol%), and indistinguishable hydrothermally altered lithologies (<23 vol%). Conspicuous locally derived lithic populations occur around La Mareta (alkali basalt clasts; see Bryan et al. 1998b), Montaa Pelada (palagonitised lapilli-tuff) and Montaa Fasnia (basalt scoria). Many blocks have fresh, partly loose curviplanar flakes surrounding a central sub-round faceted core. Similar sharpedged flakes are also abundant within the lapilli tuff matrix of the breccia. Some lithic blocks are made of jigsaw-fit fragments supported in tuff matrix, and some blocks are cracked, with the cracks infilled with pumiceous lapilli-tuff. Member 8 contains abundant rounded, pale green juvenile phonolite pumice lapilli (2–20 vol%). These are absent from some of the coarser, openwork lithic breccias. In addition, Member 8 marked the first appearance of a second type of juvenile material; approximately 50– 100 cm above the base of Member 8, irregular-shaped blocks of dense banded tephriphonolite (Fig. 5) pumice and obsidian (<40-cm diameter; ~ 1.5 vol%) occur. These blocks are variably vesicular, with pods, streaks and bands, 1-mm thick, of black and green glass. Such banded material is absent from lower members of the Poris Formation (for geochemistry see Edgar et al. 2002). Interpretation: the change from deposition of the widespread lithic-poor ignimbrite of Phase 5 to widespread emplacement of lithic breccias during Phase 6 marks a major change in eruptive style. However, the gradational nature of the upper and lower contacts of Member 8 indicates that changes were gradational and that members 7, 8 and 9 all were deposited from the same sustained pyroclastic density current. The ubiquitous influx of large lithic blocks records a peak in the competence of this current and indicates near-source entrainment of blocks as a result of conduit erosion, nearsource erosion, and/or rock avalanching from growing and decrepitating caldera fault scarps (Fig. 7; see Branney

and Kokelaar 1994). The widespread occurrences of moulds of transported trees together with the upward increase in the abundance and size of lithic clasts suggest widespread deforestation and erosion of the upper flanks (in-situ fossils at the base of the formation indicate that at the start of the eruption lower flanks were covered with sparse shrubs, not forest). Some lithologies (syenite, welded tuff, hydrothermally altered lithologies, and some varieties of basalt and phonolite) were clearly erupted from the vent, while others (alkali basalt, phonolite, scoria, palagonite) were derived locally by substrate erosion by the pyroclastic density current on the lower volcano flanks. The latter, together with the presence of abundant scour surfaces at the base of, and within, the lithic breccias indicate erosion also occurred on lower flanks during the start of Phase 6. This is consistent with waxing flow conditions (Branney and Kokelaar 2002). Multiple normal-graded breccia layers record successive pulses of breccia deposition, so the pyroclastic current must have been unsteady (fluctuating). Normal grading of lithic clasts above loaded horizons may reflect decreases in the maximum size of clast being supplied by the pyroclastic current, i.e. due to waning flow competence or waning clast entrainment near source. Alternatively they may form by the sinking of lithic clasts through loose deposit, in which larger clasts tend to sink to deeper levels than smaller ones (Branney and Kokelaar 2002 fig. 5–7). The subrounded, faceted shape of the accidental lithic blocks with their associated splintery rock flakes indicate mechanical spalling due to thermal stresses during and/or, in the case of the in-situ brecciated relations immediately after, hot emplacement. The abrupt appearance of dense banded tephriphonolite pumice and blocks of obsidian in Member 8, alongside the ubiquitous phonolite pumice, defines a geochemical zonation within the Poris Formation (an ‘entrachron’ of Branney and Kokelaar 2002) and reflects a significant change in the volcano’s plumbing system. Significantly, this new appearance occurs just above the first appearance of the extensive breccias, and it is likely that they are related. By analogy with widespread coarse lithic breccias within ignimbrites inferred to relate to caldera collapse elsewhere (e.g. see Druitt and Bacon 1986; Hildreth and Mahood 1986; Scott et al. 1996; Allen and Cas 1998b; Bryan et al. 1998a) we infer that Phase 6 was the climactic caldera collapse phase of the Poris eruption, and that caldera faulting and conduit modification caused an increase in eruptive mass flux, with major entrainment of lithic blocks from conduit walls and forming caldera fault scarps. The increase in eruptive mass flux probably increased the depth of draw-down into the zoned magma chamber (e.g. Blake and Ivey 1986) so that resident tephriphonolite magma in a zoned chamber mingled and erupted for the first time. In ignimbrites elsewhere, first appearances of more mafic juvenile magma at levels close to the appearance of coarse lithic breccias have been interpreted in a similar way (CarrascoNfflez and Branney 2003).

