Dyke apertures record stress accumulation during sustained volcanism

Abstract

The feedback between dyke and sill intrusions and the evolution of stresses within volcanic systems is poorly understood, despite its importance for magma transport and volcano instability. Long-lived ocean island volcanoes are crosscut by thousands of dykes, which must be accommodated through a combination of flank slip and visco-elastic deformation. Flank slip is dominant in some volcanoes (e.g., Kilauea), but how intrusions are accommodated in other volcanic systems remains unknown. Here we apply digital mapping techniques to collect > 400,000 orientation and aperture measurements from 519 sheet intrusions within Volcán Taburiente (La Palma, Canary Islands, Spain) and investigate their emplacement and accommodation. We show that vertically ascending dykes were deflected to propagate laterally as they approached the surface of the volcano, forming a radial dyke swarm, and propose a visco-elastic model for their accommodation. Our model reproduces the measured dyke-aperture distribution and predicts that stress accumulates within densely intruded regions of the volcano, blocking subsequent dykes and causing eruptive activity to migrate. These results have significant implications for the organisation of magma transport within volcanic edifices, and the evolution and stability of long-lived volcanic systems.

Full text



Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports
Dyke apertures record stress
accumulation during sustained
volcanism
Samuel T. Thiele1,2*, Alexander R. Cruden1, Steven Micklethwaite1, Andrew P. Bunger3,4 &
Jonas Köpping1
The feedback between dyke and sill intrusions and the evolution of stresses within volcanic systems
is poorly understood, despite its importance for magma transport and volcano instability. Long-lived
ocean island volcanoes are crosscut by thousands of dykes, which must be accommodated through
a combination of ank slip and visco-elastic deformation. Flank slip is dominant in some volcanoes
(e.g., Kilauea), but how intrusions are accommodated in other volcanic systems remains unknown.
Here we apply digital mapping techniques to collect > 400,000 orientation and aperture measurements
from 519 sheet intrusions within Volcán Taburiente (La Palma, Canary Islands, Spain) and investigate
their emplacement and accommodation. We show that vertically ascending dykes were deected
to propagate laterally as they approached the surface of the volcano, forming a radial dyke swarm,
and propose a visco-elastic model for their accommodation. Our model reproduces the measured
dyke-aperture distribution and predicts that stress accumulates within densely intruded regions of
the volcano, blocking subsequent dykes and causing eruptive activity to migrate. These results have
signicant implications for the organisation of magma transport within volcanic edices, and the
evolution and stability of long-lived volcanic systems.
Magma plumbing systems comprise temporally and spatially interconnected dykes, sills and magma chambers
that exert fundamental controls on volcanic behaviour1. ey inuence and reect edice processes, and so
provide a record of volcano dynamics and long-term evolution that is essential for the development of predictive
models. Dierent volcanoes exhibit a range of magma plumbing styles due to variations in geological setting.
For example, the mechanical properties of the volcanic basement, and exure of the underlying lithosphere,
can exert a rst-order control on stresses within and beneath volcanic edices, and hence the organisation of
their magma plumbing systems2,3. Similarly, magma plumbing systems have been suggested to be inuenced
by topographic stresses4–6, upli due to the emplacement of magma at depth7,8, edice instability3,9,10, remote
tectonic stresses11,12 and basement structures13.
In this contribution we investigate the mechanisms that control and accommodate dyke injections in the shal-
low plumbing system of Volcán Taburiente, which forms the northern part of La Palma (Canary Islands; Fig.1).
is area is of interest as many thousands of intrusions1,11,14 are exceptionally well exposed along spectacular
cli sections, and because these intrusions may record the self-organisation of a radial magma plumbing system
into a discrete ri zone15. e eventual collapse of the edice aer ca. 400 kyr of activity16 could also be linked
to this plumbing system reorganisation.
Geological setting. e Canary Islands are constructed on slow moving (< 2cm/year) Jurassic oceanic
crust west of the coast of Morocco. As a result, volcanic islands in this region tend to be long-lived compared
to e.g., Hawaiian volcanoes, due to lower plate-velocities, and they undergo limited subsidence because of the
stiness of the underlying oceanic crust17. e intraplate setting of the Canary Islands and their distance from
Atlantic mid-ocean ridge spreading centres also results in low horizontal dierential stresses compared to other
tectonic settings11,18.
OPEN
           Helmholtz
        
Germany.        
USA.            
USA. *email: sam.thiele@monash.edu
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
It is broadly accepted that magmatism at La Palma (and the other Canary Islands) results from an upwelling
mantle plume that triggers mantle melting through decompression and the addition of heat17. However, magma
production rates are much lower than at other hotspots17,19, and as a result many of the islands (including La
Palma) do not appear to host long-lived shallow magma reservoirs. is hypothesis is supported by pyroxene
phenocryst and melt inclusion geobarometry20,21 and geophysical observations from eruptions at the neighbour-
ing El Hierro in 201122.
Instead, eruptions on La Palma appear to be fed by magma that accumulated and crystallised pyroxene and
olivine within deep, long-lived reservoirs in the upper mantle (~ 20–30 + km depth) before being emplaced for a
much shorter period of time (weeks to months) in the uppermost mantle or lower crust (10–15km depth), and
nally migrating upwards along rapidly propagating dykes to erupt (Fig.1).
