Journal of Petrology | Volume 43 | Number 8 | Pages 1469-1503 | 2002
© Oxford University Press 2002
The Fish Canyon Magma Body, San Juan Volcanic Field, Colorado: Rejuvenation and Eruption of an Upper-Crustal Batholith
OLIVIER BACHMANN1,*,
MICHAEL A. DUNGAN1 and
PETER W. LIPMAN2
1SECTION DES SCIENCES DE LA TERRE DE L’UNIVERSITÉ DE GENÈVE, 13, RUE DES MARAÎCHERS, 1211 GENEVA 4, SWITZERLAND
2US GEOLOGICAL SURVEY, 345 MIDDLEFIELD ROAD, MENLO PARK, CA 94025, USA
Received
May 11, 2001;
Revised typescript accepted
January 22, 2002
 |
ABSTRACT
|
|---|
More than 5000 km
3 of nearly compositionally homogeneous crystal-rich
dacite (
68 wt % SiO
2:
45% Pl + Kfs + Qtz + Hbl + Bt + Spn +
Mag + Ilm + Ap + Zrn + Po) erupted from the Fish Canyon magma
body during three phases: (1) the pre-caldera Pagosa Peak Dacite
(an unusual poorly fragmented pyroclastic deposit,
200 km
3);
(2) the syn-collapse Fish Canyon Tuff (one of the largest known
ignimbrites,
5000 km
3); (3) the post-collapse Nutras Creek Dacite
(a volumetrically minor lava). The late evolution of the Fish
Canyon magma is characterized by rejuvenation of a near-solidus
upper-crustal intrusive body (mainly crystal mush) of batholithic
dimensions. The necessary thermal input was supplied by a shallow
intrusion of more mafic magma represented at the surface by
sparse andesitic enclaves in late-erupted Fish Canyon Tuff and
by the post-caldera Huerto Andesite. The solidified margins
of this intrusion are represented by holocrystalline xenoliths
with Fish Canyon mineralogy and mineral chemistry and widely
dispersed partially remelted polymineralic aggregates, but dehydration
melting was not an important mechanism in the rejuvenation of
the Fish Canyon magma. Underlying mafic magma may have evolved
H
2O–F–S–Cl-rich fluids that fluxed melting
in the overlying crystal mush. Manifestations of the late up-temperature
magma evolution are: (1) resorbed quartz, as well as feldspars
displaying a wide spectrum of textures indicative of both resorption
and growth, including Rapakivi textures and reverse growth zoning
(An
27–28 to An
32–33) at the margins of many plagioclase
phenocrysts; (2) high Sr, Ba, and Eu contents in the high-SiO
2 rhyolite matrix glass, which are inconsistent with extreme fractional
crystallization of feldspar; (3) oscillatory and reverse growth
zoning toward the margins of many euhedral hornblende phenocrysts
(rimward increases from
5·5–6 to 7·7–8·5
wt % Al
2O
3). Homogeneity in magma composition at the chamber-wide
scale, contrasting with extreme textural and chemical complexities
at the centimeter–millimeter scale, is consistent with
a dynamic environment, wherein crystals with a variety of growth
and resorption histories were juxtaposed shortly before eruption
by convective currents.
KEY WORDS: Monotonous Intermediates; ignimbrite; silicic magmatism; magma rejuvenation
|
INTRODUCTION
|
|---|
The Fish Canyon Tuff is the product of the largest documented
pyroclastic eruption (
Lipman et al., 1970;
Whitney & Stormer, 1985;
Lipman, 2000) and the archetypal example of a group of
voluminous silicic ignimbrites referred to by
Hildreth (1981) as the ‘Monotonous Intermediates’. These dominantly
dacitic magmas are important for the understanding of crustal
magmatism, not only because they are the largest known manifestations
of explosive volcanism on Earth (a few hundreds of km
3 to >5000
km
3 for a single eruption), but also because such ash-flow sheets
provide a link between the plutonic and volcanic realms. Indeed,
they resemble erupted batholiths in the sense that they have
comparable volumes, occur in the same tectonic settings, and
are characterized by high crystal contents (
40–50%), near-solidus
mineral assemblages, and a lack of evidence for compositional
zoning in pre-eruptive magma chambers (
Francis et al., 1989;
de Silva, 1991).
