Volume 127, Issue 6 e2022JB024056
Research Article
Open Access

Inception and Evolution of La Corona Lava Tube System (Lanzarote, Canary Islands, Spain)

I. Tomasi

Corresponding Author

I. Tomasi

Department of Geosciences, University of Padua, Padova, Italy

Correspondence to:

I. Tomasi,

[email protected]

Search for more papers by this author
M. Massironi

M. Massironi

Department of Geosciences, University of Padua, Padova, Italy

Search for more papers by this author
C. M. Meyzen

C. M. Meyzen

Department of Geosciences, University of Padua, Padova, Italy

Search for more papers by this author
R. Pozzobon

R. Pozzobon

Department of Geosciences, University of Padua, Padova, Italy

INAF-Osservatorio Astronomico di Padova, Padova, Italy

CISAS, University of Padova, Padova, Italy

Search for more papers by this author
F. Sauro

F. Sauro

Department of BiGeA, University of Bologna, Bologna, Italy

Search for more papers by this author
L. Penasa

L. Penasa

Department of Geosciences, University of Padua, Padova, Italy

INAF-Osservatorio Astronomico di Padova, Padova, Italy

Search for more papers by this author
T. Santagata

T. Santagata

Virtual Geographic Agency (VIGEA), Reggio Emilia, Italy

Search for more papers by this author
M. Tonello

M. Tonello

Department of Geosciences, University of Padua, Padova, Italy

Search for more papers by this author
G. D. Santana Gomez

G. D. Santana Gomez

Vulcan Vertical Espeleologiasy Barrancos, Arrecife, Spain

Search for more papers by this author
J. Martinez-Frìas

J. Martinez-Frìas

Instituto de Geosciences-IGEO (CSIC-UCM), Madrid, Spain

Search for more papers by this author
First published: 06 June 2022

This article was corrected on 7 JULY 2022. See the end of the full text for details.

Abstract

Growing interest in studying large terrestrial lava tubes is motivated in part by their analogy with their extra-terrestrial counterparts. However, on Earth, the formation of such structures is still poorly understood. Here, the lava tube system of La Corona (Lanzarote, Canary Islands, Spain) is studied to identify how pre-existing stratigraphy can govern a lava tube's evolution. Combining terrestrial laser scanner technology with field observations and geochemical analyses of the pre-existing lava enabled us to reconstruct the three-dimensional geometry of the lava tube system, the paleo-surface trough which it developed, and the volcanic series into which it carved its path. We show that a pyroclastic layer played a key role in the development of the lava tube. The layer—Derived from late Quaternary Strombolian activity—Is traceable along almost the full length of the tube path and defines the paleo-topography. The excavation process mostly happens because of the mechanical strength of the substrate, that controls the widening of the growing lava tube. Other influential parameters controlling erosion include slope variations of the paleo-surface (i.e., knickpoints), and the lava physical properties. Since weak layers such as regolith are a common feature of extra-terrestrial lava flows, the processes seen at La Corona to the may be highly relevant to the development of planetary lava tube systems.

Key Points

  • The presence of a weak pyroclastic layer in pre-existing lava flows favors the enlargement of lava tubes on Earth

  • The excavation process of lava tubes is enhanced by the thermo-mechanical erosional action of molten lavas at bedrock knickpoints

  • The large lava tube at La Corona may provide analogs for extra-terrestrial lava tube structures

Plain Language Summary

Lava tubes are a promising subject for future planetary exploration. Within this framework, improved knowledge of how these lava caves form and evolve in the post-cooling phase is crucial. The best way to achieve such insights is to focus on their terrestrial analogs on volcanic islands (i.e., Canary Islands, Hawai'i, Iceland, etc.). Here we study the large lava tube complex of La Corona (Lanzarote, Canary Islands) to constrain the different stages of its development. Its genesis depended on the presence of a weak pyroclastic layer, favoring its initial emplacement. Its subsequent development depended largely on the duration of the thermal contact and the mechanical interaction between molten lavas and pre-existing lava flows, as well as the paleo-topography emphasized by the pyroclastic layer and also by the chemical and physical properties (e.g., solidity, state of weathering) of the pre-existing lava pile. The presence of a weak layer as observed at the La Corona tube system is of interest for understanding the processes by which extra-terrestrial lava tubes develop among lava sequences where weak layers such as ash layer, rubble and complex ancient surface are likely to be commonly found.

1 Introduction

During effusive volcanic eruptions lava tubes work as thermally efficient conduits, where the minimization of heat loss allows the transport of lava flows over long distances up to several tens of kilometres. Ultimately, the length of a lava tube will be controlled mainly by the duration of the eruption, effusion rates, and thermal efficiency and stability of the lava tube system (Hon et al., 1994; Kempe et al., 2010; Keszthelyi and Self, 1998; Pasquarè et al., 2008; Peterson et al., 1994).

Two main modes of development have been reported for these structures: (a) Crusting over, when the roof is formed at the top of a lava channel by obstruction and/or levee accretion; and (b) inflation, when lava is injected below previously cooled lava sheets (Bravo, 1964; Francis, 1993; Greeley et al., 1971b; Greeley and Hyde, 1972; Hon et al., 1994; Kempe et al., 2010). At first inflation acts within thin sheets of lava, but then the pathway can evolce into elliptical lava tube systems as described by Cooper and Kauahikaua (1992) or as deep inflation by Sauro et al., 2020. Of these two modes, the most frequent is inflation, which typically occurs in flat areas (slopes <2°) (Bravo, 1964; Kempe, 2012; Pasquarè et al., 2008; Peterson et al., 1994). Inflation is associated with both pahoehoe and aa flows, and the magnitude of the inflation strongly depends on effusion rates (Calvari and Pinkerton, 1999; Jones et al., 2018; Kempe, 2019).

Lava tubes are not only found on Earth. Indeed, analogous structures such as pit chains and skylights have been identified on surfaces of other rocky bodies of the Solar System, such as Mars and the Moon (Cushing, 2012; Greeley, 1971a, Greeley, 1973; Greeley and Spudis, 1981; Haruyama et al., 20092012; Horz, 1985; Sauro et al., 2020; Titus et al., 2021). Lava tubes on Earth and other planetary bodies differ in their size due to diverse effusion rates, sometimes as an effect of the different gravity (weaker gravity resulting in higher effusion rates). This difference leads to the formation of pyroducts (a synonym for lava tubes, a term introduced initially by Coan, 1844) which are smaller in diameter on Earth (10–30 m) than on Mars (250–400 m) and the Moon (500–1,100 m) (Sauro et al., 2020). Despite this different scaling, the largest lava tubes on Earth are considered to be very useful analogs to their extra-terrestrial counterparts. In this regard, the La Corona lava tube in Lanzarote (Canary Islands) is of particular relevance, with a total length of about ∼7.6 km (∼9.7 km of total cave development) and a width reaching up to ∼28 m for some of its sections (Bravo, 1964; Carracedo et al., 2003; Montoriol-Pous and De Mier, 1969; Sauro et al., 20192020). It represents one of the largest pyroducts known on Earth, being comparable, in terms of volume, only to a few others such as the Kazumura Cave of Kilauea Volcano in Hawaiʻi (Allred and Allred, 1997; Kempe, 2012), the Undara System in Queensland (AUS) (Atkinson et al., 1975), the Víðgelmir and Bùri in Iceland (Detay et al., 2011; Water, 2006; Wood, 1974) and others in Arizona and Utah (Bunnell, 2013). The Ka Corona lava tube is one of the volcanic products of the Canary Island Seamount Province (CISP), a magmatic province that forms a disseminated hotspot track sitting atop old Atlantic Ocean seafloor (150–175 Ma, Roeser, 1982; Klitgord and Schouten, 1986; Roest et al., 1992; Figure 1), ∼1,300 km long and ∼350 km wide, parallel to the north-western African continental margin (Hansen Machin and Pérez Torrado, 2005; Hoernle and Carracedo, 2009; van den Bogaard, 2013). Most of the CISP has been generated ∼30 Ma, while the absolute motion of the African plate has been nearly stationary (ca. 8–10 mm/yr) (Carracedo et al., 2003; Gaina et al., 2013; Silver et al., 1998; van den Bogaard, 2013). This slow plate velocity makes this volcanic environment one of the best terrestrial analogs for the single-plate tectonics on Mars (Dañobeitia and Canales, 2000; Meyzen et al., 2015).