As is typical of ocean island volcanoes, each of the Canary Islands is constructed from several distinct volcanic
edices. La Palma consists of an uplied Pliocene seamount (~ 4–2Ma) overlain by a succession of sub-aerial
edices: Garaa (~ 2–1.2Ma), Taburiente (1.2–0.56Ma), Bejenado (0.56–0.49Ma), and Cumbre Vieja (0.56Ma
to present)16. Lavas and pyroclastic products from Volcán Taburiente cover most of the north of the island, with
Garaa and seamount complex rocks exposed only along the bottom of deeply incised ravines. e southern part
of the island is formed by the currently active Cumbre Vieja volcano, a N–S oriented ridge lined with Holocene
scoria cones and lava ows16.
Each of these volcanic edices are separated by signicant erosional unconformities, generally related to large
edice collapse events15. e Garaa edice is thought to have collapsed to the south-west at ~ 1.2Ma, forming
a large depression that was rapidly lled by Volcán Taburiente11,17. From ~ 0.8Ma Taburiente volcanism appears
to have migrated southwards, extending the edice’s southern ank and forming an elongate, N-S oriented
ridge. is ridge collapsed towards the west at ~ 0.56Ma, aer which post-collapse volcanism rapidly formed
the Bejenado edice16. However volcanism continued to migrate southwards, leaving the collapse escarpment
(now dened by the Cumbre Nueva ridge15) unlled and forming the currently active Cumbre Vieja edice in
the south of the island.
Dykes in the shallow parts of the Taburiente edice have radial orientations1,11,14, indicating that the least
compressive stress σ3 is oriented circumferentially around a central point. Circumferential orientations of σ3
can be caused by a source of pressure at the swarm centre (e.g., a magma chamber) or by a radially decreasing
topographic load (as is typically imposed by a volcanic edice), although topographic stresses tend to dominate
at shallow depths23. Similar radial dyke swarms have been reported from Mt Somma/Vesuvio (Italy)24, Summer
Coon (USA)25, Oki-Dozen (Japan)26 and Lyttleton (New Zealand)27.
While dykes that are injected into elongate ri zones such as those in Hawaii can be accommodated by lateral
ank displacement28, the accommodation mechanism for radial swarms is less clear. With this in mind, we pre-
sent data on the orientation and thickness of dykes within the Volcán Taburiente, quantify the bulk-strain they
induced, and consider their inuence on intra-edice stresses and the evolution of the magma plumbing system.
Results
Orientation and thickness. Erosion of the Cumbre Nueva collapse scarp has incised deep into Volcán
Taburiente to form an arcuate series of ~ 1km high clis known as Caldera Taburiente15,29 (Fig.1). We have
taken advantage of this landscape to map the spectacularly exposed shallow magma plumbing system in unprec-
edented detail using 14 unmanned aerial vehicle (UAV) surveys conducted over ~ 2–50 hectare areas. Emerging
Figure1. Location and prominent topographic features of La Palma, Canary Islands. Schematic cross section
A–A′ through Volcán Taburiente is based on various geobarometry studies20,21. Deection of dykes in the upper
parts of the edice to propagate laterally is inferred from results of this study. Map created using QGIS 2.18
(https ://www.qgis.org).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
three-dimensional (3-D) digital mapping techniques30,31 were then applied to extract > 400,000 orientation and
thickness measurements from 519 sheet intrusions.
As expected, these measurements highlight the generally radial orientations of the intrusions. We constrain
the focal point of the radial dyke swarm using strike measurements and a maximum likelihood estimator (see
Supplementary Method), which delineates an area in the southern part of Caldera Taburiente ~ 1.5km north of
Bejenado (see Supplementary Fig.S1). Surveys from the northern side of the Caldera (Las Pareditas, Risco Liso,
Hoyo Verde and Los Cantos) also suggest a population of thicker and somewhat shallower dipping intrusions
striking NW and crosscut by the radial dykes (see Supplementary Figs.S1, S2).
Many sheet intrusions in Caldera Taburiente have geometries that suggest they propagated laterally. ese
include basal terminations, step-overs and broken bridge structures with shallow dipping axes (see Supplemen-
tary Fig.S2). Field observations of stretched vesicles and striated chilled margins associated with strongly aligned
plagioclase ow fabrics32 indicate ow lineations that plunge gently (0°–40°) both towards and away from the
dyke swarm focal point (see Supplementary Fig.S1). As such, we interpret that dykes ascending vertically from
below Volcán Taburiente were deected to propagate laterally and radially as they approached the surface (Fig.1).
A large range of thicknesses were measured for both dykes (dip > 45°) and inclined sheets (dips < 45°, includ-
ing true sills). Measurements from dykes have a mode thickness of ~ 0.6m and a long-tailed distribution, extend-
ing to ~ 5m (Fig.2a). To avoid biases due to measurements near dyke tips, a ‘maximum aperture’ dataset was
created by removing measurements below the 75th percentile and those above the 90th percentile (assumed to
reect erroneous measurements or abnormally thick dyke sections). For clarity, we use the term ‘thickness’ to
describe the collection of all measurements and ‘aperture’ to refer to the estimates of each intrusion’s maximum
thickness. e aperture measurements have a mode of ~ 1m and a long-tailed distribution similar to the thick-
ness data (Fig.2a).
e thickness and aperture of inclined sheets are similar to the dykes, although they tend towards higher
values (Fig.2a) with modes of ~ 0.6 and 1.25m respectively. Unlike the dykes, however, the aperture distribution
of inclined sheets contains anomalously thick subpopulations (> 30m thick, e.g. Supplementary Fig.S2). is
irregularity could be because fewer inclined sheets were sampled, but might also suggest that the apertures of
sills and other shallow-dipping intrusions are limited by dierent factors to dykes.