Some of the most controversial issues in the study of silicic magmatism are the thermal evolution and residence times of large upper-crustal magma bodies and whether silicic magmas are generated dominantly by crystal fractionation or by melting of pre-existing crustal material. In particular, the question of residence time of magmas in the upper crust has been debated (see Halliday, 1990; Mahood, 1990; Sparks et al., 1990; Reid et al., 1997; Brown & Fletcher, 1999; Reid & Coath, 2000), following the inference, based on Rb–Sr and 40Ar/39Ar data from the Long Valley system (Halliday et al., 1989; van den Bogaard & Schirnick, 1995), that upper-crustal silicic magma bodies may remain partly molten for periods >1 Myr. In acknowledgement of the difficulty of maintaining a large magma body near its liquidus in the upper crust for such a long period, an alternative model was proposed in which the magma chamber is episodically reheated, with periods of crystallization alternating with periods of partial remelting following new magma inputs; that is, ‘defrosting’ of a crystal mush (Mahood, 1990).
The present reinvestigation of the Fish Canyon magmatic system, which has been extensively studied in the past (Lipman et al., 1970, 1997; Whitney & Stormer, 1985; Johnson & Rutherford, 1989a; Riciputi et al., 1995), has led to a new model for its petrologic evolution. A rich catalog of previously unrecognized textural and geochemical features, which are well preserved in recently identified non-fragmented samples of Fish Canyon magma of the Pagosa Peak Dacite and Nutras Creek Dacite (Bachmann et al., 2000), serve as the basis for inferring that the Fish Canyon magma body was remobilized following a voluminous injection of mafic magma, which reheated and partially remelted an upper-crustal near-solidus crystal mush of batholithic dimensions. Similar models have been proposed for the Kos Ignimbrite (Keller, 1970), the Long Valley system (Mahood, 1990), and the southwest moat rhyolites of the Valles Caldera (Wolff & Gardner, 1995), as well as the continuing Soufrière Hills eruption on Montserrat (Murphy et al., 2000; Couch et al., 2001). The fact that these rhyolitic to andesitic eruptive products, with vastly different volumes, seem to have originated by the same mechanism may indicate that thermal rejuvenation of near-solidus upper-crustal intrusions shortly before eruption is a fairly common process in silicic systems. Moreover, this model does not require long upper-crustal residence times for largely molten magma bodies.
|
GEOLOGICAL SETTING
|
|---|
The San Juan volcanic field, located on the eastern margin of
the Colorado plateau in SW Colorado, covers an area of
25 000
km
2 with a total volume of
40 000 km
3 of intermediate to
silicic volcanic rocks. The San Juan volcanic field is the largest
erosional remnant of a nearly continuous igneous province that
extended over much of the southern Rocky Mountains during mid-Tertiary
time (
Steven, 1975). It lies at the northern tip of a succession
of Tertiary volcanic fields extending south to the Sierra Madre
Occidental, in present-day Mexico (Fig.
1). Although magmatic
activity in the Sierra Madre Occidental is plausibly related
to Oligocene subduction of the Pacific plate beneath North America,
the tectonic setting of the volcanic fields that are located
further to the north, several hundreds of kilometers inboard
from the plate margin, is more controversial. The arc-like geochemical
signature of San Juan magmas (e.g. high-K calc-alkaline series
with high La/Nb and Ba/La) has led to the interpretation that
low-angle subduction was associated with this magmatic activity
(
Lipman et al., 1972,
1978), but magma generation from lithospheric
mantle previously modified by subduction has been proposed as
an alternative (
Davis et al., 1993).
View larger version (63K):
[in this window]
[in a new window]
|
Fig. 1. Map of southwestern North America, emphasizing the alignment of mid-Tertiary volcanic fields extending from the apparently subduction-related magmatism of the Sierra Madre Occidental to the San Juan volcanic field, which is characterized by high-K calc-alkaline eruptive products with arc-like geochemical affinities, despite its unusually large distance from the western margin of the North American plate (>700 km). Modified from Davis et al. (1993).
|
|
Tertiary activity in the San Juan volcanic field began with a period of >4 Myr of andesitic magmatism (33·5–29·5 Ma) in the form of large stratovolcanoes (Lipman et al., 1970), which accumulated to thicknesses of 1–2 km across the entire area of the volcanic field. Voluminous explosive eruptions began before 29 Ma, and at least 17 large silicic ash-flow sheets (100–5000 km3) and related post-collapse lavas were erupted during 3 Myr from three nested caldera clusters in the west, central, and SE part of the field (Fig. 2; Steven & Lipman, 1976). Caldera-forming activity started in the west and SE caldera clusters and then converged on the central part of the field at 28·5 Ma. The Fish Canyon magmatic system belongs to a sequence of nine major ash-flow tuffs and related lavas erupted during 2·5 Myr (28·5–26 Ma) from the extremely productive central San Juan cluster, reaching a magma production rate of >4000 km3/Myr (Lipman et al., 1996). Andesitic activity was prevalent between caldera-related eruptions, as intermediate composition lavas and clast-rich laharic breccias are interlayered with the ignimbrites. San Juan magmatism ended with a Mio-Pliocene bimodal suite, the Hinsdale Formation, which was associated with crustal extension of the Rio Grande Rift (Lipman et al., 1970, 1978).