411

Load structures in the lithic breccias Spectacular load-and-flame structures and podiform breccia bodies are common both at the base of and within Member 8 (Fig. 11a–c; logs 2, 7, 13, 14 of Fig. 6). In places (e.g. La Mareta; Fig. 1) they are asymmetric indicating downslope shear of unconsolidated material (Fig. 11a). The load structures penetrate up to 2 meters into the underlying lapilli-tuff and vary from isolated bulbous convex-down structures to more complex chaotic structures, pods and lenses (Fig. 11a). Finer-grained, pods and domains of lithic lapilli occur within and around loaded lithic breccia (e.g. Tajao). Most blocks in the load structures do not show a preferred long-axis orientation, but clasts along the margins of many load structures, and in the apices of flame structures show preferred subvertical a-axis or b-axis alignments. Isolated blocks, up to 1.5 m in diameter, in Member 8 have sunk individually through underlying unconsolidated lapilli-tuff (i.e. Member 7). These occur below both loaded and non-loaded contacts, and they commonly show diffuse vertical trails of Member 8 ignimbrite or lithic lapilli (e.g. Fig. 11c). Small (<20 cm long) elutriation pipes occur in Member 8 ignimbrite (mLTpip) above loaded lithic breccia horizons at one locality (log 7 on Fig. 6a). Interpretation: the load structures indicate that emplacement of Member 8 occurred while the underlying parts of the ignimbrite (Member 7) were still completely loose, uncompacted and gas rich. The downslope-sheared nature of many of them suggests that the loading took place before downslope movement had ceased. The discrete pods of lithic breccia may have formed in a number of ways: (1) by loading of discrete depositional clusters, lenses or dunes of blocks (such as formed at Mount Mazama, Druitt and Bacon 1986); (2) by vigorous segregation and elutriation during rapid sedimentationinduced fluidization (Druitt and Sparks 1982; Branney and Kokelaar 2002), or (3) by post-depositional segregation of an originally coherent layer of breccia into isolated load balls via initial development of convective instabilities at its lower contact (Allen 1984; see Fig 5.7 of Branney and Kokelaar 2002). The elutriation pipes above the breccias indicate that loading and compaction caused a sufficiently vigorous upward gas flow to cause gas channeling and localized elutriation of the overlying uncompacted ignimbrite, probably enhanced by the thermal expansion of interstitial air. Clasts near apices of flame structures were rotated and orientated by a combination of the frictional drag exerted by the sinking block and by related fluid-escape (e.g. see Postma 1983). The lithic lapilli trails above isolated blocks formed by sedimentation fluidization of ignimbrite during the sinking of the block through the underlying loose and uncompacted ignimbrite. Rapid lateral grainsize variations and the occurrence of discrete finer-grained domains of lapilli may reflect the different loading behavior of different grainsize populations (i.e. unsteady deposition), or segregation caused by loading of an initially heterogeneous grain size popula-

tion, due to hindered settling and sedimentation-fluidization processes (see Branney and Kokelaar 2002). Phase 7: waning current(s) and close of the explosive eruption The final stages of the explosive eruption are recorded by an ignimbrite (Member 9). Member 9: ignimbrite lithofacies (mLT, dbLT) and ash layer with accretionary lapilli (mTacc) Member 9 widely rests upon Member 8 and comprises two beds: a lower, widely distributed layer of ignimbrite and an upper, thin, geographically restricted accretionary lapilli-bearing ash layer. It passes gradationally down into the lithic breccias of Member 8 and its top is marked by a paleosol. At Gimar, Member 9 rests disconformably upon Member 6 (Fig. 11d). It comprises massive, diffusebedded and pumice-rich lapilli-tuff (lithofacies mLT; dbLT; pLT). At many localities, this lower bed typically comprises a thin (20–40 cm), massive layer with abundant inverse-graded framework supported pumices and normal-graded lithic lapilli. However, in some paleovalleys it reaches 5-m thick (e.g. at La Mareta, Gimar, Magua; see Logs 15–17 on Fig. 6b), and its base is marked by discontinuous scours and lithic-rich lenses. In the Gimar valley, it directly overlies the ashfall layer of Member 6, and contains abundant moulds of allochthonous trees (Fig. 10; Log 17 on Fig. 6b). Well-rounded lapilli of both phonolite pumice and banded tephriphonolite pumice occur. Several thin, discontinuous fine-grained tuff horizons with sparse accretionary lapilli occur at some localities (e.g. La Mareta; Fig. 1). At one locality (El Arrecife, Fig. 1) this ignimbrite passes upwards into 16-cm thick layer of fine tuff (mTacc) that contains scattered accretionary lapilli, 8 mm in diameter, with ash cores and several concentric laminae, totaling 1–2 mm in thickness. Interpretation: Phase 7 was a waning phase that followed the climax of the Poris eruption, during which the lithic supply to the pyroclastic density current diminished (Fig. 7). Waning flow is recorded by the normal grading of lithic clasts and inverse grading of pumices (for mechanism see Branney and Kokelaar 2002 p 43–45). The current was still density stratified and deposited coarser, thick massive ignimbrite in remaining paleovalleys, and an extensive, laterally variable pumiceous ignimbrite veneer elsewhere, similar to those of earlier ignimbrite phases. For the first time, the Poris eruption funneled pyroclastic density currents down the southern side of the Gimar valley. This suggests that changes in near-source (e.g. caldera rim) topography, or in vent location, accompanied caldera collapse. The abundant moulds of allochthonous trees at Gimar indicate that the change of flow direction caused significant deforestation of Las Caadas slopes above Gimar.