A striking relationship can be observed between intrusion altitude, a rough proxy for its depth of formation,
and thickness. Below ~ 1400m above sea level, the intrusions have a median thickness of ~ 1m, but by ~ 1800m
this has increased to nearly 2m, where it appears to plateau (Fig.2b). Based on a paleo-altitude estimate
of ~ 2500m around the rim of present-day Caldera Taburiente (the current elevation of its highest point), and
assuming that most of the dykes formed when the edice was similar to its current size, this region of increasing
dyke thickness corresponds to below surface paleo-depths of ~ 1100 to 700m. Above ~ 1800m altitude (< 700m
paleo-depth) the frequency of > 2.5m thick dykes also appears to increase (Fig.2b).
Excess magma pressure. e ratio between dyke aperture and length has been used by several authors to
estimate excess pressure driving dyke emplacement26,33–35, where excess pressure is the dierence between total
magma pressure and the normal stress resisting initial dyke opening. We have applied this method to our dataset
by selecting 52 dykes that are exposed in their entirety in the UAV surveys, such that their tip-to-tip ‘span’ could
be measured. Assuming the dykes propagated sub-horizontally, as suggested previously, these span measure-
ments were projected onto a vertical plane to give measurements of the dyke height (h). Excess pressure (Pexcess)
was then estimated using the aperture (a) measurements; assuming that the dykes are much longer than they are
Figure2. Kernel density estimates (a) describing the thickness and aperture (dened as the 75th to 90th
percentiles of thickness measurements for each dyke) of all dykes (i, ii) and inclined sheets (iii, iv). ickness
measurements were also divided into 50m high bins (b) based on their elevation and the interquartile range
(red lines), median (black line) and a kernel density estimate of the underlying distribution (grey) plotted. is
shows a clear increase in thickness between ~ 1400 and 1800m above sea level. No clear trend in thickness can
be seen above this altitude, although unusually thick dykes appear to become more common.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
high and vertical pressure variations are negligible, Pexcess can be related to a and h using the plane-strain solution
to a pressurised elliptical crack33,35:
is excess pressure estimate will vary proportionally to the elastic properties of the volcanic edice. e
bulk Young’s modulus (E) is rather dicult to ascertain, and in this context presumably depends on the ratio
of compliant pyroclastic material to sti basalt ows. Values of 1 to 5 GPa have been used by previous authors
for studies on basaltic volcanoes25,33,34, and hence we evaluate excess pressure over this range. Poisson’s ratio (v)
was kept xed at 0.25.
e results (Fig.3) show substantial variation in estimated excess pressure, probably because (1) propagation
directions may not have been strictly horizontal, meaning the dyke ‘height’ will be under- or over-estimated, (2)
solidied apertures may not represent the fracture aperture during dyke propagation, and (3) a range of actual
magma pressure is likely. Regardless of these limitations, the results indicate excess pressures of ~ 10 to 60MPa,
similar to estimates from other studies25,33,34.
Induced strain. e continuity of exposure in Caldera Taburiente allowed us to estimate the vertical and
tangential strain induced by the intrusions in Volcán Taburiente, using a cylindrical coordinate system with its
origin at the previously described maximum likelihood radial centre. Tangential strain estimates range between
1 and 10%, with maximum values observed in the north of the caldera (Fig.4a). An increased number of intru-
sions in this area is noticeable both in the eld and in aerial imagery, suggesting that these results are reasonable.
Vertical strain (Fig.4b) follows a similar but weaker trend, suggesting relatively modest endogenous growth
from shallow intrusions (1–2%). e large spike in the 90th percentile of estimated vertical strain at the Hoyo
Verde site results from a single very thick (> 30m) inclined sheet (Supplementary Fig.S2), and is probably only
a local eect.
Strain accommodation mechanisms. is strain must have been accommodated by a combination
of (1) accumulated elastic stresses in the edice, (2) ductile and plastic deformation such as compaction and
faulting, and/or (3) ank movement along a basal detachment or ductile layer. Mechanisms (1) and (2) can be
combined in a Maxwell visco-elastic model that estimates the edice stresses that result from intrusions in the
absence of a basal detachment and associated ank slip.
First, we dene a polar coordinate system with r and θ corresponding to the radial distance and rotation
(strike direction) relative to the focal point of the dykes, respectively (Supplementary Fig.S3). Assuming that
radial and vertical stresses are transmitted to the volcano’s free surface (σr = σz = 0), stress caused by the injection
of the radial dykes (σθ) will be uniaxial in the circumferential direction. To avoid unreasonably large strains near
the centre of the dyke swarm, we also assume that ascending dykes are deected towards areas of lowest stress and
so, as an ensemble, induce approximately constant circumferential stress. us, the rate of dyke-injections (fd)
varies in proportion with 2πr such that the rate of circumferential strain (εθ) is spatially uniform (Supplementary
Fig.S3). is is convenient as it makes our choice of r arbitrary (we use 5km). Similar accommodation-stress
induced migration of volcanism has been demonstrated by Derrien and Taisne36, who simulated repeated dyke
injection into an analogue volcano.