View larger version (77K):
[in this window]
[in a new window]
|
Fig. 2. Location map showing the preserved extent of the San Juan volcanic field and the approximate distribution of the outflow facies of the Fish Canyon Tuff (modified from Steven & Lipman, 1976). Ash-flow tuff volcanism in the San Juan volcanic field was concentrated in three caldera clusters. The source of the Fish Canyon eruptions is the La Garita caldera, the second (after buried source of Masonic Park Tuff) and largest of a series of eight calderas that form the central San Juan caldera cluster. The northernmost segment of the La Garita caldera was formerly referred to as the Cochetopa Park caldera (CP), but recent field work (by Lipman, in 1999–2000) has shown that this is a late, subsidiary collapse structure associated with the Fish Canyon Tuff eruption.
|
|
The changes in eruptive style and chemical composition recorded by the products of the San Juan volcanic field reflect progressive crustal hybridization and emplacement of a composite high-level batholith during long-term intracontinental magmatism (Riciputi et al., 1995). During the early stages, the eruptive products consisted mainly of andesitic lava flows, produced by differentiation of mantle-derived basalts involving both fractional crystallization and crustal assimilation (Colucci et al., 1991; Riciputi et al., 1995). Volumetrically minor units with more evolved compositions appear in some sections, but basaltic magma apparently did not reach the surface during Oligocene magmatism. The relatively abrupt appearance of silicic ash-flow sheets at 29–30 Ma, emplaced during large caldera-forming eruptions, marks the ascent of magma bodies to shallow levels and increased chemical contributions from the upper crust after several millions of years of crustal heating by injection of mantle-derived magmas. During the period dominated by explosive silicic volcanism, mantle input continued, as indicated by the andesitic units interlayered with the ignimbrites. This led to progressive hybridization of the crustal column within the main zone of magma transport and injection (Lipman et al., 1978; Riciputi et al., 1995). This long-term progression toward more evolved magma compositions and shallower magma storage has been interpreted as representing the emplacement and differentiation of a large composite batholith in the upper crust. This interpretation is corroborated by the presence of a large negative Bouguer gravity anomaly (down to -330 mgal) in the San Juan region (Plouff & Pakiser, 1972).
Recently, the eruptive history of the Fish Canyon magmatic system has been shown to be more complex than previously thought. The Fish Canyon magma chamber produced three separate but compositionally identical units with different eruptive styles in rapid succession 28 Myr ago (Lipman et al., 1997; Bachmann et al., 2000). The Fish Canyon Tuff, with a total volume of 5000 km3, was erupted during the collapse of the 100 km x 35 km La Garita caldera as a highly fragmented crystal-rich deposit containing only scarce small pumices. In contrast, the precursory Pagosa Peak Dacite eruption produced 200 km3 of poorly fragmented pyroclastic deposits, which are preserved around the southeastern edge of the La Garita caldera (Fig. 2). This unit, which was first recognized as a separate eruptive phase in 1995, and which is interpreted as the product of a low-energy fire-fountain eruption (Bachmann et al., 2000), contains large low-vesicularity pumices, up to 4 m in diameter, referred to as magma ‘blobs’. Unlike bulk Fish Canyon Tuff, the Pagosa Peak Dacite blobs closely approximate Fish Canyon magma compositions. Furthermore, they preserve magmatic textures, which have been largely destroyed in the Fish Canyon Tuff by shattering of phenocrysts (particularly feldspars) and dispersal of crystal fragments. The Nutras Creek Dacite is a small post-Fish Canyon lava flow (<1 km3) on the north flank of the resurgent dome (Fig. 2), which has also yielded non-fragmented samples of Fish Canyon magma.
|
PREVIOUS INVESTIGATIONS OF THE FISH CANYON SYSTEM
|
|---|
The regional extent of the Fish Canyon Tuff, its large volume,
distinctive petrologic character with respect to many ash-flow
sheets (uniform dacitic composition, high phenocryst content,
11-phase mineral assemblage), and relation to large-scale caldera
collapse were initially recognized more than 30 years ago (
Lipman et al., 1970;
Lipman, 1975;
Steven & Lipman, 1976).