412

The final parts of the pyroclastic density current were ashrich. Accretionary lapilli fell into the current probably from a buoyant co-ignimbrite plume (see below). Member 9 in the Gimar valley has previously been interpreted as the deposits of an earlier phase of the Poris eruption (the ‘Abona ignimbrite’ of Egdar et al. 2002) and a correlative to Member 7. However, whereas Member 7 contains phonolite pumice, Member 9 also contains banded pumice lapilli, and so must post-date the entrachron (Fig. 6; sensu Branney and Kokelaar 2002) that records the initial eruption of abundant banded pumice during Phase 6 (Member 8). This shows that this deposit post-dates Member 8, and led to the recognition of the disconformity at its base, where it rests directly on Member 6 (Fig. 11d). Phase 8: post-eruption reworking, clastic dyke emplacement and pedogenesis Post-eruption reworking of the Poris tephra in the Bandas del Sur is recorded by impersistent layers of volcaniclastic sediments.

the dykes, and discontinuous lenses and pods of pebbles occur near the center of some dykes. Interpretation: the clastic dykes record downslope extensional deformation of the semi-lithified Poris ignimbrite shortly after the Poris eruption. Unconsolidated, wet sediment derived from Member 10 infilled the opening cracks in the compacted ignimbrite. The orientation of laminae in the dykes suggests rapid streaming of sediment through the opening fissure, possibly fluidized as pore water was sucked in to the dilating fissure. The extensional deformation may have been triggered by aftershocks and fault readjustments following caldera collapse. Similar dykes cut through ignimbrite of the La Caleta Formation around Tajao (Fig. 1), and the phenomenon may be common on Tenerife. Soft-state deformation of ignimbrites elsewhere similarly may relate to seismic shock during caldera collapse (e.g. Matahina ignimbrite in New Zealand, Bailey and Carr 1994; Scafell caldera in England, Branney and Kokelaar 1994; Rotorua caldera in New Zealand, Milner et al. 2002; Taal caldera, Philippines, MJB pers. observation). Distinguishing silicic phreatomagmatic eruptions from fine silicic ash deposits

Sediments of Member 10 Phreatomagmatic characteristics Member 10 comprises volcaniclastic sandstones, pebbly sandstones and conglomerates that sharply overlie and overstep Members 8 and 9 with an irregular erosion surface (e.g., Magua; Fig. 1). Fine to medium-grained vitric sandstones show ripple cross-lamination with thin silt drapes, and are intercalated with stratified pebbly sandstones, and thin lenticular pebble lags. Clast types are similar to those found in underlying members. Member 10 locally passes up into a paleosol that marks the top of the formation. Interpretation: erosion and reworking of the Poris Formation by ephemeral streams immediately followed the Poris eruption. This was presumably enhanced by the absence of vegetation and the availability of abundant unconsolidated new Poris tephra. The general paucity of post-eruption sediments is inferred to reflect widespread downslope transport of sediment into the sea. The landscape stabilized during the subsequent repose period as indicated by the soil at the top of the Formation. Clastic dykes Several bifurcating dykes, 50 cm wide and ~ 20-m long, cut as much as 9 m down through members 8 and 9 of the Poris ignimbrite (e.g. at Magua, Fig. 1). They trend NE, parallel to the strike of the paleoslope, and their margins are straight to irregular. They are filled with pumiceous sand and silt derived from the overlying Member 10. Most of the fill is massive, but locally the dyke margins show a steep lamination parallel to side of

The pellet- and accretionary lapilli-bearing layers of finegrained (generally <100 nm) ash in the Poris Formation resemble the products of phreatomagmatic explosivity. Members 1–6 have been interpreted as recording rapid fluctuations between magmatic and phreatomagmatic explosivity at the vent (Bryan et al. 1998b) and as the deposits of discrete ‘pyroclastic surges’ generated by phreatomagmatic explosivity at the vent (Edgar et al. 2002). We interpret the field relations of these layers to indicate that some of the fine ash layers are ignimbrite veneers and others are co-ignimbrite ashfall layers, as follows. Beds 2A, 3A and 4A are thin, fine-grained, ignimbrite veneers on ridges and they grade laterally into thick valley fills of massive lapilli-tuff ignimbrite, locally via layers, up to 1-m thick, of fine-grained tuff plastered against steep valley sides (e.g. Member 7 at Magua). This suggests that coarser lapilli were preferentially channeled along valley floors within the denser lower parts of density stratified currents, while upper parts of the current were rich in fine ash, more laterally extensive, and draped fine ash layers higher up valley sides and across the topographic highs. Similar veneer deposits occur in ignimbrites from Laacher See volcano, Germany (Schumacher and Schmincke 1990) and in the Oruanui ignimbrite (Wilson 2001). In contrast, beds 2B, 3B and 4B overlie ignimbrites, lack tractional stratification, drape irregular topography, commonly comprise frameworksupported pellets, are extremely widespread, and are interpreted as co-ignimbrite ashfall deposits that settled during pauses between successive pyroclastic density currents traveling across the Bandas del Sur (e.g. during