Within this context, our work aims to identify the processes that shaped the La Corona lava tube in order to understand the origin, evolution and degradation of this subterranean structure. In order to disentangle the processes, we have carried out a study of the satellite images of the northern region of Lanzarote, followed by field exploration of inner portions of the La Corona lava tube and adjacent areas. A laser scanner survey allowed us to proceed to the three-dimensional reconstruction of the main conduit portions. Sampling was carried out inside the pyroduct to characterize the geochemistry of the various lavas providing constraints on the timing of and genetic relationships between the different lava flows.

2 Geological Overview

The Canarian archipelago comprises an elongated arcuate chain of seven major volcanic islands and numerous islets and seamounts located off the northwest coast of Africa (Hoernle and Carracedo, 2009; Figure 1). The origin of the island chain is thought to be hotspot-related (Ancochea Soto et al., 1996; Carracedo, 1998; Dañobeitia and Canales, 2000; Duncan, 1984; Morgan, 1983). The hotspot track progressed toward the south-west, as shown by the irregular succession of decreasing island ages (Figure 1). The lack of a perfect age-distance correlation within hotspot-related archipelagos in slow-moving plates such as the Canary archipelago (Figure 1a) reflects the long-term spatial focusing of the volcanic activity within a restricted area (Meyzen et al., 2015).

Details are in the caption following the image

Geological overview. (a) Simplified map of the Canary Islands Seamount Provinces (CISP) location, including islands and associated seamounts, relative to Africa over the last 60 Ma. The stars track the location of the Canarian hotspot assuming a fixed position. Africa is held fixed to show relative motion of hotspot (Carracedo et al., 1998; Holik et al., 1991). At ∼30 Ma, when the plate velocity drastically slows down (Gaina et al., 2013), the real position of the hotspot activity is significantly shifted from its predicted position. The archipelago construction stages do not reflect a regular progression in volcanism age toward the west. Ages in italic indicate the oldest dated subaerial volcanism of each island. (b) Simplified geological map of Lanzarote. Red frame indicates the study area in the northern region of Lanzarote, while the red dashed line marks the path of La Corona lava tube. The geological unit contacts have been established from our study and earlier work by Hoernle and Carracedo (2009).

2.1 Lanzarote

2.1.1 Geological Setting of Lanzarote Island

Lanzarote island is the north-eastern extension of the Fuerteventura-Lanzarote volcanic platform, as the sea depth in the narrow La Bocaina strait between the two islands does not exceed 40 m. Indeed, Lanzarote and Fuerteventura are the subaerial parts of a single volcano built along a possible fissure striking northeast-southwest parallel to the African coast. Lanzarote could be the longest-lived island of the archipelago, since the oldest subaerial volcanism is dated at ∼15.5 Ma (Figure 1b; Carracedo et al., 1998; Coello et al., 1992; van de Bogaard, 2013), while the last activity took place in 1824. It is old enough for erosion to have drastically smoothed its relief compared to other Canary Islands, reaching a mere 671 m at the highest point.

Early submarine remnants of the shield building stage do not outcrop on Lanzarote (Hoernle and Carracedo, 2009). The oldest accessible massifs (Los Ajaches and Famara) are of Miocene age and represent late cycles of the shield stage (Hoernle and Carrecedo, 2009). Los Ajaches Massif was emplaced between ∼15.5 and ∼5.7 Ma in the southern part of the island (Hoernle and Carrecedo, 2009, Figure 1b), while a distinct magmatic event generated the Famara Massif between ∼10.2 and ∼6.3 Ma in the north-eastern part of the island (Abdel-Monem, Watkins and Gast, 1971; Carracedo et al., 2003; Coello et al., 1992; Hansen Machin and Pérez Torrado, 2005; Cabrera Vega, 2010; van den Bogaard, 2013, Figure 1b). Hiatus periods separating the three main different stages of Famara shield subaerial growth are marked by the presence of soils and calcretes deposits particularly well-developed at the top of each succession (Cabrera Vega, 2010; Carracedo et al., 2003; Coello et al., 1992; Hansen Machin and Pérez Torrado, 2005; Lomoschitz et al., 2016).

The more recent post-erosional stage took place in the Quaternary as a massive outpouring of fluid basalt, fed by a northeast-southwest alignment of volcanic vents, that flooded most of the surface of the island (Cabrera Vega, 2010; Carracedo et al., 2003; Coello et al., 1992; Hansen Machin and Pérez Torrado, 2005; van den Bogaard, 2013). In the north of Lanzarote (where our study area is located), this activity crowns the massif of Famara with the build-up of the volcanic system of La Quemada-La Corona-Los Helechos between ∼91 and ∼21 ka (Hansen Machin and Pérez Torrado, 2005; Hoernle and Carrecedo, 2009, Figures 1 and 2). The last volcanic phase corresponds to the Historic eruptions occurring in the central region of the island in 1730–1736 and 1824 (Gómez-Ulla et al., 2018; Hoernle and Carrecedo, 2009, Figure 1b).

2.1.2 La Corona Volcano

La Corona volcano is one of the edifices of the La Quemada-La Corona-Los Helechos volcanic alignment (Figures 1b and 2e). Although these edifices were thought to be derived from a single eruption for many years (Luis and Quirantes, 1984), Los Helechos and La Quemada emplaced around ∼91 ka, while La Corona erupted later during the Last Glacial Maximum (LGM) around ∼21.6 ka (Figure 2b; Carracedo et al., 2003; Hansen Machin and Pérez Torrado, 2005).

Details are in the caption following the image

Reconstruction of the evolution of the East coast of Famara massif, in relation to the transgression-regression phases that allowed the formation and subsequent flooding of La Corona lava tube system: (a) Erosion and excavation of the Mio-Pliocenic shield of Famara; (b) quaternary re-activation of the northern region. The lava flows of La Quemada and Los Helechos form a widespread lava field over the eroded plain of the Mio-Pliocenic shield; (c) Eruption of the volcano La Corona and formation of the lava tube; (d) The rising sea level, in the present interglacial, has left the submerged lava tube at the current level (modified from Carracedo et al., 2003); (e) Simplified geological map of the northern region of Lanzarote. The map presents the locations of interest in the studied area in the municipality of Harìa (Lanzarote, Canary Islands) and the location of the collected samples.

The higher (than Canary average) dimensions of La Corona volcano (cone width (Wco) = 1.500 m; crater width (Wcr) = 500 m; cone height (Hco) = 269 m) can only be generated by voluminous eruptions and high effusive rates (Carracedo et al., 2003). Over its eastern flank, adjacent to its main crater, a secondary crater is visible within a 200 m wide channel-like depression terminating in a potential sinkhole at the bottom of the volcanic edifice (Figure 3). A similar process occurred at Fagradalsfjall (Geldingadalir) Volcano in Iceland during an eruption in June 2021 (see video, from 01:16).

Details are in the caption following the image

Aerial and satellite images of the lava tube path. (a) View of the eastern downslope of La Corona volcano taken from the secondary crater (see red star in b). The red circle marks the inflation sink position of the lava tube, while the white dashed line indicates the lava tube path. Large circular to elongated cavities in the background correspond to the Jameos. (b) Plan-view of the lava field and skylights of the La Corona tube system. (a) Jameos de los Molinos; (b) Jameos de Arriba; (c) Jameo de Francisco Leòn; (d) Jameo de los Prendes; (e) Jameo de la Gente; (f) Jameo Tacho; (g) Jameo Cumplido; (h) Jameo Agujerado; (i) Jameo de la Puerta Falsa; (j) Jameo Redundo; (k) Jameo de los Verdes; (l) Jameo de los Lagos; (m) Jameos del Agua (see also in Supporting Information S1). The nomenclature adopted is updated, and sometimes differs from that used by Montoriol-Pous and De Mier, (1969). The dash-dotted black line highlights the topographic step at the base of the Famara massif.