Based on these assumptions, we can relate εθ with σθ and the edice’s bulk properties (shear modulus G and
viscosity µ) using the Maxwell visco-elastic constitutive equation (see Supplementary Method for derivation):
(1)
P
excess =
aE
2h(1−v
2
)
Figure3. Height and aperture of the 52 dykes for which measurements could be obtained (a), coloured using
overpressure estimates calculated using Eq.(1) and a Young’s modulus (E) and Poisson’s ratio (v) of 2.5 GPa and
0.25 respectively. e top 25% of estimates were considered to be outliers (unlled circles) as they result from
dykes with unusually small length-aperture ratios. Kernel density estimates of the calculated excess pressure
using dierent Young’s moduli highlight signicant uncertainty and/or variation (b), although they suggest that
typical values between 10 and 60MPa seems plausible. Note that outliers (top 25% of calculated excess pressure)
were excluded during the kernel density estimation.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
Strain rate εθ can be related to the frequency fd at which dykes intrude across a given circumference (dened
by radius r) and their aperture a, rewriting Eq.(2) as:
Using Eq.(1), we can also relate dyke apertures to the excess magma pressure, which we here dene as some
initial excess pressure P0 minus the accumulated accommodation stresses σθ. Substituting into Eq.(3) and re-
arranging into a conventional form for a rst-order, linear ordinary dierential equation we get:
noting that:
is dierential equation (Eq.(6)) can be solved analytically to give the accommodation stress as a function
of time t (assuming that the initial σθ = 0), namely:
Values for G and v and µ are somewhat constrained by the literature25,33,34, although uncertainties exist due
to their scale-dependence and the highly fractured and heterolithic nature of volcanic materials. Based on the
abundance of compliant breccia and pyroclastic materials in Volcán Taburiente, we have chosen to use a Young’s
modulus of 2 GPa and Poisson’s ratio of 0.25, which corresponds to a shear modulus of 0.8 GPa. Viscosity is
less well constrained, but is generally thought to be on the order of 1022–1023 Pas for basaltic rocks at shallow
depths and low temperatures9. We tried a range of µ between 1 × 1022 and 5 × 1022 Pas (Supplementary Fig.S3).
e model is insensitive to Poisson’s ratio.
e remaining variables in Eq.(1) (P0, fd and h) can be related directly to the observed dyke aperture distri-
bution and nal bulk strain. P0 is related to the largest observed aperture using Eq.(1), and given we are using
E = 2 GPa, our eld observations (Fig.3) suggest that it should be 60–70MPa. All three terms can also be related
directly to the observed nal strain, the accommodation stress asymptote, and hence the mode of the aperture
distribution (see the Supplementary Method for more details). us, we can estimate these terms by tting them
to our measurements of dyke aperture and spacing (see Supplementary Fig.S3).
Finally, we perturbed the overpressure at each timestep (using a numerical solution to Eq.(4)) such that
it followed a normal distribution with a mean of P0 and a standard deviation that was optimised (along with
(2)
˙
ε
θ=
1
2G
˙σθ+
1
µ
σ
θ
(3)
fd
a
2πr
=
1
2G
˙σθ+
1
µ
σ
θ
(4)
˙
σ
θ+2G

1
µ
+k

σθ=2GkP0
,
(5)
k
=fd
2h(1−v
2
)
2πrE .
(6)
σ
θ=k

P0
1
µ
+k−
P0
1
µ
+k
e−2G1
µ+kt
.
Figure4. Map showing the median (solid line), 10th and 90th percentiles (dotted lines) of tangential (a) and
vertical (b) strain estimated using 1m spaced scan lines extracted from each digital outcrop model. ese show
a general tangential strain of ~ 2–4% through much of the caldera and substantially higher values (6–10%) in the
north. A similar pattern is observed for vertical strain, although the net vertical extension is much less (~ 1–2%).
Data from Los Andenes has been omitted from this analysis due to the limited number of intrusions at that site.
Note the dierence in the strain plot scale between (a,b).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
P0, fd and h, as previously described) to t the observed aperture distribution. Overpressure was modelled as a
stochastic variable to account for natural variations in the yield stress at which the source magma chamber fails
during each dyking event37 and variations in magma buoyancy and viscosity.
e estimated apertures (Fig.5) show a close t to our observations, and the optimised parameters are geo-
logically reasonable with the exception of dyke height (53m), which is signicantly less than we would expect.
Using an elastic modulus of 2 GPa in Eq.(1) and excess pressures of 65–80MPa, > 100m high dykes would be
10s of metres thick, which has not been observed.
is discrepancy can be explained if h instead represents an eective elastic height, which is signicantly less
than the actual height of large dykes due to (1) partial closure prior to solidication, and (2) bonding between
either side of the fracture due to unbroken rock bridges (which are commonly observed on La Palma and else-
where as ‘step-overs’; Supplementary Fig.S2). Unbroken rock-bridges tying together nominal fracture surfaces
are commonly observed in laboratory experiments examining uid-driven rock fracture; a striking example is
shown in Fig.24 of Hampton etal.38. However, the quantitative impact of these rock-bridges on fracture aperture
remains an unresolved topic of ongoing research.