Hildreth (1981) called attention to the anomalous character of voluminous
silicic units, which lack compositional zoning (Monotonous Intermediates),
in light of the prevailing view that most such magmas were derived
by differentiation from subjacent mafic parental magmas by crystal
fractionation. Such an origin was, and still is, thought to
produce chemical and thermal gradients or discontinuities in
shallow magma chambers. However, erupted volumes of homogeneous
silicic magma greater than a few thousand km
3 require extraordinarily
large volumes of parental mafic magma. In effect, the question
of how Monotonous Intermediate tuffs originate raises the same
issues as does the origin of batholithic granitoid intrusions.
The pre-eruptive state of the Fish Canyon magma chamber was investigated by Stormer & Whitney (1985) and Whitney & Stormer (1985), who concluded that this chamber lacked compositional and thermal gradients and that the phenocrysts had equilibrated at 7–9 kbar before rapid ascent to shallow levels. Both conclusions were based on whole-rock and coexisting phase compositions in bulk samples of the Fish Canyon Tuff sensu stricto, which is notable for: (1) an extreme scarcity of pumices (fiamme) sufficiently large to permit direct determinations of magma composition; (2) shattering and dispersion of phenocrysts and glassy matrix during eruption and emplacement; (3) nearly pervasive devitrification of the matrix. Following criticism of these conclusions by Grunder & Boden (1987), Johnson & Rutherford (1989a, 1989b) used Fish Canyon Tuff as a starting material to determine the P, T, XH2O, and fO2 at which the phenocryst assemblage equilibrated with residual melt and to experimentally calibrate the Al-in-hornblende geobarometer (Hammarstrom & Zen, 1986). Both methods yielded a pressure of 2·4 ± 0·5 kbar, strengthening arguments for upper-crustal residence and low-pressure equilibration of the Fish Canyon mineral assemblage, in accord with caldera collapse.
The large, easily extracted blobs of Fish Canyon magma in the Pagosa Peak Dacite and the recognition of the effusive origin of the Nutras Creek Dacite (Bachmann et al., 2000) provide samples of unmodified magma composition, which have allowed us to test the assertion that the pre-eruptive Fish Canyon magma chamber lacked thermal and chemical gradients (Whitney & Stormer, 1985). These samples preserve textural information critical to the formulation of our new model for the genesis and evolution of the Fish Canyon magma body. Because crystals were less broken during the relatively low-energy eruptions, numerous textural relationships that were not previously appreciated are evident (see Figs 10, 12, 13 and
15–18
,
below). In particular, resorption textures are not limited to quartz, but also are widespread in feldspar phenocrysts, and evidence of grain boundary melting is preserved when the two feldspar phases or feldspars and quartz are in contact (see Figs 12, 17 and 18, below).
View larger version (201K):
[in this window]
[in a new window]
|
Fig. 10. Textures in feldspars and quartz in non-fragmented Fish Canyon magma. (a) Sanidine, plagioclase and quartz phenocrysts in the Nutras Creek Dacite, all showing anhedral outlines and internal melt channels connected to the matrix (Bfc 115c; crossed polars). (Note also plagioclase phenocrysts, to the left and to the right of the large sanidine, which are euhedral and lack resorption textures.) (b) Large, anhedral sanidine in the Pagosa Peak Dacite (PCB1; partially crossed polars). (c) Complexly zoned plagioclase in the Pagosa Peak Dacite, wherein a euhedral mantle displaying oscillatory growth zoning has overgrown a partly resorbed core (Bfc 12; crossed polars). (Note the prominent ‘calcic spike’ immediately outboard of the resorption surface that marks the boundary between the core and the overgrowth.) (d) Corroded plagioclase grain in the Pagosa Peak Dacite (Bfc 171; partially crossed polars). Euhedral terminations are never observed on sanidine and quartz, but are the rule rather than the exception on plagioclase.
|
|
View larger version (202K):
[in this window]
[in a new window]
|
Fig. 12. Two types of plagioclase inclusions in Fish Canyon sanidines. (a, b) Multiple isolated, anhedral inclusions in optical continuity, showing a range of sizes from a few microns to 0·5 mm (PCB1 and Bfc 171; crossed polars). In (b), inclusions are concentrated at the rim of the sanidine and they are spatially associated with large oscillatory variations in Cn content (Ba) of the host sanidine. In contrast, the core (delimited by an undulatory resorption surface) is both unzoned and free of inclusions. (c) and (d), examples of sanidine containing larger subhedral–euhedral plagioclase laths. These are typical of the second type of inclusions. This second population is commonly associated with incipient dissolution of both phases (i.e. grain boundary melting) along the contact with the host sanidine (Bfc 12 and Bfc 91; crossed polars). The sanidine in (c) contains both populations.