The edifice construction of La Corona (Figure 2c) happened into two main phases: 1) an initial Strombolian episode during which the volcanic cone built up, and lapilli deposits were radially spread in a wide area around the main vent and 2) a second phase dominated by effusive lava emission. The cone morphology suggests a moderate intensity of Strombolian explosions with a greater accumulation of pyroclasts close to the eruptive centre forming the steep slopes (28–30°) of the volcanic edifice (Bravo, 1964; Carracedo et al., 2003). The subsequent effusive phase can be subdivided into two different episodes of lava emission. The first event (from here onwards C1) was characterized by huge volumes of basaltic pahoehoe-type lava, that flowed down the eastern slope burying the pyroclastic material of the Strombolian phase and the previous La Quemada-Los Helechos lava flows (∼91 ka, Carracedo et al., 2003). This massive discharge of lava enlarged the areal extent of the island toward the northeast by ∼100 m beyond the former coastline (Hansen Machin and Pérez Torrado, 2005; Figure 2e). Subsequently, a second effusive episode releasing a greater volume of viscous lavas of predominant aa-type (Figure 2d) took place (from here onward C2). C2 only partially covered the lava field previously created by C1 (Figure 2e). These two main effusive events produced a broad fan-shaped basaltic plateau of ∼18 km2, called by the local s the Malpaìs de La Corona. The lava tube, digging the south eastern side of the volcano, is of uncertain attribution, and may have been formed during any of the emission episodes of C1 and C2.

2.1.3 La Corona Lava Tube System

The La Corona lava tube originates from a sinkhole, located at the base of the south-eastern side of La Corona volcano and heads from there toward the coast with a northwest-southeast direction (Figure 3). It has a length (Lt) of ∼7.6 km (of which ∼6.0 km are subaerial and ∼1.6 km submerged), with a total extension to ∼9.7 km when side branches and upper levels are considered. The conduit has a mean width (Wt) of ∼13.7 m and average height of ∼10 m and a total difference in topographical elevation of ∼374 m (∼310 m above current sea level and ∼64 m below it) (Sauro et al., 2020).

The tunnel's sinuous path can be easily tracked owing to the numerous skylights and collapses (Jameos in the local dialect and Pukas in Hawaiian) forming a northwest to southeast chain recognizable both from the surface and aerial views (Figure 3).

The La Corona lava tube is not walkable for its entire length; numerous collapses obstruct the cave passages in the upper part. Each of the major Jameos networks (Jameos de los Molinos, Jameos de Arriba and Jameo de Francisco Leòn, Figure 3) in the vicinity of the La Corona crater (<1 km) located on the plateau of Famara massif (>250 m) prevents humans from accessing the interconnections between sections. The last ∼1.7 km of the tube are submerged below sea level, because of the marine transgression that occurred after the LGM (∼21 ka; Bard et al., 1990; Labeyrie et al., 1987). The most accessible and walkable portions of the tube are stretched over six sections (Figure 3): (a) Jameo de los Prendes—Jameo de la Gente (∼1,170 m); (b) Jameo de la Gente—Jameo de la Puerta Falsa (∼1,165 m); (c) Jameo de la Puerta Falsa—Jameo de los Verdes (∼1,290 m) and (d) Jameo de los Lagos/Perdido (∼730 m); (e) los Jameos del Agua (∼350 m); (f) Tùnel de la Atlantìda (Carracedo et al., 2003; Isler, 1989; Mendo and Ortega, 1988; Montoriol-Pous and De Mier, 1969; Sauro et al., 2019; Smith, 2010, Figure 4).

In some places, the tube has developed a multi-level architecture. Superposed levels can be formed by two (not mutually exclusive) processes: (a) smaller branches reconnect with a principal one beneath (Bravo, 1964), (b) a smaller lava tube forms at the base of a partially drained larger tube (multi-levels/tube-in-tube structure; Bunnell, 2013; Shick, 2017). The latter mechanism is visible in several places, in particular along the la Gente—la Puerta Falsa stretch, where an upper conduit runs below Jameo Tacho, Jameo Cumplido and Jameo Agujerado, and is not reachable from the tube-in-tube structure beneath (Figure 5d).

From Cueva de los Lagos toward the coast, due to its lower elevation, the tube becomes exposed to marine ingression and thereby to tidal changes of the water level. The final submerged segment of the lava tube, called Tùnel de la Atlantìda (Carracedo et al., 2003; Isler, 1989; Mendo and Ortega, 1988) starts from Jameos del Agua (nowadays a tourist site, Figures 3m and 7) and continues for over ∼1.6 km under the ocean, reaching a depth of −64 m ca. below sea level, where the tube ends abruptly, forming a large spherical cavity (10 × 10 m) (Carracedo et al., 2003; Chappell and Shackleton, 1986). There are no discernible signs of magma-seawater interactions along most of the tube extension (∼7.6 km). This suggests that, at the time of the eruption and therefore of the emplacement of the lava tube, the sea was in a phase of regression and the coastline was about 80–120 m lower than today. Thus, the eruption took place while the marine platform was a dry land (Carracedo et al., 2003; Sauro et al., 2019). This emplacement framework is coherent with the paleoclimatic data collected by Labeyrie et al. (1987), that places the sea level variation between 18 and 21 ka, during the LGM event.

3 Materials and Methods

3.1 Geological Mapping

A new geological map at a scale of 1:25,000 m for a wide area covering approximatively 70 km2 in the north-eastern part of Lanzarote was created using high-resolution images (DTM 5 m from the Lidar topography provided by the Spanish Geological Service) and field survey observations (Figure 2e). The field work has been carried out both at the surface and inside the lava tube.

One of our primary goals was to first check geological boundaries deduced from preliminary interpretations of remotely sensed imagery. Photogrammetric acquisitions with Unmanned Air Vehicles (UAVs) at the surface and laser scanning surveys within the tube were also performed during the field campaign. The UAV surveys were constrained by GPS points within the tube collapses themselves and at their edges; these edge points were used to constrain the underground LiDAR data to the surface by point-cloud co-registration. This procedure allowed integration of the tube into a geographic-projected reference system and correlation of the contacts between units within the tube, with those observed and mapped at the surface.

3.2 Laser Scanning

Terrestrial Laser Scanning (TLS or T-LiDAR) technology was employed to reconstruct the three-dimensional model of the volcanic cave system, using two different devices: a Leica HDS 7000 laser scan mounted on a tripod and a wearable Leica Pegasus Backpack for acquisition during motion within the cave.

During the acquisition phase the Leica HDS 7000 instrument recorded 1,016,727 pts/s, with a maximum range of 187 m. Since the Leica HDS 7000 is a rather heavy and delicate instrument, for some of the more difficult to explore portion of the tube, we used the Leica Pegasus Backpack, which is more manageable (∼12 kg weight), easily wearable, and optimized for indoor data capture. In this case the instrument recorded 600,000 pts/s, at a maximum range of 50 m.

We performed a total of 440 scans, to map ∼5.6 km of cave passages, including the different sections connected to each other through the sinkholes. The main part of the tube system, from Jameo de los Prendes to Cueva de los Verdes (Plate 3 in Supporting Information S1), was mapped with a resolution of 0.25 m. To be able to georeference the point clouds, without any GPS information (since the survey was conducted underground), the point clouds representing various tube sections have been mutually co-registered, initially with the UAVs photogrammetric surveys and then integrated to the LiDAR topography of Canary Island (gridded at 5.0 m). The analysis and treatment of LiDAR data were carried out using CloudCompare (CC) software and the Python programming language. Working on point cloud clusters enabled us to obtain both perpendicular and transverse sections of the lava tube (in Supporting Information S1). In addition, the three-dimensional reconstruction of the tube allowed the recognition of several internal morphologies such as flow ledges (or benches, Figure 5a), cupolas (dome-like heightening in a lava tube's ceiling, Figure 4b) and windows (openings between different levels, Figure 4c) as cataloged in Bunnell (2013).

Details are in the caption following the image

Schematic cross-sections of the anchialine portions of La Corona lava tube system. (a) Jameo de los Lagos, (b) Cueva de los Lagos; (c) Jameos del Agua lagoon (dash-dotted transversal lines indicate the areal extent occupied by the tourist complex); (d) Túnel de la Atlántida (inset); (e) Escondido lagoon; (f) Dome room; (g) Montaña de Arena “La Duna”. Shaded areas in blue mark the position of current sea level. While horizontal dashed line indicates the possible sea level during tube formation. Vertical scale on left axis is exaggerated (modified from Wilkens et al., 2009). The zoomed section of Túnel de la Atlántida (d) is modified from Javier Lario, “Proyecto Sublantida” (https://gama.uned.es/proyecto-sublantida/).