Discussion
Because petrological constraints indicate magma is not stored at shallow depths below La Palma20,21, we suggest
the Taburiente dykes are radial because of topographic stresses rather than a shallow pressure source. e tempo-
ral progression of dyke orientations from NW striking towards radial observed from deeper parts of the edice
(Supplementary Fig.S1) is also consistent with deection due to increasing edice load. Early intrusions would
have formed in a NW orientation because of a combination of NNW-oriented regional maximum horizontal
stress, post-collapse stress modication39 and topography associated with the Garaa volcano. As the Taburiente
edice grew (probably quite rapidly16), topographic stresses progressively came to dominate regional stresses,
favouring more radial dyke orientations.
Due to the topographic source of radial stress in Volcán Taburiente, dykes need not have propagated away
from the dyke swarm focal point. Indeed, this is unlikely given, (1) the lack of evidence for shallow storage,
and (2) the variable plunge-directions of observed ow indicators. Instead, we suggest that dykes ascending
through the crust re-oriented into radial orientations as they entered the region below Volcán Taburiente where
topographic stresses became dominant. In our conceptual model (Fig.1), dykes propagate laterally as they are
deected away from regions of elevated topographic stress40, and possibly also due to decreasing buoyancy at shal-
low depths, forming blade-shaped intrusions (e.g. Supplementary Fig.S2). Similar interactions between ascend-
ing dykes and topographic loads have been suggested based on numerical models5 and eld observations12,25.
e inverse correlation between dyke thickness and depth results from a shi in the balance between conn-
ing pressure (resisting fracture opening) and internal magma pressure, although a decrease in host-rock elastic
modulus at shallow depths could also play a role. We suggest that the distinct range of paleo-depths (1.1–0.7km)
over which the increase in aperture occurs could be explained by the transition from supercritical to gaseous H2O
at pressures of ~ 22 MPa41 (depths ~ 750–900m). It is also plausible that less overpressured dykes are arrested or
deected before they reach the upper levels of the edice, and hence shallower dykes tend to be thicker.
Based on the distribution of preserved eruptive centres, it has been suggested16 that the northern anks of
Volcán Taburiente were formed by two diuse ri-zones trending NE and NW. Including the collapsed Cumbre
Nueva ridge in the south, this would give the Taburiente edice a ‘three-pointed-star’ type geometry that has
long been proposed as typical for the Canary Islands7. However, based on our strain estimates (Fig.4), we nd
no evidence for a concentration of dykes along the NE or NW margins of the Caldera.
Figure5. A numerical solution to the previously described Maxwell accommodation model using the elastic
parameters described in the text, randomly varying magma pressure sampled from a normal distribution, and
calibrated using Powell minimization to t observed aperture data. e optimised parameters (µ = 3.0 × 1022,
Fd = 2.1 days/ka, h = 53m and P = Ν(µ = 66, σ = 8.8) MPa) give results that closely match the observations. e
nal bulk strain of 5.5% and equilibrium excess pressure of 17.5MPa were not used to calibrate the model, and
match eld estimates (Figs.3, 4) excellently. Note, however, that this is probably not a unique solution and so
there may be other parameter combinations that reproduce the observed aperture data equally well.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
Instead, our data suggests that even though dykes in the Taburiente magma plumbing system are radial,
N–S orientations are favoured, causing denser dyke swarms and larger (~ 8% vs ~ 3%) extensional strain in the
Hoyo Verde and Los Cantos areas. Measurements of dykes crosscutting the seamount complex also suggest a
dominant N–S trend14. It seems plausible that this orientation was favoured from early in the growth of Volcán
Taburiente due to NNW–SSE oriented regional horizontal stress11 and topographic loads associated with the
Garaa edice and its subsequent collapse. Abundant proximal pyroclastic deposits exposed immediately above
the basal unconformity of Volcán Taburiente at Los Cantos support this hypothesis, as they suggest signicant
early volcanic activity in the north of the Caldera.
e location of the focal point of the radial dykes provides an additional constraint on the geometry of Volcán
Taburiente, as it should lie at the centre of the topographic stress eld and hence below the edice paleo-summit.
Its position in the south of Caldera Taburiente suggests a more elongate edice than is indicated by modern-day
topography. We therefore conclude that our dyke orientation measurements and the strain induced by their
emplacement are best explained by a somewhat elongated Taburiente edice. is topography would further
encourage the emplacement of N–S dykes4, helping to explain the strain localisation observed in the north of
Caldera Taburiente and the late-stage growth of the N–S oriented Cumbre Nueva ridge.