413

Phase 2). We ascribe the absence of fine ashfall layers within flow-unit 4 to a lack of significant pauses in flow during which pellets could accumulate. A general absence of accretionary lapilli from within the climactic parts of flow unit 4 (Members 7–8), may reflect the significantly higher mass flux of the density current, so that the relative effects of atmospheric moisture were minimal at these distances from source (see below). If the Poris eruption was phreatomagmatic, large volumes of external water must have been readily available, such as in a former caldera lake or ice-filled caldera at Las Caadas. The paucity of basalt lithic clasts in the fine ash population suggests that the water did not meet the erupting magma in a basalt aquifer. SEM backscatter imaging shows that most of the shards (<100 m) are bi-cuspate and tri-cuspate bubble-wall, and pumiceous types with relatively few blocky shards (Fig. 9d). The 3-d morphology of the shard populations cannot readily be quantified as a result of lithification involving cementation and alteration, but the predominance of the aforementioned shard types suggests that the erupting magma was pumiceous foam at the time of fragmentation, presumably as a result of exsolution of magmatic volatiles. There is little direct evidence for a phreatomagmatic fragmentation mechanism, although this cannot be excluded. Bubble-wall shards in fine silicic ashes elsewhere have been inferred to record phreatomagmatic eruptions in which lake or seawater at the vent gained rapid access to, and explosively interacted with, pumiceous silicic magma that was already vesiculating, fragmenting and erupting as a result of magmatic exsolution (e.g. Self and Sparks 1978; Allen and Cas 1998a), however the mechanisms and extent of water involvement in such cases remain poorly constrained (see Houghton et al. 2000). Role of atmospheric moisture The preservation of abundant fine-grained ash in the Poris Formation (both in the ignimbrites and the pellet-bearing fine-ash layers) may reflect high levels of atmospheric moisture during the Poris eruption, rather than reflecting phreatomagmatic fragmentation. Evidence for moisture during the deposition of the fine-ash rich lithofacies is indicated by: (1) abundant ash pellets and accretionary lapilli; (2) ash-coatings on pumice and lithic lapilli (‘armored lapilli’ of Waters and Fisher 1971); (3) steep plastering of fine ash layers onto vegetation, commonly followed by partial flaking off to produce decimeter-sized intraclasts of fine ash seen at the base of vegetation moulds (e.g. in Member 3 at Tajao); and (4) trapped porespace and vesicles (e.g. within Bed 2B) formed as moist ash pellets and clusters partially disaggregated (see Rosi 1992). High moisture levels in Poris ash plumes would have enhanced ash agglomeration (formation of ash clusters, pellets and accretionary lapilli) so that the fine ash was deposited on Tenerife rather than being more widely

dispersed. Thermal lofting will hinder proximal settling of ash onto a hot, newly deposited ignimbrite sheet, unless the ash has agglomerated to form aggregates with larger settling velocities than those of the individual ash particles. This may account for the absence of coignimbrite ashfall layers on some ignimbrites. Where such layers do occur (e.g. ‘layers 3’ of Sparks et al. 1973) the proximal deposition may have been favored by moist ash agglomeration, even though delicate ash clusters may not be visibly preserved. As Tenerife is surrounded by ocean the atmosphere is commonly moist, and orographic clouds and precipitation are common on higher slopes. The thermal lofting of coignimbrite ash plumes during the Poris eruption may have induced condensation of entrained atmospheric moisture. Moist ash fallout with ash pellets and accretionary lapilli has been observed from eruptions considered to be magmatic, rather than phreatomagmatic, in conditions of high atmospheric humidity (e.g. Sakurajima volcano, Japan, Gilbert and Lane 1994; Unzen volcano, Japan, Watanabe et al. 1999; and at Soufri re Hills volcano, Montserrat, Cole et al. 2002). The Poris ignimbrite sheet generally thickens towards the coast. The moisture content of the co-ignimbrite ash plumes may have been enhanced by thermal evaporation of surface seawater where the pyroclastic density currents crossed the coast. At the present coast the ignimbrite locally overlies ephemeral stream sediments, preserves subaerial plant fossils and shows no evidence of being marine (Brown 2001). Therefore, we infer that the pyroclastic density currents must have encountered a shoreline beyond the location of the present coastline, and so the exact nature of the interaction between the currents and the ocean are not known. Passage of the density currents to the sea along a 50 km stretch of the Bandas del Sur coast is likely to have caused thermal evaporation of seawater, and increased the moisture content of the coignimbrite ash plume. Such events have been documented on a smaller scale at Montserrat (Cole et al. 2002; Mayberry et al. 2002) and ash pellets formed in the associated ash plumes (Ritchie et al. 2002). Thermal expansion of steam also may have enhanced the elutriation of fine ash, as proposed by Sigurdsson and Carey (1989). The abundant fine ash layers and lithofacies in the Poris deposits, therefore, may be the result of deposition enhanced by ash agglomeration in ash plumes where atmospheric moisture was supplemented by steam evaporated from seawater as the pyroclastic density currents crossed the coast. The Poris Formation appears to be richer in fine ash, pellets and accretionary lapilli than most other Bandas del Sur ignimbrites. This may be because the earlier Poris pyroclastic density currents were smaller and separated by pauses during which the fine ashfall layers were deposited. Thin fine-ashfall layers, some with ash pellets, do occur in several other Bandas del Sur ignimbrite sheets (e.g. in the Abades, Fasnia and La Caleta Formations; see Brown et al. 2003) and these may be co-ignimbrite ashfall deposits formed in the same way as those in the Poris