3.3 Sample Selection, Location and Analytical Procedure

3.3.1 Magmatic Rocks

The sampling strategy was primarily aimed at collecting the whole spectrum of lithological units intersected by the lava tube along most of its length. We will refer to these units as pre-existing lava flows. These samples were collected above or below a recurrent red pyroclastic layer at three cross-sections, located at distances of approximately ∼1.5 km (SUB1), ∼2.3 km (PL1, PC1) and ∼2.6 km (LPUB1, LPLB1, LPCV1) from the sink as shown in Figures 2 and 6a6c. The second goal was to sample the “cave material” (PC1 and LPCV1, Figures 6a–6c), namely the inner flows of the tube (e.g., floor deposits, lining walls, etc.). A sample (CRB1) from the upper Famara unit was also collected to allow a geochemical comparison with samples from the tube (Figure 2). Samples were sorted with respect to phenocryst abundance, mineralogical assemblage, vesicularity and extent of alteration. After this categorization, major and trace element analyses were performed on seven samples (Table 1). A brief description of their petrography is given in Supporting Information S1. Major and trace element contents were measured at the Service d’Analyses des Roches et des Mineraux (SARM-CNRS) at the Centre de Recherches Pétrographiques et Géochimiques (CRPG, Nancy, France) by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) and Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES) (Table 1). Analytical errors are less than 2% for SiO2, Al2O3, Fe2O3, MgO and CaO, and 5%–15% for Na2O, TiO2, MnO, P2O5 and K2O. Long-term precision (% RSD) for the procedure is typically better than 15% for trace elements and 5%–10% for Rare Earth Elements (REE).

Table 1. Chemical Analysis of Samples From the Pre-Existing Lava Sequence of the Tube Walls and From Famara Massif
Famara massif Los Helechos lava flows La Corona lava flows Lining walls lava flows
Samples CRB1 PL1 LPLB1 SUB1 LPUB1 PC1 LPCV1
Type of rocka Bsn Bsn Bsn B-alk B-thl B-alk B-alk
Major elements (wt. %)
SiO2 41.88 42.40 43.39 45.59 48.97 48.10 46.79
TiO2 2.95 3.24 2.98 2.55 2.21 2.93 2.83
Al2O3 13.66 11.89 12.26 12.66 13.64 14.04 13.31
Fe2O3T 12.45 13.38 13.02 12.01 11.68 11.60 11.81
CaO 12.17 10.71 10.30 10.32 9.25 10.30 10.37
MgO 9.07 12.78 12.22 11.53 9.12 7.36 9.38
MnO 0.19 0.19 0.19 0.17 0.16 0.15 0.16
K2O 0.51 0.46 0.99 0.76 0.64 0.87 1.02
Na2O 3.25 3.45 3.39 3.03 3.24 3.62 3.29
P2O5 0.95 0.81 0.73 0.76 0.36 0.48 0.52
LOI 2.14 −0.09 0.10 −0.10 0.29 0.12 0.00
Total 99.232 99.22 99.55 99.27 99.55 99.56 99.47
Trace elemets (in ppm)
Sc 29 26 25 24 22 24 25
V 273 269 259 241 189 307 532
Cr 310 488 533 491 355 313 401
Co 56 80 84 113 81 93 75
Ni 135 315 308 309 221 133 208
Cu 76 50 58 32 74 86 60
Zn 112 121 120 114 117 116 111
Ga 22 21 21 20 21 23 22
Cs 0.84 0.48 0.52 0.25 0.33 0.28 0.31
Rb 5.95 23.05 18.47 17.01 12.77 15.94 19.73
Ba 453 461 425 423 211 301 337
Th 7.33 8.24 7.51 8.92 3.75 5.60 6.05
U 1.91 1.99 2.15 2.47 1.06 3.16 1.89
Nb 76 67 60 58 28 49 49
Ta 5.33 4.93 4.30 3.75 2.05 3.88 3.67
Sr 929 853 820 856 490 671 752
Zr 393 270 242 211 153 269 225
Hf 7.71 6.29 5.55 4.64 3.73 6.25 5.34
La 58 54 50 58 23 36 40
Ce 127 110 101 114 49 74 81
Pr 15.19 12.90 11.87 13.18 6.05 8.87 9.56
Nd 60 51 47 51 26 36 38
Sm 11.56 9.90 9.37 9.80 6.19 7.99 8.03
Eu 3.49 3.04 2.91 3.00 2.08 2.58 2.55
Gd 9.13 7.91 7.55 7.84 5.51 6.88 6.81
Tb 1.21 1.07 1.02 1.06 0.80 0.95 0.94
Dy 6.50 5.66 5.53 5.67 4.48 5.18 5.08
Ho 1.17 1.01 0.97 0.99 0.81 0.92 0.90
Er 2.72 2.32 2.25 2.30 1.92 2.12 2.10
Tm 0.35 0.30 0.29 0.29 0.25 0.28 0.27
Yb 2.05 1.72 1.68 1.71 1.46 1.59 1.59
Lu 0.30 0.25 0.24 0.24 0.21 0.22 0.23
Y 30 25 25 25 21 23 24
Pb 2.34 2.42 2.21 2.15 1.34 2.08 2.15
  • Note. Types of rocks: Bsn = Basanite; B-alk = Alkaline basalts; B-thl = Tholeiitic basalts. The first line (Famara massif, Los Helechos lava fows, etc.) indicate the layer where the basalts were sampled, you can identify those as: 'Sampling site'. The second line (CRB1, PL1, LPLB1, etc.) indicate the IDs given to the samples.
  • a Bsn = Basanite; B-alk = Alkaline basalts; B-thl = Tholeiitic basalts.

4 Results

4.1 Subsurface Morphologies

The massif of Famara in the northern part of Lanzarote is made up of at least three different volcanic stages interlayered by erosive discordant surfaces, sedimentary deposits and soils (Dóniz Páez et al., 2002; Hansen Machin and Pérez Torrado, 2005; Meco et al., 2003), and is overlapped by the Quaternary volcanic group of La Quemada-La Corona-Los Helechos (Figures 1b and 2e).

The La Quemada lava flows stretch from the volcanic cone of La Quemada (de Orzola) through the north-eastern side of Famara massif to reach as far asthe northern coast of Orzola (Figure 2e). The Los Helechos volcanic sequence begins with pyroclastic deposits (outcropping inside the quarry of Maguez and along the sides of the Mesa de Los Llanos nearby) followed by widespread lava flows emitted from the vent alignments of Los Helechos which is formed from north-east to south-west by La Cerca, El Helecho and La Quemada de Maguez vents (Figure 2e). Los Helechos lavas cover a wide area that - from the volcanic cones in the Maguez district—Reach the eastern coast at the villages of Arrieta and Punta Mujeres (Figure 2e). A small lava flow on the western side of the Los Helechos volcanic edifice crosses the physical boundary constituted by the Famara cliff to reach the western coast (Figure 2e).

La Quemada and Los Helechos lava fields were covered by the effusive events of La Corona, creating a new fan-shaped area that extended beyond the current coastline (Carracedo et al. (2003); Hansen Machin and Pérez Torrado, 2005). While entrenching the south-eastern slopes of La Corona volcano, the lava tube system of La Corona may have intersected to some extent the lava fields generated by the La Quemada-La Corona-Los Helechos alignment (Figure 2e). At the entrance to collapses, the pre-existing lava succession has been clearly cross-cut by repeated injections of molten lava flows. This constitutes an unquestionable indication of an inflation-type formation process (sensu Kempe, 2012).