|
|
View larger version (190K):
[in this window]
[in a new window]
|
Fig. 13. Fish Canyon sanidines showing compositional zoning associated with plagioclase inclusions in optical continuity (all images with crossed polars). Each band has an innermost Ba-rich zone (light grey), which gives way outward to dark grey sanidine with lower Ba concentrations. (a) Plagioclase inclusions in optical continuity enclosed in the zoned outer part of a large sanidine (inclusions are absent in the homogeneous core—boundary marked by arrows). (b) Truncation of earlier growth bands by an undulatory surface (arrows) suggesting dissolution (Bfc 41). Three optically continuous plagioclase inclusions are preserved in the outermost rim. (c) and (d), two sanidines showing numerous bands in association with plagioclase inclusions in optical continuity (Bfc 31 and 100a). Minerals were partly fractured during polishing.
|
|
View larger version (107K):
[in this window]
[in a new window]
|
Fig. 15. Granophyric overgrowths on Fish Canyon feldspars (both images with crossed polars). (a) Granophyric rim on a large crystal in the Nutras Creek Dacite associated with multiple broad oscillatory zones, which are spatially associated with plagioclase inclusions in optical continuity and numerous melt pockets (Bfc 115). (b) Granophyre filling a fracture in a sanidine from the late-erupted intracaldera Fish Canyon Tuff. These fractures are thought to result from crystal shattering during decompression of the magma chamber as a result of early eruptions of the Fish Canyon magmatic system (see main text; Bfc 191).
|
|
View larger version (99K):
[in this window]
[in a new window]
|
Fig. 16. Discontinuous plagioclase mantles on sanidines in the Pagosa Peak Dacite. (a) The multiple grains of plagioclase intergrown with the outer margin of the associated sanidine are all in optical continuity with each other and with the plagioclase crystals located within this sanidine (Bfc 171; plane-polarized light; lower right margin of the sanidine crystal is truncated at the edge of the thin section). (b) Large grain of resorbed sanidine rimmed on one end by multiple grains of plagioclase, all in optical continuity. We interpret the attachment sites of these grains on sanidine as epitaxial nucleation sites. Plagioclase grains, which are not in contact with sanidine in the plane of this thin section, are still likely to have attained their orientation by epitaxial nucleation. We interpret this grain as an example of the initial stages of formation of Rapakivi-like texture wherein plagioclase mantles form during sanidine dissolution.
|
|
View larger version (78K):
[in this window]
[in a new window]
|
Fig. 17. Evidence for melting of quartz inclusions in feldspar hosts. (a) High-magnification view of the melt channels of the large Pagosa Peak Dacite sanidine shown in Fig. 10b (PCB1). Birefringent multigrain aggregates within melt channels are SiO2, which we interpret as the product of late reprecipitation from extremely SiO2-rich melt after melting of quartz inclusions. (b) Complex melt inclusion in a Pagosa Peak Dacite plagioclase. The inner glassy zone marked by the arrow has the composition of 95–96 wt % SiO2 (+ 1·8–2 wt % Al2O3 and <1 wt % CaO, Na2O and K2O) whereas the surrounding darker glass has the same composition as typical matrix (Bfc 59; plane-polarized light). We interpret this occurrence as melting of a quartz inclusion and host plagioclase within a melt channel that was connected in three dimensions to the surrounding melt. It should be noted that the volume of melted quartz is approximately equal to the volume of melted plagioclase within this melt pool.
|
|
View larger version (63K):
[in this window]
[in a new window]
|
Fig. 18. (a, b) Electron microprobe traverse across a plagioclase–sanidine grain boundary preserved in a microplutonic fragment, designated by the line on the photomicrograph at the top left. This profile shows wave-like compositional zoning on both sides of the contact, interpreted as diffusive Na–K exchange across the grain boundary induced by a thermal fluctuation. (Note the irregular contact, highlighted by the dashed line, suggesting a nearly horizontal interface between the two phases.) (c) Ternary diagram showing the three feldspar pairs chosen on the microprobe profile for the thermometry exercise. T1 and T3 have essentially the same compositions and overlap in this diagram. T2 is shifted toward higher Or in sanidine and lower An in plagioclase, inducing a clockwise rotation of the tie-line, which crosses the other two. Temperatures are calculated using the Solvcalc program of Wen & Nekvasil (1994) and the calibration of Elkins & Grove (1990). (Note grain boundary melting at the contact between the two phases.)