414

ignimbrite. Absences of such layers (e.g., within the Granadilla ignimbrite, and in Members 7–9 of the Poris Formation) may indicate that the pyroclastic density currents were widespread and sustained, so that there were no pauses during which ash could fallout upon the Bandas del Sur. In such cases, the fine ash fell back into the sustained pyroclastic density current and/or was dispersed far across the Atlantic Ocean. Where there is abundant atmospheric or standing water (e.g. at a flooded caldera or volcanic island) the abundance of fines in a pyroclastic deposit in some cases may relate more to within-plume processes (e.g. moisture-enhanced agglomeration of fine ash) than it does to source processes, such as the degree of explosive fragmentation. The proposed relationship between the abundance of fine ash layers and the impersistence of the pyroclastic density currents is less likely to apply at volcanoes where there is little atmospheric moisture or standing water because, in the absence of ash agglomeration, thermal lofting may hinder the deposition of fine grained coignimbrite ash on newly-formed hot ignimbrite sheets, for example in arid climates. We conclude that grainsize characteristics, ash pellets and accretionary lapilli, are at best equivocal evidence for phreatomagmatic explosivity, and in some cases may be attributed to meteorological moisture. It is likely that all gradations exist between (1) true phreatomagmatic eruptions in which explosive fragmentation results directly from interaction between water and the erupting magma, (2) water-enhanced explosivity, in which water–magma interactions modify magma that is already fragmented by volatile exsolution, and (3) ‘dry’ magmatic explosivity followed by fine-ash deposition where atmospheric moisture, in some cases enhanced by the evaporation of standing water crossed by pyroclastic density currents, favors agglomeration and deposition of fine ash. At present it is difficult to distinguish between these possibilities using deposit characteristics, although the true phreatomagmatic end-member may give rise to less vesicular shard populations (e.g. Wohletz 1983). Accretionary lapilli in the Poris ignimbrites In the Poris Formation, the ash pellets occur within the ashfall layers (Beds 2B, 3B, 4B and 6B), whereas the accretionary lapilli occur within the ignimbrites. Pellets in the ashfall layers lack fine-grained laminations and show evidence of partial amalgamation with adjacent pellets. This indicates that they were moist, malleable aggregates at the time of deposition. They probably formed by hydrostatic and electrostatic agglomeration within moist, low-concentration co-ignimbrite ash plumes (see Schumacher and Schmincke 1991; Gilbert and Lane 1994; ch. 16 of Sparks et al. 1997) that developed from the lowconcentration uppermost parts of the pyroclastic density currents, and we infer that they were deposited shortly after passage of the pyroclastic density current, and/or within its wake, and they have locally impacted, or

loaded, into the tops of the ignimbrites that were still loose and unconsolidated (Fig. 8). We propose that the accretionary lapilli started to form in the same way, but that in this case the pellets dropped into upper turbulent parts of the pyroclastic density currents where they accreted their fine-grained concentric laminations. They were ultimately deposited through the currents’ lower flow boundary, sometime after the deposition of ignimbrite had started. In Member 6, the widespread breakage of accretionary lapilli in the pyroclastic current is shown by abundant accretionary lapilli fragments in the ignimbrite. This indicates that the accretionary lapilli, in contrast to the pellets, lithified rapidly prior to deposition, perhaps due to elevated temperatures within the lower levels of the pyroclastic current. Early salt or zeolite cementation may have played a role (e.g. Gennarro et al. 2000). The intact and fragmented accretionary lapilli ultimately settled through the lower flow-boundary zone of the pyroclastic density current, to be deposited along with lapilli and ash to form the various accretionary lapilli-bearing lithofacies (e.g. mLTacc, dbLTacc). Large accretionary lapilli reported in ignimbrites elsewhere (e.g. the Cana Creek Tuff of Australia, McPhie 1986; the Ito ignimbrite of Japan, Ui et al. 1992) may have formed in a similar way. Erupted volume The Poris ignimbrite was emplaced over a highly irregular topography, and it has been subjected to significant posteruption erosion. A DRE volume of 3–4 km3 has been estimated for the Poris Formation (Edgar et al. 2002). However, its original onshore limits in the southwest, northeast and northwest of the volcano are poorly constrained, and there is abundant evidence that the flanks of Las Caadas volcano acted as chutes that funneled most of the eruptive products directly into the ocean, bypassing the subaerial slopes (Brown 2001). The greatest volume from the Poris eruption is likely to have been transported into the ocean by pyroclastic density currents (including offshore dispersal of co-ignimbrite ash) because the onshore topography lacks a basin that could act as a major depocentre for ignimbrite accumulation. Rather, the onshore Poris ignimbrite sheet has the form of a sourceward-thinning veneer that represents just a thin proximal feather edge of what is probably a much larger deposit. The volume of this feather edge is a reflection more of depositional control and available accommodation space afforded by the local and regional slopes than it is a measure of the eruption volume. A >700 m thickness of volcaniclastic material detected by seismic profiling off the southern coast of Tenerife (Oceanographic Institute Madrid data reported in Bryan et al. 1998b) is three times the thickness of the equivalent subaerial deposits (Bandas del Sur Formation) and probably includes material from the Poris eruption. Bathymetric data published by Watts and Masson (2001) also indicate the existence of a submarine apron