The huge amount of debris inside both the collapses and the conduit indicate breakdowns (collapses of ceiling and lining walls; Bunnell, 2013), which occurred mainly during and after the cooling phase of the lava tube itself. Often, the conduit walls are covered by a coating of thin layers of lava (lining walls in Bunnell, 2013; Harter, 1972; or glaze in Shick, 2017) accreted onto the underlying coarser basalts during the passage of the molten lava flows. Once the first layer has formed, repeated temporal fluctuations in lava height and hence in the volume rate of flow can deposit additional layers, glazing the tube walls, and covering up the pre-existing lava flows. If the stream-flow rate remains stable, deposits on the walls will grow into flow ledges. Flow ledges are terrace-shaped lava accumulations, preferentially formed at the external wall of meandering lava tubes (Figure 5a). Morphologies such as flow ledges grow by overbank events involving thin lava layers (Kempe, 2012). In the case of prolonged and gradual decreases of flow rates, the flow ledges might grow up as a “secondary ceiling” above the flowing lava, forming a tube-in-tube structure (Figure 5d; Kempe, 2012 and Shick, 2017, for similar examples in Hawaii). This process can be recognized by the presence of multiple levels of tubes that run in parallel, one above the other, often vertically connected through cavities called windows (Figure 5c).

The La Corona lava tube shows a single-type large tunnel structure, evolving toward a multi-level architecture in some sections. From Jameo de los Prendes to Jameo Tacho it has a single section (15–20 m wide, Plate 2 in Supporting Information S1), while from Jameo Tacho to Jameo de la Puerta Falsa the section is divided into two conduits (4–5 m wide, Plate 2 in Supporting Information S1) and, in some sporadic segments, into three conduits (3–4 m wide, Figure 5c and Plate3 in Supporting Information S1). In this particular section, the upper conduits are not fully accessible because of collapses that obstruct their entry points, while the lower conduit runs undisturbed until Jameo de Puerta Falsa. Some sections between Jameo de la Puerta Falsa and Jameo de los Verdes are similarly divided into two superimposed conduits (5–10 m wide, Plate 3 in Supporting Information S1). At Cueva de los Verdes, the different conduits merge into a single large tube, reaching a maximum height of 37 m, often divided into two to three levels separated by secondary ceilings. In Jameo de los Verdes, the tube is interrupted by a breakdown, but can be accessed again at Jameo de los Lagos (Figure 3). From there, the tube mainly continues as a single conduit up to the seven underground lagoons (from which the cave takes its name). Recently it was discovered that on top of the second underground lake a second superimposed conduit develops for 70 m with heights of up to 15 m. From Cueva de los Lagosit was originally possible to reach the cave complex of Jameos del Agua, where progress is blocked by tourism infrastructure. Then, from Jameos del Agua, the tube “sinks” under the present-day sea level for 1.6 km as the Tùnel de la Atlantìda.

The tube is also characterized by the presence of knickpoints in the floor level, that facilitate the downcutting processes (Allred and Allred, 1997; Kauahikaua et al., 1998; Keszthelyi and Self, 1998 and Kempe, 2012; for similar examples in Hawaii). The most evident knickpoint is the one that marks the topographic step between Famara massif and La Corona lava field plane (Malpaìs de La Corona). Other knickpoints were identified by the three-dimensional model of the tube. They are probably related to sharp changes in the paleo-topography reflected in the tube channel slope.

In association with knickpoints or enhanced tube curvatures, cupolas can develop in the tube ceiling. They form where lava under pressure caused melting upwards into a hardened ceiling (Figure 5b). Under sustained flow, the cupola formation process might breach the ceiling creating a window, toward an eventual upper tube (Figure 5c), or even a skylight. Lower levels, representing the last active part of the tube, may overflow into upper levels through windows, leaving deposits along the flow ledges that can grow and eventually seal the windows (Figure 5d).

Details are in the caption following the image

Views of the inner part of La Corona lava tube system: (a) Flow ledge (or bench) and, at the floor level, a small tube-in-tube structure (ph. Robbie Shone); (b) cupola; (c) window: An opening between two conduits (ph. Gustavo (d) Santana Gomez); (d) multi-level structures (or tube-in-tube structures). The formation of a new secondary roof is often due to lava overflow into upper levels through windows, leaving deposits along the flow ledges that can grow and eventually seal the windows.

4.2 Linings and Entrenched Units

Lining wall breakdowns are more frequent in the initial part of the tube, exposing pre-existing lava flow sequences cross-cut by the tube emplacement. A reddish deposit of lapilli in an interlayered position in the lava flow pile is conspicuous when examining pre-existing lava flows sections (Figure 6). This pyroclastic deposit has a variable thickness, ranging from 40 cm to 1 m. The pyroclastic level is composed by layered pyroclastic fallout deposits, that do not show any cross stratifications or other evidence of pyroclastic surges or pyroclastic flow arrangement (Figure 6). The sequence is characterized by parallel layers of pyroclasts sorted by size ranging from mm to cm scales. Very similar in aspect to the dark pyroclastic material that covers the La Corona volcano sides, the red layer appears repeatedly in most segments of the tube, if not hidden by lining walls (Figure 6). For these reasons, we consider it to be a reference stratigraphic level. The reddish color of the lapilli layer is due to oxidation driven by the presence of water vapor and oxygen within the lapilli intergranular porosity at high temperature during the flow of molten lavas within the tube (Waters et al., 1990; and see Shick, 2017 for similar examples in Hawaii).

Details are in the caption following the image

Sample locations along transversal sections of the tube and red lapilli layer identified as marker level inside the lava tube system: (a) Upstream branch of the Jameo de los Prendes—See Section 1.03 in Plate 1, in Supporting Information S1; (b) downstream branch from the Jameo de los Prendes to Jameo de la Gente—See Section 1.26 in Plate 1, in Supporting Information S1; (c) upstream branch from the Jameo de la Puerta Falsa to Jameo de la Gente—See Section 2.06 in Plate 2, in Supporting Information S1; (d) Deposit of fallout material due to the initial Strombolian activity of La Corona volcano; (e) the red layer is often covered by the lining walls and is exposed by the breaking down of those latter; (f) exposed outcrop below the flow ledge in Jameo de la Puerta Falsa. The inset shows a zoom in of the outcrop.

Lava sheets under and above this reference level show clear differences in morphology, thickness, phenocryst content and alteration state. The lava flows underneath the pyroclastic level (from here onwards termed lower units) are more massive and thicker (60–140 cm) than the much porous upper flows (from here onwards termed upper units). They are also more altered and fractured and show “onion skin” shells typical of spheroidal weathering (e.g., Chatterjee and Raymahashay, 1998; Ollier, 1971), indicating rock-water interaction before the emplacement of the pyroclastic layers. However, the absence of any trace of soils and/or calcretes—Which are typical of the Famara complex—Suggests that these flows must be largely subsequent to the Mio-Pliocenic Famara event. The upper, pre-existing, lava flows have instead a moderate individual thickness (40–80 cm), an aphyric texture and do not show any evidence of spheroidal weathering, indicating a limited duration of exposure to exogenous agents. However, they are less massive and present a higher degree of vesicularity (avg. 6% for lava flows' cores in the upper unit vs. less than 1% for cores in the lower units, see in Supporting Information S1). Neither of the basaltic successions (lower and upper units) show any signs of interaction with seawater during the active stream-flowing of molten lava.

4.3 Paleo-Topography Derived From Three-Dimensional Reconstruction

As previously described in Section 2.1.1, the La Corona eruption began with Strombolian explosions, that spread a layer of pyroclastic material over a wide area surrounding the volcanic edifice. This blanket of lapilli was then successively covered by effusive lava flows from both first (C1), and second (C2) major phases. The external deposits of dark lapilli were mapped by remote sensing and field surveys (Figure 2e), while the positions of the lapilli deposits within the tube have been reported inside the reconstructed three-dimensional internal tube morphology retrieved by TLS. Since GPS measurements are not possible underground, the lapilli layer locations within the tube were retrieved using the distances from the Jameos entrances, relative height of the lapilli deposit within the tube, photos and campaign notes. In this way, the inner path of the lapilli layer is constrained in relation to both its underground exposures within the tube and outcropping deposits at the surface, as displayed in the geological map (Figure 2e).

Using ArcGis—Kriging Tool and Cloud Compare software programs, we were able to generate a mesh surface which, to a good approximation, represents the paleo-topography covered by the layer of lapilli of the Strombolian event. This surface was obtained by the interpolation of the red lapilli layer within the tube and the dark lapilli deposits outcropping at the surface. As expected, the reconstructed paleo-topography is rather smooth and gentle, similar to the current one. The presence of such flat paleo-relief might be related to the very long hiatus between the eruptions of the last Miocene and early Pleistocene units, which would have enabled erosion to drastically reduce most of the pre-Pleistocene island relief. In addition, the reconstructed surface (Figure 7), highlights how the pyroclastic horizon runs along almost the entire lava tube (see Supporting Information S1).