|
|
|
MODAL ABUNDANCES
|
|---|
The Fish Canyon magma is a crystal-rich (40–45% crystals)
dacite characterized by a low-variance mineral assemblage, consisting
of two feldspars, quartz, hornblende, biotite, Fe–Ti oxides,
apatite, sphene, zircon, and rare pyrrhotite, in a microlite-free
interstitial melt of high-silica rhyolite (76·7–77·7
wt % SiO
2, recalculated to 100% anhydrous, as are all major
element compositions reported in the text of this paper). These
attributes characterize all three stratigraphic units that make
up the Fish Canyon magmatic system. Modal abundances of phenocrysts
are similar among Pagosa Peak Dacite blobs and Nutras Creek
Dacite samples. Phenocryst abundances in these samples of non-fragmented
Fish Canyon magma are generally lower than in the Fish Canyon
Tuff (Table
1). The coarse grain size of some phases (up to
5 mm), the heterogeneous distribution of minerals on the centimeter
scale, and uncertainties in distinguishing quartz and sanidine
in thin section all impede highly accurate modal counting. Variations
in measured modal feldspars, quartz, and hornblende among Pagosa
Peak Dacite samples (Table
1) probably reflect the difficulty
of obtaining accurate estimates as much as real variability.
Despite these obstacles, the impact of glass winnowing during
ash-flow emplacement is perceptible (Table
1). One outflow Fish
Canyon Tuff sample (‘Elep’) is
11% more crystal
rich than any magma blob from the Pagosa Peak Dacite.
|
WHOLE-ROCK CHEMISTRY
|
|---|
Whitney & Stormer (1985) reported whole-rock major element
compositions for Fish Canyon Tuff (outflow tuff facies from
the region SSE of the La Garita caldera) ranging in SiO
2 wt
% from
62·5 to 67·5. On the basis of mass-balance
calculations tied to glass compositions and crystal contents,
they concluded that the Fish Canyon magma is a high-SiO
2 dacite
(quartz latite in their nomenclature) with
67–68 wt %
SiO
2, and that the compositions with lower SiO
2 are the products
of variable syn-eruptive crystal–ash fractionation (glass
elutriation). We have analyzed 17 bulk Fish Canyon Tuff samples
(16 outflow and one intracaldera) with widespread geographic
coverage, including one with an extremely high crystal content
(
55%; Elep, Table
2). Our data range from 65·4 to 68·5
wt % SiO
2, so we have not been able to reproduce the wide spectrum
of SiO
2 values reported by
Whitney & Stormer (1985). For
most major elements (Fig.
3), our Fish Canyon samples define
linear correlations with SiO
2 [compare with fig.
3 of
Whitney & Stormer (1985)],
and these data fall on trends extrapolated
from glass compositions (electron microprobe analyses) and the
average composition of Pagosa Peak Dacite magma blobs [compare
samples 1, 83, 55 and 34, Table
2 and analysis 10, table
6 of
Lipman (1975)], which we infer to be representative of bulk
magma compositions. The three least crystal-rich tuff samples
fall close in major element composition to the average Pagosa
Peak Dacite, as do a large fiamme from the northern intracaldera
facies of the Fish Canyon Tuff (sample 129, Table
2) and the
post-collapse Nutras Creek Dacite (sample 115, Table
2).
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 3. Major element Harker variation diagrams for the Fish Canyon magmatic system (wt % oxides normalized to anhydrous values). The data plotted in these diagrams are those reported in Table 2 (Pagosa Peak Dacite, Nutras Creek Dacite and granodioritic xenolith) plus additional Fish Canyon Tuff samples for which we do not have the complete range of trace element data.
|
|
On the basis of these data, we infer that the major element range of Fish Canyon magma compositions was narrowly restricted (67·5–68·5 wt % SiO2), but not homogeneous within the limits of analytical precision. As both early-erupted and late-erupted magma compositions are similar, and there are few systematic variations in mineral chemistry among the eruptive phases, our data confirm the inference that the Fish Canyon magma chamber was not characterized by strong compositional and thermal gradients (large variations in Cl are due to degree of devitrification). Small, fine-grained mafic magmatic inclusions are found in the late-erupted intracaldera facies of the northern collapse depression (QMI 2, Table 2; see the section ‘Magmatic inclusions’). Although these inclusions are hybridized (mixing with Fish Canyon magma), the linear trends defined by glass and whole-rock compositions of Fish Canyon material do not project toward the analyzed inclusion for most elements (Fig. 3). This observation is in accord with the virtual absence of quenched fragments of andesitic magma in more than 250 Fish Canyon samples that we have examined.
Variations of some trace elements in non-fragmented samples (i.e. magmatic compositions) span wider ranges than do major elements in the same samples. Although the Nutras Creek Dacite and pumice from the intracaldera Fish Canyon Tuff are indistinguishable from the average Pagosa Peak Dacite blobs in terms of almost all trace elements (Ba alone is low in pumice), there are differences among Pagosa Peak Dacite blob samples of the order of 15% relative for Y and heavy rare earth elements (HREE) (La/Yb 18–21) and 10% for Zr and Hf. Given the presence of multiple accessory phases (apatite, zircon, sphene), it is difficult to assess the role of modal differences among analyzed splits of material vs first-order magmatic processes in generating these variations.