415

extending 10–20 km from the southern coast. We thus speculate that the eruption volume may have been an order of magnitude larger than that represented by the onshore deposits. Eruption Narrative The Poris Formation records an explosive, calderaforming phonolitic eruption at Las Caadas caldera at 273 ka. The eruption was multiphase, compositionally zoned, and began with a Plinian eruption column, which increased in intensity with time and deposited tephra over much of southeastern Tenerife (Phase 1). At least three separate pyroclastic density currents were then emplaced across a 10–15 km strip of the Bandas del Sur, and coignimbrite ash clouds deposited ash pellets over much of the central and southeast parts of the island (Phase 2). A Plinian eruption column is not thought to have existed during this time. A subsequent, brief Plinian phase (Phase 3) marked the start of a larger, more sustained pyroclastic density current, that gradually waxed so that the mass flux and dispersal increased with time (Phases 4 and 5), climaxing with subsidence at Las Caadas caldera and the widespread emplacement of lithic breccias (Phase 6). Tephriphonolite magma erupted for the first time during this phase. The paroxysmal pyroclastic density current caused widespread erosion, and deforestation of upper flanks of the volcano. Caldera subsidence-induced changes in vent and/or caldera wall morphology started to funnel the waning pyroclastic density current into the Gimar valley (Phase 7). The eruption was followed by rapid erosion, sedimentary reworking, formation and breaching of temporary ignimbrite-dammed lakes, delivery volcaniclastic sediment into the ocean, and minor downslope sliding of semi-lithified ignimbrite (Phase 8). Several lithofacies have the appearance of phreatomagmatic deposits, but no conclusive evidence for phreatomagmatic explosivity has been found, and the abundant pellets and accretionary lapilli are inferred to be the result of the eruption columns and/or fine-grained ‘phoenix’, or co-ignimbrite, ash plumes interacting with atmospheric moisture. The moisture in the ash plumes may have been enhanced by heat-induced evaporation of seawater where hot pyroclastic density currents crossed the coast into the ocean. The volume of the eruption was considerably larger than that of the preserved ignimbrite sheet, because large volumes were transported into the ocean: we infer eruption volumes of the same order-ofmagnitude as the other large eruptions from Las Caadas (i.e. tens of km3). Acknowledgements RJB was funded by a NERC Studentship (GT 4/144/97) at the University of Leicester. P. Kokelaar, J. Zalasiewicz, S. Self provided many useful comments on the thesis chapter that this is taken from. Critical reviews by J. Stix and E. Calder were greatly appreciated.

References Ablay G, Hrliman M (2000) Evolution of the north flank of Tenerife by recurrent giant landslides. J Volcanol Geotherm Res 103: 135–159 Allen JRL (1984) Sedimentary structures: their character and physical basis. Developments in Sediment 30B: 343–393 Allen SR, Cas RAF (1998a) Rhyolitic fallout and pyroclastic density current deposits from a phreatoplinian eruption in the Aegean Sea, Greece. J Volcanol Geotherm Res 86: 219–251 Allen SR, Cas RAF (1998b) Lateral variations within coarse coignimbrite lithic breccias of the Kos Plateau Tuff, Greece. Bull Volcanol 59: 356–377 Allen SR, Stadlbauer E, Keller J (1999) Stratigraphy of the Kos Plateau Tuff: product of a major Quaternary explosive eruption in the eastern Aegean, Greece. Int J Earth Sci 88: 132–56 Ancochea E, Fuster JM, Ibarrola E, Cendrero A, Coello J, Hernan F, Cantagrel JM, Jamond C (1990) Volcanic evolution of the island of Tenerife (Canary Islands) in the light of new K-Ar data. J Volcanol Geotherm Res 44: 231–249 Bacon CR (1983) Eruptive history of Mount Mazama and Crater Lake caldera, Cascade Range, USA. J Volcanol Geotherm Res 18: 57–115 Bailey RA, Carr RG (1994) Physical geology and eruptive history of the Matahini Ignimbrite, Taupo Volcanic Zone, North Island, New Zealand. NZ J Geol Geophys 37: 319–344 Blake S, Ivey GN (1986) Magma mixing and the dynamics of withdrawal from stratified reservoirs. J Volcanol Geotherm Res 30: 200–229 Bogaard P, Schmincke H-U (1984) The eruptive centre of the late Quaternary Laacher See Tephra. Geologische Rundsch 73: 935–982 Branney MJ, Kokelaar BP (1994) Volcanotectonic faulting, softstate deformation and rheomorphism of tuffs during development of a piecemeal caldera, English Lake District. Bull Geol Soc Amer 109: 507–530 Branney MJ, Kokelaar P (2002) Pyroclastic density currents and the sedimentation of ignimbrites: Geol Soc London, Memoirs 27, 152 pp Brown RJ (2001) The eruption histories and depositional mechanisms of the Poris ignimbrite of Tenerife and the Glaramara tuff of the English Lake District. PhD Thesis, University of Leicester, England, 171 pp (unpubl) Brown RJ, Barry TL, Branney MJ, Pringle MS, Bryan SE (2003) The Quaternary pyroclastic succession of southern Tenerife, Canary Islands: explosive eruptions, related subsidence and sector collapse. Geol Mag 140: 265–288 Bryan SE, Cas RAF, Mart J (1998a) Lithic breccias in intermediate volume phonolitic ignimbrites, Tenerife (Canary Islands): constraints on pyroclastic flow depositional processes. J Volcanol Geotherm Res 81: 269–296 Bryan SE, Mart J, Cas RAF (1998b) Stratigraphy of the Bandas del Sur Formation: an extracaldera record of Quaternary phonolitic explosive volcanism from the Las Caadas edifice, Tenerife (Canary Islands). Geol Mag 135: 605–636 Bryan SE, Cas RAF, Mart J (2000) The 0.57 Ma Plinian eruption of the Granadilla Member, Tenerife (Canary Islands): an example of complexity in eruption dynamics and evolution. J Volcanol Geotherm Res 103: 209–238 Bryan SE, Mart J, Leosson M (2002) Petrology and geochemistry of the Bandas del Sur Formation, Las Caadas Edifice, Tenerife (Canary Islands). J Pet 43: 1815–1856 Carrasco-Nunez G, Branney MJ (2003) Progressive assembly of a massive ignimbrite with normal to reverse compositional zonation: the Zaragoza ignimbrite of central Mexico. Bull Volc (in press) Cole PD, Calder ES, Sparks RSJ, Druitt TH, Young SR, Herd R, Harford C, Norton G, Robertson R (2002) Pyroclastic flow deposits formed during the 1996–99 eruption of Soufriere Hills Volcano, Montserrat, W. I. In: Druitt TH, Kokelaar BP (eds) The eruption of Soufriere Hills Volcano, Montserrat, from 1995 to 1999. Mem Geol Soc London 21, pp 231–262