Details are in the caption following the image

Paleo-topography reconstruction. (a) Surface reconstruction of the paleo-topography based on the position of the red pyroclastic level; (b) simulation of the interaction between the reconstructed surface and the current topography; (c) perpendicular section of the tube. Note that the reconstructed surface cuts the tube in several points. All reconstructions are done by using Cloud Compare software.

Other detailed observations are:
  1. Approximately 30–40 m upstream from the collapse of Jameo de los Prendes, the red layer is visible at the top of the tube vault, before being buried behind the lining walls (Sections 1.01 in Plate 1, Supporting Information S1) and reappears at floor level next to the Jameo entrance (Figure 6a; Sections 1.03 in Plate 1, Supporting Information S1).

  2. In the upstream branch of Jameo de la Gente, the red layer roughly crosscuts the upper height of the tube (Figure 6b; Plate 1 in Supporting Information S1).

  3. From Jameo de la Gente to Jameo de la Puerta Falsa (Figure 6c; Plate 2 in Supporting Information S1), the reconstructed layer is often covered by the lining walls, but on some occasions, reddish lapilli are visible right under aligned flow ledges.

  4. Downstream, in the section from Jameo de la Puerta Falsa to Jameo de los Verdes (Plate 3 in Supporting Information S1), where the tube divides into a tube-in-tube structure, the reconstructed layer lies at the top of the lower tube. Indeed, pockets of reddish lapilli appear under the flow ledges responsible for the tube-in-tube structure.

  5. In Jameos del Agua, the layer of reddish lapilli lies in the lower half of the main conduit, forming a more hollowed side ledge (which was largerly covered by concrete consolidation during the adaptation of the site for tourism).

Therefore, it is certain that the reddish lapilli layer is present along the whole terrestrial part of the tube development. In cross-sections, it can be found at the ceiling, close to the ceiling, in the middle or lower part, but never at the lower floor level.

4.4 Geochemical Results

We document here geochemical variations for aphyric samples from the tube linings and the pre-existing lava flows cross-cut by the tube. Their variations are compared with earlier data, acquired from the same units retrieved from GEOROC, during previous studies (Carracedo et al., 2003; Ibarrola, 1970; Ibarrola and Lopez, 1967; Lundstrom et al., 2003; Santin, 1969; Thomas et al., 1999). Our goal is to identify any significant chemical compositional differences between pre-existing units that could help us to pin-point the nature of the units intersected by the lava tube underground.

4.4.1 Chemistry of Quaternary Volcanism Compared to the Mio-Pliocene Famara Complex

The Mio-Pliocenic samples of the Famara complex, CRB1 and literature data (Carracedo et al., 2003; Ibarrola, 1970; Ibarrola and Lopez Ruiz, 1967; Lundstrom et al., 2003; Santin, 1969; Thomas et al., 1999), were used as a benchmarks to compare our samples with those relating to Quaternary volcanism (Figure 8). When major element chemistry is considered in relation to MgO (Figure 8), the Famara field is indeed well distinguishable from those related to the Quaternary volcanism. In particular, progressively from the Mio-Pliocene samples to the Quaternary samples there is a tendency for the proportion of SiO2 to increase, while CaO, and to a lesser extent FeOT and TiO2, decrease (Figure 8).

Details are in the caption following the image

Major element variations (wt %) versus MgO (wt %) for lavas from the northern region of Lanzarote including samples from this work and those of Carracedo et al. (2003). Fields have been drawn from literature data of [1] Ibarrola and Lopez Ruiz, 1967; [2] Santin, 1969; [3] Ibarrola, 1970; [4] Thomas et al., 1999; [5] Lundstrom et al., 2003; [6] Carracedo et al., 2003. All data were re-normalized to 100% (all iron as FeOT).

The composition of all our samples taken from the units above the pyroclastic level (upper units) fit with the composition of the Quaternary volcanism reported in the literature (Los Helechos, La Quemada, La Corona) and fall outside the values obtained from the Famara field, with the only exception being the Na2O-MgO content. The lining wall samples (PC1, LPCV1) are in contrast compositionally different from the homologous ones reported in literature, but still clearly distinguishable from la Famara field.

The samples taken from beneath the pyroclastic horizon (lower units' samples, PL1 and LPLB1) are the only ones that systematically overlap with the Famara field (Figure 8). Nonetheless, their potential association with the Famara series is ruled out by trace element analysis (Figure 9). The Famara spiderdiagram is distinct from those of all the other units (above and beneath the pyroclastic horizons as well as the lining walls) because of its characteristic positive slope in the trend from Ba to Ta and the noticeable absence of a pronounced negative anomaly in Zr-Hf (Figure 9).

Details are in the caption following the image

Trace element concentrations in lavas from La Corona tube system normalized to the primitive-mantle values of Sun and McDonough (1989). Light green and dark green fields respectively denote data from Thomas et al. (1999) and Lundstrom et al. (2003).

4.4.2 Major Element Chemistry of the Quaternary Volcanism

In a total alkali (Na2O + K2O) versus silica (TAS) diagram (Figure 10), there is a clear systematic difference in the nature of lavas located below and above the lapilli reference layer. Pre-existing lavas underneath the lapilli layer (PL1 and LPLB1) are classified as basanites. LPLB1, a sample from the lower units, is the most alkali-enriched sample of this group, overlapping with previous data published for Los Helechos (Carracedo et al., 2003; Lundstrom et al., 2003) and La Quemada (Carracedo et al., 2003; Ibarrola and Lopez Ruiz, 1967, Figure 10). In contrast, wall lining deposits (PC1, LPCV1) and lava flows (SUB1, LPUB1) located above the reference layer systematically fall in or near the basaltic field (Figure 10), defined by earlier published data for La Corona samples (Carracedo et al., 2003; Ibarrola and Lopez Ruiz, 1967; Lundstrom et al., 2003; Thomas et al., 1999). Another systematic difference between compositions of the samples from the upper andthe lower units is the MgO range (Figure 8). MgO among the lower units samples (PL1 and LPLB1) has a 12.8%–12.2% range, with a mean of 12.5%, substantially higher than the 11.5%–9.1% range 10.3% and mean of of the upper units (SUB1 and LPUB1) and the 7.4%–9.4% range, and 8.4% mean, of the lining wall (LPCV1 and PC1) samples. Note that the lining wall samples are among the most differentiated samples of our data set.

Details are in the caption following the image

Total alkalis versus SiO2 (wt %) classification (TAS; Le Bas et al., 1986) for samples from the northern region of Lanzarote including samples collected for this study and those of Carracedo et al. (2003). Fields have been drawn from literature data [1] Ibarrola and Lopez Ruiz, 1967; [2] Santin, 1969; [3] Ibarrola, 1970; [4] Thomas et al., 1999; [5] Lundstrom et al., 2003; [6] Carracedo et al., 2003.

When major element chemistry is considered in relation to MgO, other geochemical differences become even clearer. Samples from lower units (PL1 and LPLB1) display higher FeOT and lower SiO2 and Al2O3 content than those of the upper units (SUB1 and LPUB1) and lining walls (LPCV1 and PC1). In detail, the lower unit sample LPLB1 lies within or closely adjacent to the field defined by previously published Los Helechos data (Carracedo et al., 2003; Lundstrom et al., 2003), while sample PL1 shows no overlap (Figure 8). The high TiO2 values place the lower units samples above those of both the La Quemada (Carracedo et al., 2003; Ibarrola and Lopez Ruiz, 1967) and La Corona compositional fields (Carracedo et al., 2003; Ibarrola and Lopez Ruiz, 1967; Lundstrom et al., 2003; Thomas et al., 1999), and they fall in the range of samples identified as from Los Helechos but with slightly higher MgO contents (Carracedo et al., 2003; Lundstrom et al., 2003). As expected, samples from the upper units (SUB1 and LPUB1) fall or plot closely adjacent to the field defined by earlier data for La Corona units (Carracedo et al., 2003; Ibarrola and Lopez Ruiz, 1967; Lundstrom et al., 2003; Thomas et al., 1999, Figure 8). The composition of our lining wall samples (LPCV1 and PC1) is enriched in Na2O and TiO2 compared to the upper units (SUB1 and LPUB1) (Figure 8). However, as already noted, they are remarkably compositionally unlike the previous lining samples published by Carracedo et al. (2003), even if their trace element patterns are closer to those of La Corona rather than Los Helechos. The differences with previous data published on other lining samples might potential be due to different degrees of contamination while the lava was flowing within the tunnel due to thermal erosion of pre-existing lava flows.