On the basis of the linear arrays defined by the Fish Canyon Tuff samples and glass in major element variation diagrams (Fig. 3) and the elemental enrichments and depletions illustrated in Fig. 4, we concur with Whitney & Stormer (1985) that most of the compositional variability among Fish Canyon Tuff samples is likely to be the result of syn-eruptive crystal–ash fractionation, not the presence of substantial pre-eruptive compositional gradients in the magma chamber. In addition, although there are previously unrecognized quenched andesite inclusions in the intracaldera facies, volumetrically important mingling or mixing with a more mafic component does not appear to be a factor in generating even the small observed variability. In light of the probability that the magma chamber was weakly heterogeneous, such heterogeneities do not appear to have been erupted in a systematic fashion, with the possible exception of the preferential occurrence of andesitic inclusions in late-erupted intracaldera tuff. We emphasize that the sampling of multiple eruptive phases of the Fish Canyon magmatic system in this study provides a more solid basis than previous datasets for documenting variations in magma composition and interpreting them.
View larger version (11K):
[in this window]
[in a new window]
|
Fig. 4. Trace element enrichment–depletion diagram; the most crystal-rich Fish Canyon Tuff sample (Elep) is normalized to a Pagosa Peak Dacite blob sample (55), which is representative of non-fragmented Fish Canyon magma. This diagram illustrates general consequences of crystal–ash fractionation during pyroclastic flow transport and emplacement. Fine particles of ash are preferentially lost during transport, and tuff samples are enriched in crystals with respect to non-fragmented magma. Elements concentrated in one or more of the nine solid phases are enriched (>1) by as much as 30–40% relative to the magma composition, whereas those few elements that are incompatible with respect to all the mineral phases are depleted (<1) in the tuff.
|
|
As noted by Whitney & Stormer (1985), the Fish Canyon magma is similar for most major elements to average granodiorite (Nockolds et al., 1978). Moreover, the products of the Fish Canyon magmatic system closely approximate the average composition of the upper continental crust proposed by Taylor & McLennan (1985; Fig. 5). In comparison with the latter, the Fish Canyon magma is higher in Ba, Th, K2O (and total alkalis), and light REE (LREE) in accord with the relatively high-K nature of San Juan volcanism. The Fish Canyon magma is also notable in having a more pronounced negative Nb–Ta anomaly, and therefore an apparent arc-like signature (Ba/Nb 50 vs 22 for the crust; La/Nb 2·5 vs 1·2), but we emphasize that the Fish Canyon composition has relatively high Nb and Ta (16–18 ppm Nb and 1·1–1·2 ppm Ta) in comparison with dacitic magmas from modern continental margin arcs. We concur with Whitney & Stormer (1985) that the similarity of the Fish Canyon magma to average granodiorite (and average upper crust) implies that the processes involved in the generation of the Fish Canyon magma were similar to those that are typically responsible for the generation of batholithic intrusions.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 5. Mantle-normalized multi-element diagram comparing the compositions of the eruptive products of the Fish Canyon magmatic system with (1) an estimate of the composition of the upper crust by Taylor & McLennan (1985), and (2) a granodioritic xenolith with Fish Canyon mineralogy and mineral chemistry (collected from intracaldera Fish Canyon Tuff). Concentrations are normalized to the primitive mantle composition of Sun & McDonough (1989). Representative non-fragmented samples from the Pagosa Peak Dacite, Fish Canyon Tuff, and Nutras Creek Dacite, as well as the analysis of the granodioritic xenolith, are plotted to illustrate the absence of large systematic differences between the early- and late-erupted products of the Fish Canyon magmatic system. The shaded field encompasses all the analyses listed in Table 2, and these account for the total variability of the Fish Canyon magma, as well as modifications caused by crystal–ash fractionation during transport (Figs 3 and 4).
|
|
|
PHENOCRYST TEXTURES AND CHEMISTRY
|
|---|
Our re-examination of phenocryst textures and compositions in
the Fish Canyon magma has revealed greater textural variability
in feldspars, and greater compositional variability in feldspars
and ferromagnesian phases than previously reported by
Whitney & Stormer (1985) and
Johnson & Rutherford (1989a). The
following sections discuss each phase separately, and first-order
interpretations of the observed chemical variations are provided,
before these results are integrated in a petrologic model.