Combining these interpretations of Volcán Taburiente’s shallow magma plumbing system with our Maxwell
visco-elastic model for radial dyke accommodation, we propose that a combination of topographic and remote
tectonic stresses governed dyke-propagation paths, while dyke induced strain was accommodated visco-elasti-
cally. Accommodation stresses associated with this visco-elastic deformation will have caused stress to evolve
towards an isotropic state within heavily intruded portions of the edice. Subsequent dykes would have been
deected away from these regions of high stress6,36, causing volcanic activity to migrate. In our model (Fig.5)
the accommodation stress reaches a maximum at about the same time that eruptive activity began to focus on
the southern ank of Volcán Taburiente, forming the Cumbre Nueva ridge. We propose that this shi in activity
occurred as dykes were deected southwards by a ‘stress-plug’ formed near the centre of the radial dyke swarm
(below the summit of Volcán Taburiente), noting that escape to the north would have been blocked by stresses
related to the older Garaa edice16.
e transition from radial to focused dyking below the Cumbre Nueva ridge can thus be explained by accom-
modation stresses blocking dyke ascent below the summit of Volcán Taburiente and regional tectonic stress
favouring N–S oriented dykes. If widespread, this process of stress-plug development followed by lateral escape
may help to explain the apparent self-organisation of many volcanic systems into elongate volcanic ridges or ri
zones. e competition between stress plug development due to intrusive activity and counteracting topographic
stress changes due to eruptive activity and/or edice instability could also be a signicant control on the evolu-
tion of magma plumbing systems and distribution of associated volcanism.
e close t between eld measurements and the apertures predicted by our Maxwell visco-elastic model
highlight the inuence of accommodation stress on dyke apertures. Volcanic systems typically have long-tailed
dyke aperture distributions, which have previously been interpreted to reect the distribution of magma cham-
ber failure pressures33. Our results show that this need not be the case. Instead, the build-up of accommodation
stress can also cause long-tailed dyke aperture distributions, regardless of the distribution of source overpres-
sures. ese eects must be considered should aperture data be used to estimate source properties33,37,42, as these
methods typically assume lithostatic stress.
We conclude that the stresses induced by successive intrusions accumulate in volcanic edices and inuence
dyke propagation paths and apertures. e stress eld within heavily intruded regions may be signicantly dif-
ferent to that predicted by a lithostatic model, because dyke-induced deformation increases horizontal stress
magnitude and in doing so reduces the dierential stress. Stress plug development by this mechanism may be
an important and widespread control on the spatio-temporal distribution of volcanism in long-lived volcanic
systems.
Methods
Digital mapping. To circumvent limited access to steep and unstable exposures within Caldera Taburiente,
14 unmanned aerial vehicle (UAV) surveys were conducted over ~ 2–50 hectare areas. Survey sites were chosen
to capture a representative distribution of the available exposure within Caldera Taburiente and a range of dif-
ferent depths. Imagery was collected using a DJI Phantom 4 Pro and its integrated camera (20–megapixel CMOS
sensor) along horizontal ight lines ~ 30–60m from the cli faces using horizontal and ~ 30 degree downward
oriented viewing angles and vertical and horizontal overlaps of ~ 80%. ese image sets were then reconstructed
using a structure-from-motion multi-view-stereo (SfM-MVS) workow30,43 to create a database of 3-D digital
outcrop point cloud models at ground sampling distances of ~ 2–5cm. Specic details of each of the surveys
is included in the Supplementary Method and the dataset can be downloaded from https ://doi.org/10.26180
/5d688 c17f2 ed2.
e upper and lower surfaces of 519 intrusions within these models were then digitised from the high-reso-
lution 3-D point clouds using the Compass plugin in Cloud Compare31. Continuous exposure within the survey
areas means that all intrusions > 20cm thick could be identied, allowing the number and spacing of intrusions
in each survey area to be characterised. Structure-normal estimates (SNEs)30 constraining dyke orientation and
true thickness were also generated at each point along the intrusion contacts. ese were manually vetted to
remove invalid results in non-planar sections of the dyke or where data quality was poor30, aer which 12 million
SNEs from ~ 400,000 locations remained, covering ~ 60% of the 66km of digitised dyke margins.
Strain estimation. Dykes in each survey were divided into horizontal scan-lines spaced at intervals of 1m
vertically to investigate the bulk-strain associated with their emplacement. Assuming Mode I opening, the dila-
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
tion vector of each dyke along a scan line was calculated from the SNEs by multiplying the structure normal vec-
tor by the dyke thickness. ese opening vectors were summed along each scan line, and the resultant expressed
in radial and tangential components based on the maximum likelihood radial centre. e same method was used
to estimate vertical strain at each location, although in this case vertical scan lines were used and both dykes and
inclined sheets were included.
Visco-elastic modelling. Python code for our visco-elastic modelling and the generation of Supplemen-
tary Fig.S3 and Fig.5 is included in the Supplementary Method, along with the derivation of our Maxwell
constitutive equation (Eq.(2)).
Data availability
e digital outcrop models and associated structural measurements analysed during this study are available on
the FigShare repository, https ://doi.org/10.26180 /5d688 c17f2 ed2. e python code used to perform our analyses
is included in the Supplementary Information les.
Received: 1 November 2019; Accepted: 2 September 2020
References
1. Burchardt, S., Walter, T. R. & Tuen, H. Growth of a volcanic edice through plumbing system processes—Volcanic ri zones,
magmatic sheet-intrusion swarms and long-lived conduits. In Volcanic and Igneous Plumbing Systems (ed. Burchardt, S.) 89–112
(Elsevier, Amsterdam, 2018).