4.4.3 Trace Element Chemistry of the Quaternary Volcanism

Primitive-mantle-normalized incompatible element patterns for our samples (Figure 9) reveal subtle compositional differences between upper and lower units. While samples from lower units (PL1, LPLB1) have subparallel spiderdiagrams with an extremely restricted range of concentrations, the same is not true for samples from the upper units (SUB1, LPUB1). The latter exhibit two distinct spiderdiagrams, with greater enrichment in highly versus moderately incompatible trace elements for sample SUB1 relative to sample LPUB1 (Figure 9). Although spiderdiagrams of the lower unit samples (PL1, LPLB1) overlap with that of sample SUB1, a detailed examination shows a distinct fractionation in Th-U relative to Nb and a more pronounced negative Zr-Hf anomaly for SUB1 relative to samples from the lower units. If LPUB1 is also considered, it becomes clear that the steeper fractionation of Th-U relative to Nb is a general feature of the upper units relative to lower ones (Figure 9). The spiderdiagrams of the lining wall samples display a slightly lower fractionation of highly to moderately incompatible elements than SUB1 and samples from the lower units, although their overall patterns share more similarity with sample SUB1.

5 Discussion

5.1 Geochemical Constraints on the Units Crossed by the La Corona Lava Tube

There is a great compositional difference in terms of major elements between the upper and lower units enclosing the pyroclastic layer (Figure 8). Although samples from the lower unit, (PL1) do not complitely overlap with the field defined by the literature data (Carracedo et al., 2003; Ibarrola, 1970; Lundstrom et al., 2003) for the massif of Los Helechos, they represent an extension of major elements trends defined from published Los Helechos lava analyses to substantially higher MgO values (Figure 8). The other sample from the lower units (LPLB1) systematically falls within the Los Helechos field. We therefore, conclude that basanites (PL1 and LPLB1) from the lower units compositionally resemble samples from the Los Helechos sequence, confirming that the lava tube entrenched the pre-existing unit at its base. The lining sample (LPVC1) is also of clear La Corona affinity, while the other lining sample (PC1) is the most differentiated one falling in the extension of major elements trends established by previously published data for La Corona. Their primitive-mantle-normalized incompatible element patterns closely resemble that of sample (SUB1) from to the upper lava flow and which is interpreted to be an expression of the La Corona C1 magmatic event.

5.2 Lava Tube Genetic Mechanism and Morphologies

The structural and geological study conduited on the northern region of Lanzarote and in particular the field observation of the La Corona lava tube system coupled with the geochemical analyses raised the following considerations:
  1. The pyroclastic deposits at the surface mark the first Strombolian event of La Corona, which covered the Los Helechos sequences;

  2. The pyroclastic layer inside the tube can be correlated with early La Corona pyroclastic deposits onto at the surface, allowing the definition of a reliable paleo-topography at the time of the La Corona Strombolian event and before the C1 first stage of lava emplacements;

  3. The spatial positioning of the lapilli surface within the three-dimensional tube reconstruction shows (Figure 7) how the lapilli layer characterizes the pyroduct through the entire length of the tube.

All these points suggest that the lapilli deposit in-between the Los Helechos and La Corona lava flows, may have greatly facilitated the early inflation process of the La Corona tube system, as reported for other lava tubes in Hawaii (Greeley et al., 1998). With regard to the early stages of its development, our data suggest that the lava tube emplacement happened after the pyroclastic deposit and during the effusive stage of the La Corona event. In particular from field and geophysical data (Torrese et al., 2021), it is evident that the tube was excavated underneath 10–20 m of lava flows and was well-developed within ∼900 m from the main vent. Since the pyroclastic horizon is visible at 15 m of depth at Jameo de los Prendes (∼1 km down from the first Jameo de los Molinos, Figures 3b and 6a, and Section 1.03 in Plate 1, Supporting Information S1) and in view of its regional position and dip (Figure 7), it is very likely that the same layer is at a similar depth upslope, although in that case overlying the Famara massive. This implies that the tube was already progressing within the pyroclastic deposits just 20 m from the putative sinkhole at the base of La Corona volcano (Figure 3).

We therefore envisage that the general process of initial inflation would have begun as the effusion rate increased and the lava stream flooded laterally from the primary and secondary vents. The steep slope gradient promoted the thermomechanical excavation at the base of the volcanic edifice where the remnant of a potential sinkhole is visible (Figures 3 and 12). Such a process has indeed been reported by several authors as a viable mechanism for forming a sink pond (or plunge pool, sensu Allred and Allred, 1997; Bunnell, 2013; Greeley et al., 1998; Kauahikaua et al., 1998; Kempe, 1997; Keszthelyi and Self, 1998; Smith, 1998). It is very likely that the initial tube formation began with the deepening of the sink pond. The hot lava made its way through lava sheets stack inflation and downward thermal erosion through the pre-existing lava flows down to the level of pyroclastic layer, which was reached within a distance of 250 m from the sink. This suggests that the inflation between lava flows, as defined by earlier studies (Greeley et al., 1998; Hon et al., 1994; Kempe, 1997; Kauahikaua et al., 1998; Keszthelyi and Self, 1998), took place during a very early phase and mainly in proximity of the volcanic edifice. Then, most of the flow was afterward injected between the pyroclastic layer and the previously emplaced La Corona unit (C1). The repeated inflation of molten lava below not long before emplaced the C1 flows, allowed narrow streams to gradually incise the buried and unconsolidated pyroclastic layer (Figure 11), forming a series of tiny flattened elliptical ducts parallel to each other. Eventually one of these tubelets become dominant, while others stagnated. Once the lava flow reached the topographic step between the Famara massif and the current La Corona lava field overlying the former Los Helechos unit, it formed a lava fall. There the lava flow over-excavated deeper into the pre-existing lava field, reaching the pyroclastic layer interbedded between the C1 and Los Helechos flows. The general process would be similar to that described by Kauahikaua et al. (1998) in Hawaii, where the lava advanced as a complex, anastomosing system of small tubes.

Details are in the caption following the image

Thermo-mechanical erosional stages of the lava tube in cross-sections. (a) The primary effusive phase (C1) has covered the pyroclastic deposit of the initial Strombolian event; (b) the inception of the tube by inflation (possibly by lava flows from either C1 or C2 units) starts exploiting the pyroclastic layer; (c-d-e) progressive erosion and enlargement of the tunnel; (f) post-cooling phase. On the walls are visible different layers of linings and flow ledges. The floor is covered by blocks and debris, remnants of the breaking down of the ceiling and lining walls.

This is confirmed by recent Electrical Resistivity Tomography (ERT) surveys orthogonal to the tube in Torrese et al., 2021. The identified series of elliptically-shaped channels have a width ranging from 7 to 13 m with a ceiling depth from the surface of 14–29 m. Most of them appear to be presently isolated from the main tube and were partially undrained and therefore short-lived features. A minor proportion remained active and experienced widening by lateral thermo-mechanical erosion of the weak pyroclastic layer by persisting stream-flows of molten lava. This is in full agreement with the theoretical and experimental treatment of thermal-erosion by Jarvis, (1995) and Kerr, (2001). The enlarged tunnel could have then grown by capturing nearby conduits, especially when the lava started to entrench down through thermal erosion, forming a much larger tube. The inflation directions of the lateral spreading were governed by topographical variations (i.e., slopes) and temporal and spatial variations in the volume rate of flow alternatively favoring different branches then conveyed to the main tube, as a result of consecutive downward erosion (Kempe, 2019). Once the pyroclastic layer was consumed, the new main tube continued its expansion through thermo-mechanical erosion by complete or partial melting and assimilation of the pre-existing substrate constituted by the upper and lower units as proposed in the pioneering work in Lanzarote by Bravo, (1964); in the Hawaiian context by Kempe et al. (2010), and Bauer et al. (2013) and in a komatiitic lava context by Williams et al. (200120042011).