416 Druitt TH, Bacon CR (1986) Lithic breccia and ignimbrite erupted during the collapse of Crater Lake Caldera, Oregon. J Volcanol Geotherm Res 25: 1-32 Druitt TH, Sparks RSJ (1982) A proximal ignimbrite breccia facies on Santorini. J Volcanol Geotherm Res 13: 147–171 Edgar CJ, Wolff JA, Nichols HJ, Cas RAF, Mart J (2002) A complex quaternary ignimbrite-forming phonolite eruption: the Poris Member of the Diego Hernandez Formation (Tenerife, Canary Islands). J Volcanol Geotherm Res 118: 99–130 Fierstein J, Hildreth W (1992) The Plinian eruptions of 1912 at Novarupta, Katmai National Park, Alaska. Bulletin of Volcanology 54(8): 646–684 Gennaro M, Cappelletti P, Langella A, Perrotta A, Scarpati C (2000) Genesis of zeolites in the Neapolitan Yellow Tuff: geological, volcanological and mineralogical evidence. Contrib Min Pet 139: 17–35 Gilbert JS, Lane SJ (1994) The origin of accretionary lapilli. Bull Volcanol 56: 398–411 Hildreth W, Mahood GA (1986) Ring-fracture eruption of the Bishop Tuff. Geological Society of America Bulletin 97: 396403 Houghton B, Wilson CJN, Smith RT, Gilbert JS (2000) Phreatoplininan eruptions. In: Sigurdsson H, Houghton B, McNutt SR, Rymer H, Stix J (eds) Encyclopedia of volcanoes. Academic press, New York, pp 513–525 Inman D (1952) Measures of describing the size distribution of sediments. J Sed Pet 22: 125–145 Kneller B, Branney MJ (1995) Sustained high-density turbidity currents and the deposition of thick massive sands. Sedimentol 42: 607–617 Le Bas MJ, Le Maitre RW, Streckeisen A, Zanettin B (1986) A chemical classification of volcanic rocks based on the Total Alkali Silica diagram. J Petrol 27: 82–133 Mart J, Mitjavila J, Araa V (1994) Stratigraphy, structure and geochronology of the Las Caadas caldera (Tenerife, Canary Islands). Geol Mag 131: 715–727 Mayberry GC, Rose WI, Bluth GJS (2002) Dynamics of volcanic and meteorological clouds produced on 26 December (Boxing Day) 1997 at Soufri re Hills Volcano, Montserrat. In: Druitt TH, Kokelaar BP (eds) The eruption of Soufri re Hills Volcano, Montserrat, from 1995 to 1999. Mem Geol Soc London 21, pp 539–555 McPhie J (1986) Primary and redeposited facies from a largemagnitude rhyolitic, phreatomagmatic eruption: Cana Creek Tuff, Late Carboniferous, Australia. J Volcanol Geotherm Res 28: 319–350 Milner DM, Cole JW, Wood CP (2002) Asymmetric, multipleblock subsidence at Rotorua caldera, Taupo Volcanic Zone, New Zealand. Bull Volcanol 64: 134–149 Postma G (1983) Water escape structures in the context of a depositional model of a mass flow dominated conglomeratic fan-delta (Abrioja Formation, Pliocene, Almeria Basin, SE Spain). Sedimentol 30: 91–103 Ritchie LJ, Cole PD, Sparks RSJ (2002) Sedimentology of deposits from the pyroclastic density current of 26 December 1997 at Soufri re Hills Volcano, Montserrat. In: Druitt TH, Kokelaar BP (eds) The eruption of Soufri re Hills Volcano, Montserrat, from 1995 to 1999. Mem Geol Soc London 21, pp 435–456 Rodehorst U, Schmincke H-U, Sumita M (1998) Geochemistry and petrology of Pleistocene ash layers erupted at Las Caadas edifice (Tenerife). Proc Ocean Drill Progr, Sci Results 157: 315–328 Rosi M (1992) A model for the formation of vesiculated tuff by the coalescence of accretionary lapilli. Bull Volcanol 54: 429–434 Ross CS, Smith RL (1961) Ash-flow tuffs, their origin geological relations and identification: Bull US Geol Surv 37: 1-81 Schumacher R, Schmincke H-U (1990) The lateral facies of ignimbrites at Laacher See volcano. Bull Volcanol 52: 271–285