2. McGovern, P. J. et al. Lithospheric exure and volcano basal boundary conditions: Keys to the str uctural e volution of large volcanic
edices on the terrestrial planets. Geol. Soc. Lond. Spec. Publ. 401, 219–237 (2015).
3. van Wyk de Vries, B. & Matela, R. Styles of volcano-induced deformation: Numerical models of substratum exure, spreading and
extrusion. J. Volcanol. Geothermal Res. 81, 1–18 (1998).
4. Fiske, R. & Jackson, E. Orientation and growth of Hawaiian volcanic ris: e eect of regional structure and gravitational stresses.
Proc. R. Soc. Lond. A Math. Phys. Sci. 329, 299–326 (1972).
5. Pinel, V. & Jaupart, C. Magma storage and horizontal dyke injection beneath a volcanic edice. Earth Planet. Sci. Lett. 221, 245
(2004).
6. Kervyn, M., Ernst, G. G. J., van Wyk de Vries, B., Mathieu, L. & Jacobs, P. Volcano load control on dyke propagation and vent
distribution: Insights from analogue modeling. J. Geophys. Res. Solid Earth 114, 3 (2009).
7. Carracedo, J. e Canary islands: An example of structural control on the growth of large oceanic-island volcanoes. J. Volcanol.
Geoth. Res. 60, 225–241 (1994).
8. Klügel, A., Longpré, M.-A., García-Cañada, L. & Stix, J. Deep intrusions, lateral magma transport and related upli at ocean island
volcanoes. Earth Planet. Sci. Lett. 431, 140–149 (2015).
9. Cecchi, E., de Vries, B. V. W. & L avest, J.-M. Flank spreading and collapse of weak-cored volcanoes. Bull. Volcanol. 67, 72–91 (2004).
10. Münn, S., Walter, T. R. & Klügel, A. Gravitational spreading controls ri zones and ank instability on El Hierro, Canary Islands.
Geol. Mag. 143, 257–268 (2006).
11. Fernández, C., de la Nuez, J., Casillas, R. & García Navarro, E. Stress elds associated with the growth of a large shield volcano (La
Palma, Canary Islands). Tectonics 21, 11–18 (2002).
12. McGuire, W. J. & Pullen, A. D. Location and orientation of eruptive ssures and feederdykes at Mount Etna; inuence of gravita-
tional and regional tectonic stress regimes. J. Volcanol. Geothermal Res. 38, 325–344 (1989).
13. Ganey, E. S., Damjanac, B. & Valentine, G. A. Localization of volcanic activity: 2. Eects of pre-existing structure. Earth Planet.
Sci. Lett. 263, 323–338 (2007).
14. Staudigel, H., Feraud, G. & Giannerini, G. e history of intrusive activity on the island of La Palma (Canary Islands). J. Volcanol.
Geothermal Res. 27, 299–322 (1986).
15. Carracedo, J. C., Day, S. J., Guillou, H. & Gravestock, P. Later stages of volcanic evolution of La Palma, Canary Islands: Ri evolu-
tion, giant landslides, and the genesis of the Caldera de Taburiente. Geol. Soc. Am. Bull. 111, 755–768 (1999).
16. Carracedo, J. C., Badiola, E. R., Guillou, H., Nuez, J. & Pérez-Torrado, F. J. Geology of La Palma and El Hierro, Canary islands.
Estud. Geol. 57, 175–273 (2001).
17. Carracedo, J. C. Growth, structure, instability and collapse of Canarian volcanoes and comparisons with Hawaiian volcanoes. J.
Volcanol. Geothermal Res. 94, 1–19 (1999).
18. Geyer, A., Martí, J. & Villaseñor, A. First-order estimate of the Canary Islands plate-scale stress eld: Implications for volcanic
hazard assessment. Tectonophysics 679, 125–139 (2016).
19. Acosta, J. et al. Comparison of volcanic ris on La Palma and El Hierro, Canary Islands and the island of Hawaii. In Geophysics of
the Canary Islands (eds Cli, P. D. & Acosta, J. J.) 59–90 (Springer, New York, 2005).
20. Galipp, K., Klügel, A. & Hansteen, T. H. Changing depths of magma fractionation and stagnation during the evolution of an oceanic
island volcano: La Palma (Canary Islands). J. Volcanol. Geothermal Res. 155, 285–306 (2006).
21. Klügel, A., Hansteen, T. H. & Galipp, K. Magma storage and underplating beneath CumbreVieja volcano, La Palma (Canary
Islands). Earth Planet. Sci. Lett. 236, 211–226 (2005).
22. Martí, J., Villaseñor, A., Geyer, A., López, C. & Trygg vason, A. Stress barriers controlling lateral migration of magma revealed by
seismic tomography. Sci. Rep. 7, 40757 (2017).
23. Roman, A. & Jaupart, C. e impact of a volcanic edice on intrusive and eruptive activity. Earth Planet. Sci. Lett. 408, 1–8 (2014).
24. Acocella, V., Porreca, M., Neri, M., Massimi, E. & Mattei, M. Propagation of dikes at Vesuvio (Italy) and the eect of Mt. Somma.
Geophys. Res. Lett. 33, L08301 (2006).
25. Poland, M. P., Moats, W. P. & Fink, J. H. A model for radial dike emplacement in composite cones based on observations from
Summer Coon volcano, Colorado, USA. Bull. Volcanol. 70, 861–875 (2008).