Bearing in mind the morphological differences between upper and lower units (see Section 4.1), we suppose that the enlargement mechanisms may have worked differently in the erosion of the different units. The less massive and more vesicular nature (avg. 6%) of the upper units made them more easily subject to mechanical erosion processes, in particular during sustained flow events where the conduits were under pressure flow conditions. While the low topographic gradients (slopes of ca. 2–3°) which characterize the north-eastern region of the island both nowadays and in the past contributed to reducing the flow rate, favoring a prolonged thermal contact between the overriding molten lava flow and the cold basaltic substrate of the lower units beneath. The transfer of heat into the substrate further reduces the hardness of the material already reduced by weathering, which would have encouraged its partial melting with consequent pyroduct downcutting below the pyroclastic layer (Bravo, 1964; Kauahikaua et al., 1998; Keszthelyi and Self, 1998; Shick, 2017; Williams et al., 200120042011). This long-lasting contact between hot and cold magmatic material together with the local turbulence of the flow induced by topographic irregularities of the tube floor and walls, would have promoted a combined effect of thermal and mechanical erosion of the pre-existing lava sheets (Greeley et al., 1998; Jarvis, 1995; Williams et al., 200120042011).

The downcutting along the La Corona lava tube increases downstream. In section 2 (Jameo de la Gente—Puerta Falsa; Figure 3 and Plate 2 in Supporting Information S1), the tube has a maximum of two superposed levels, and its total height does not exceed 20 m, whereas in Section 3 (Jameo de la Puerta Falsa—Jameo de los Verdes; Figure 3 and Plate 3 in Supporting Information S1), the tunnel is entrenched in the form of a canyon with depths reaching up to ∼40 m, and is composed of up to three superposed levels.Proceeding downstream the red pyroclastic layer is exposed at increasingly high levels on the tube walls with respect to its floor. The entrenchment progression was probably controlled by a process of knickpoint retreat in a similar process to what happens in river beds, and already proposed by Greeley et al. (1998), Kauahikaua et al. (1998), Keszthelyi and Self, (1998) and Kempe, (2012) for other lava tube systems. During the progressive drainage of the enlarged pyroduct, the flow activity was confined to the base of the cave. Consequently, the flow ledge accretion can lead to the formation of a second ceiling, isolating the lava flow to the inferior sector of the canyon and giving rise to the multi-level structure (e.g., Shick, 2017) clearly visible in section 2 (Jameo de la Gente—Puerta Falsa) and Section 3 (Jameo de la Puerta Falsa—Jameo de los Verdes).

The combination of these processes has likely allowed the La Corona system to reach the considerable dimensions and the multi-level structure that it has today.

6 Conclusions

The genetic processes that led to the formation of the La Corona lava tube system were mainly driven by an initial inflation process and the subsequent capture of different flows along a main drainage conduit. A lapilli layer occurring at the earliest stage of La Corona volcanic activity seems to play a pivotal role in the establishment of an efficient draining tube system. This layer outcrops in several locations at the surface as a widespread deposit of dark lapilli, frequently visible within the tube as a reddish pyroclastic horizon where the breakdown of the lining walls exposes the pre-existing flows.

The three-dimensional reconstruction of the surface covered by lapilli shows that thelapilli run along almost the entire length of the pyroduct. This leads us to suppose that the incoherent lapilli horizon has favored the excavation process of the molten lava, which in exploiting this weakness began to establish the lava tunnel itself (Figure 11). In particular surface observations have shown that the inflationary process began at a sinkhole conveying lavas at the base of the main edifice. Whitin a few hundred meters of the sinkhole, the underground hot lava was able to reach the pyroclastic layer underneath the pre-existing lava flows, allowing the tube system to be formed (Figure 12). The stream of lava ends up coalescing into a single large tube, as a result of consecutive downward erosion and the capture of nearby tubelets.

Details are in the caption following the image

Stages of La Corona volcano eruption. (a) Volcanic products of the initial Strombolian event covering the area around the volcano. (b) Effusive event C1 overlies the pyroclastic deposit. (c) Opening of a secondary vent on the eastern side of the volcano. (d) Formation of a lava pond at the base of the volcano the hot lava starts to sink through the pre-existing lava flows down to the pyroclastic layer. (e) Lava exploits the weakness of the pyroclastic level starting the excavation process. (f) Post-cooling phase, the tube is drained. Collapses occur obstructing the access to the tube.

Geochemical analyses performed on aphyric basaltic flows from above and below the lapilli layer show that the lower units are of Los Helechos affinity, while the upper units are compositionally similar to La Corona flows. The samples were collected at distances of approximately ∼1.5 km (SUB1), ∼2.3 km (PL1, PC1) and ∼2.6 km (LPUB1, LPLB1, LPCV1) from the sinkhole (Figures 2 and 6). Hence the tube developed for a considerable distance exclusively within the Quaternary age lava flow successions.

The gentle paleo-topography of the north-eastern part of the island at the time of the tube formation favored a prolonged contact between pre-existing lava flows and the molten lava flowing through them. This prolonged contact led to the melting of the tube bed while knickpoints developed at slope changes favored processes of thermo-mechanical erosion and so promoted the downcutting of the tube footwall.

To summarize, we infer that multiple causes operate in the placement and evolution of a large scale inflated lava tube. Through a combination of field work, three-dimensional reconstruction and geochemical analysis, it was possible to determine that the La Corona lava tube was emplaced among Quaternary lava flows generated during Los Helechos and La Corona activity and that the development of such huge dimensions was due to a combination of several concurrent conditions. The presence of a weak pyroclastic layer between pre-existing piled up lava flows facilitated the incipient inflation of lava in small sized tubes, whereas the gentle slope of the paleo-topography and the presence of knickpoints favored lava tube entrenchment, the formation of tube-in-tube structures and the breakdown of the ceiling or the floor of different tube levels forming a unique tube section. All of these findings and parameters will be useful for future studies of pyroducts, not only on Earth, but also on other bodies in the Solar System such as the Moon and Mars where lava tubes have been widely documented and layers of weak material (regolith or soils) in-between lava flows are thought to be common.

Acknowledgments

We are grateful to the Cabildo de Lanzarote and the Geopark of Lanzarote and Archipelago Chinijo for granting permissiton to access and collect samples in the La Corona lava tube system and to the Spanish Geological Service for providing us with Lanzarote's DTM. ESA-PANGEA partially supported this project with field activity and samples collection whereas GMAP-EPN2024 provided the tools for the digital geological mapping. LEICA geosystem provided the laser scan data of some sections of the tube realized through mobile mapping system during the PANGAEA-X 2017 campaign. We are grateful to Christophe Cloquet (SARM) for having performed the major and trace element analyses of our samples. We are grateful to Kaj Hoernle and Craig Lundstrom for sharing with us the locations of their samples, to Robbie Shone for his wonderful photos, and to Leonardo Tauro e Nicola Michelon for their technical support. We acknowledge Laurens Smets for the useful discussions on the Corona speleological survey and Prof. Javier Lario for the discussions on the Atlantida tunnel morphologies. We acknowledge the careful English and scientific review by Dr Simon Crowhurst, from the Department of Geosciences at Cambridge University. Finally, we acknowledge the two reviewers, David A. Williams and Laszlo P. Keszthelyi, for the really useful suggestions and discussions. Open Access Funding provided by Universita degli Studi di Padova within the CRUI-CARE Agreement.

    Data Availability Statement

    The datasets analyzed during the current study are provided as tables within the paper or within the accompanying online material file. In addition, the full geochemical data set, coordinates of samples and surface LIDAR are available open access at this repository link: https://doi.org/10.5281/zenodo.6572985. The full 3D model data set of the La Corona lava tube is property of the Virtual Geographic Agency (VIGEA) and is available under restricted access at the following repository link: https://doi.org/10.5281/zenodo.6573250. In order to protect the intellectual and commercial property of the La Corona lava tube 3D models, the entire data set will be made accessible through the request access form in Zenodo. Access will be granted when the requester will state in written that the data set will be used only for scientific research, excluding any commercial use.

    Erratum

    The original version of this article incorrectly stated that Figure 4 was modified from a figure by Franjo Sànchez; the correct attribution for the modified figure is Javier Lario, “Proyecto Sublantida.” This error has been corrected, and this may be considered the official version of record.