Schumacher R, Schmincke H-U (1991) Internal structure and occurrence of accretionary lapilli—a case study at Laacher See Volcano. Bull Volcanol 53: 612–634 Scott WE, Hoblitt RP, Torres RC, Self S, Martinez ML, Nillos TJ (1996) Pyroclastic flows of the June 15, 1991, climactic eruption of Mount Pinatubo. In: Newhall CG, Punongbayan S (eds) Fire and mud: eruptions of Pinatubo, Philippines. University of Washington Press, Seattle, pp 545–570 Self S, Sparks RSJ (1978) Characteristics of widespread pyroclastic deposits formed by the interaction of silicic magma and water. Bull Volcanol 41: 196–212 Sigurdsson H, Carey S (1989) Plinian and co-ignimbrite tephra fall from the 1815 eruption of Tambora volcano. Bull Volcanol 51: 243–70 Smith RT, Houghton BF (1995) Vent migration and changing eruptive style during the 1800a Taupo eruption: new evidence from the Hatepe and Rotongaio phreatoplinian ashes. Bull Volcanol 57: 432–439 Sohn YK, Chough S (1989) Depositional processes of the Suwolbong tuff ring, Cheju Island (Korea). Sedimentol 36: 837–855 Sparks RSJ (1976) Grain size variations in ignimbrites and implications for the transport of pyroclastic flows. Sedimentol 23: 147–188 Sparks RSJ, Self S, Walker GPL (1973) Products of ignimbrite eruptions. Geology 1: 115–118 Sparks RSJ, Bursik MI, Carey SN, Gilbert JS, Glaze LS, Sigurdsson H, Woods AW (1997) Volcanic plumes. Wiley, New York, 574 pp Talbot JP, Self S, Wilson CJN (1994) Dilute gravity current and rain-flushed ash deposits in the 1.8 ka Hatepe plinian deposit, Taupo, New Zealand. Bull Volcanol 56: 538–551 Ui T, Kobayashi T, Suzuki-Kamata K (1992) Calderas and pyroclastic flows in southern Kyushu. In: Kato H, Noro H (1992) 29th IGC field trip guidebook 4: volcanoes and geothermal fields of Japan, pp 245–276 Valentine GA, Wohletz KH, Kieffer SW (1992) Effects of topography on facies and compositional zonation in calderarelated ignimbrites: Geol Soc Am Bull 104: 154–165 Valentine GA, Giannetti B (1995) Single pyroclastic beds deposited by simultaneous fallout and surge processes: Roccamonfina volcano, Italy. J Volcanol Geotherm Res 64: 129–137 Vrolijk PJ, Southard JB (1998) Experiments on rapid deposition of sand from high-velocity flows. Geoscience Canada 24: 45–54 Walker G (1971) Grainsize characteristics of pyroclastic deposits. J Geol 79: 696–714 Walker GPL, Wilson CJN, Froggatt PC (1981) An ignimbrite veneer deposit—the trail marker of a pyroclastic flow. J Volcanol Geotherm Res 9: 409–421 Watanabe K, Ono K, Sakaguchi K, Takada A, Hoshizumi H (1999) Co-ignimbrite ashfall deposits of the 1991 eruptions of Fugendake, Unzen Volcano, Japan. J Volcanol Geotherm Res 89: 95– 112 Waters AC, Fisher RV (1971) Base surges and their deposits: Capelinhos and Taal volcanoes. J Geophys Res 76: 5596–5614 Watts AB, Masson DG (2001) New sonar evidence for recent catastrophic collapses of the north flank of Tenerife, Canary Islands. Bull Volcanol 63: 8–19 Wilson CJN (2001) The 26.5 ka Oruanui eruption, New Zealand: and introduction and overview. J Volcanol Geotherm Res 112: 133–174 Wolff JA, Grandy JS, Larson PB (2000) Interaction of mantlederived magma with island crust? Trace element and oxygen isotope data from the Diego Hernandez Formation, Las Caadas, Tenerife. J Volcanol Geotherm Res 103: 343–366 Wohletz KH (1983) Mechanisms of hydrovolcanic pyroclast formation: grain-size, scanning electron microscopy, and experimental studies. J Volcanol Geotherm Res 17: 31–64