26. Miura, D., Toshida, K., Arai, K.-I., Wachi, T. & Wada, Y. Oki-Dozen dike swarm: Eect of the regional stress eld on volcano-
tectonic orientations. In Evolutionary Models of Convergent Margins: Origin of eir Diversity (ed. Itoh, Y.) (InTech, Rijeka, 2017).
27. Shelley, D. Radial dikes of Lyttelton Volcano—eir structure, form, and petrography. N.Z. J. Geol. Geophys. 31, 65–75 (1988).
28. Denlinger, R. P. & Morgan, J. K. Instability of Hawaiian volcanoes. In Characteristics of Hawaiian Volcanoes (eds Poland, M. P. et
al.) 149–176 (U.S, Geological Survey, Reston, 2014).
29. Ancochea, E. et al. Constructive and destructive episodes in the building of a young oceanic island, La Palma, Canary Islands, and
genesis of the Caldera de Taburiente. J. Volcanol. Geothermal Res. 60, 243–262 (1994).
30. iele, S. T., Grose, L., Cui, T., Cruden, A. R. & Micklethwaite, S. Extraction of high-resolution structural orientations from digital
data: A Bayesian approach. J. Struct. Geol. 122, 106–115 (2019).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:17335 | 
www.nature.com/scientificreports/
31. iele, S. T. et al. Rapid, semi-automatic fracture and contact mapping for point clouds, images and geophysical data. Solid Earth
8, 1241–1253 (2017).
32. Holness, M. B. & Humphreys, M. C. S. e TraighBhànnaSgùrra Sill, Isle of Mull: Flow Localization in a Major Magma Conduit.
J. Petrol. 44, 1961–1976 (2003).
33. Becerril, L., Galindo, I., Gudmundsson, A. & Morales, J. M. Depth of origin of magma in eruptions. Sci. Rep. 3, 2762 (2013).
34. Kusumoto, S., Geshi, N. & Gudmundsson, A. Aspec t ratios and magma overpressures of non-feeder dikes observed in the Miyake-
jima volcano (Japan), and fracture toughness of its upper part. Geophys. Res. Lett. 40, 1065–1068 (2013).
35. Gudmundsson, A. Rock Fractures in Geological Processes (Cambridge University Press, Cambridge, 2011).
36. Derrien, A. & Taisne, B. 360 intrusions in a miniature volcano: Birth, growth, and evolution of an analog edice. Front. Ear th Sci.
7, 19 (2019).
37. Krumbholz, M. et al. Weibull-distributed dyke thickness reects probabilistic character of host-rock strength. Nat. Commun. 5,
3272 (2014).
38 . Hampton, J. C., Hu, D., Matzar, L. & Gutierrez, M. Cumulative volumetric deformation of a hydraulic fracture using acoustic emis-
sion and micro-CT imaging. in 48th U.S. Rock Mechanics/Geomechanics Symposium15 (American Rock Mechanics Association,
Minneapolis, 2014).
39. Maccaferri, F., Richter, N. & Walter, T. R. e eect of giant lateral collapses on magma pathways and the location of volcanism.
Nat. Commun. 8, 1097 (2017).
40. Urbani, S., Acocella, V., Rivalta, E. & Corbi, F. Propagation and arrest of dikes under topography: Models applied to the 2014
Bardarbunga (Iceland) riing event. Geophys. Res. Lett. 44, 6692–6701 (2017).
41 Cas, R. A. F. & Simmons, J. M. Why deep-water eruptions are so dierent from subaerial eruptions. Front. Earth Sci. https ://doi.
org/10.3389/feart .2018.00198 (2018).
42. Browning, J., Drymoni, K. & Gudmundsson, A. Forecasting magma-chamber rupture at Santorini volcano, Greece. Sci. Rep. 5,
15785 (2015).
43 . Dering, G., Micklethwaite, S., iele, S. T., Vollgger, S. A. & Cr uden, A. R. Review of drones, photogrammetry and emerging sensor
technology for the study of dykes: Best practises and future potential. J. Volcanol. Geoth. Res. 373, 148–166 (2019).
Acknowledgements
e authors gratefully acknowledge the sta at Parque Nacional Caldera de Taburiente for their generous support
and hospitality during collection of the eld data presented in this study. We would also like to thank JC Car-
racedo for a splendid introduction to the Canary Islands during the early stages of this work. ST was supported by
a Westpac Future Leaders Scholarship and Australian Postgraduate Award. APB wishes to acknowledge support
provided by the U.S. National Science Foundation under Grant No. 1645246. ARC is supported by Australian
Research Council Discovery Grant DP190102422. We also wish to acknowledge insightful and constructive
reviews by two anonymous reviewers.
Author contributions
S.T., S.M. and A.C. developed the ideas presented in this paper and collected the eld data. S.T., A.B. and J.K.
developed the Maxwell accommodation model. S.T. performed the analyses and wrote the python code to
produce Figs.2, 3, 4 and 5. All authors helped interpret the results and their implications, and contributed to
manuscript preparation and writing.
Competing interests
e authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https ://doi.org/10.1038/s4159 8-020-74361 -w.
Correspondence and requests for materials should be addressed to S.T.T.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.
© e Author(s) 2020
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com