Biotite
Biotite phenocrysts in Fish Canyon magma, 0·5–5 mm in size, are euhedral (Fig. 6) and have compositions typical of large dacitic crystal-rich ash-flow sheets, such as the Atana (Lindsay et al., 2001) and Cerro Galan ignimbrites (Francis et al., 1989). Grain-to-grain variations in major elements are up to 5 wt % absolute for FeO and MgO, but few inter-element correlations are apparent in the dataset as a whole, apart from slightly higher Fe/(Fe + Mg) in the intracaldera Fish Canyon Tuff and Nutras Creek Dacite (Table 3; Fig. 7). High-resolution (8–15 µm step), core-to-rim electron microprobe traverses across biotite phenocrysts from all three stratigraphic units of the Fish Canyon system display nearly flat zoning profiles for major and minor elements, with the exception of fluorine, which varies from 0·8–0·9 wt % absolute to >2 wt %, both as relatively abrupt spikes in the internal parts of the crystals and as a progressive increase towards rims. Fluorine replacement of OH- in micas depends primarily on: (1) the activity of HF in the melt during crystallization; (2) the temperature at which the F–OH exchange occurs; (3) the composition of the octahedral layer in the mica, in particular the Mg/Fe ratio (Munoz, 1984). As F variations in Fish Canyon biotites are largely independent of Mg/Fe, we suggest that temperature and/or the activity of HF in the melt varied during biotite crystallization.
View larger version (96K):
[in this window]
[in a new window]
|
Fig. 6. Hornblende and biotite phenocrysts in Fish Canyon magma. (a) Euhedral to subhedral hornblende and biotite phenocrysts in the Pagosa Peak Dacite (Bfc 171; plane-polarized light). (b) Large hornblende containing a complex core composed of an intergrowth of plagioclase, biotite, and oxides, as well as smaller hornblendes, which are compositionally indistinguishable from the overgrowth. A large euhedral sphene crystal is also included in the hornblende rim (Bfc 68; plane-polarized light). (c) Large hornblende phenocryst showing an oscillatory zoned rim and a pargasitic core (PCB1; partially crossed polars).
|
|
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 7. Major element variation diagrams for biotite (electron microprobe) in each of the stratigraphic units of the Fish Canyon magmatic system. It should be noted that fields overlap extensively for all plotted major elements.
|
|
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 21. Major element variation diagrams comparing analyses of hornblende and biotite in Fish Canyon magma (intracaldera and outflow Fish Canyon Tuff) with the same phases in the granodioritic xenoliths. In terms of hornblende composition, fields do not perfectly overlap, but are largely similar, whereas biotite compositions in the granodioritic xenoliths are slightly shifted towards lower FeOtot and TiO2.
|
|
Hornblende
Hornblende occurs in Fish Canyon magma as euhedral, generally poikilitic, phenocrysts (0·5–3 mm), without evidence for reaction with the surrounding melt (Fig. 6). An extensive electron microprobe dataset of amphibole compositions (2146 analyses) collected from multiple samples of all units of the Fish Canyon magmatic system defines significant major element variability. The vast majority (>90 %) are hornblende sensu stricto, but the most alkali-rich and silica-poor analyses reach the edenite field. The overall dispersion is extended toward pargasite by one texturally distinct resorbed core in a large hornblende phenocryst in a Pagosa Peak Dacite sample (PCB1). Excluding this pargasitic core, which we interpret as a relic from an earlier magmatic stage, variations up to 4 wt % absolute for Al2O3, and 6 wt % for SiO2 are observed. When plotted against silica, Al2O3, FeOtot, MgO, MnO, TiO2, Na2O, and K2O define linear arrays with negative slopes, except for MgO and MnO, which correlate positively with SiO2 (Fig. 8). The hornblende grown by Johnson & Rutherford (1989a) in an experiment at 2 kbar and 780°C and the composition of their natural hornblende analysis plot near the average of the new dataset.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 8. Major element variation diagrams for hornblendes from each stratigraphic unit of the Fish Canyon magmatic system. Data from the granodioritic xenolith are not shown. The natural and experimental hornblende compositions reported by Johnson & Rutherford (1989a) are plotted for comparison. The experimental product plots close to the average Fish Canyon hornblende composition for most elements. (Note the overlap between amphibole compositions from the various Fish Canyon stratigraphic units, except for the distinctive pargasitic core, which we interpret as a relic inherited from an earlier event.)
|
|
Fourteen high resolution (5–10 µm point spacing) microprobe traverses across large euhedral hornblende grains record significant internal major element variability. A majority of these traverses (nine) record rimward increases in Al2O3, FeOtot, TiO2, Na2O, and K2O, which are compensated by decreases in MgO, MnO, and SiO2. Typical near-core compositions have Al2O3 contents ranging from 5 to 6 wt % absolute, whereas rim compositions reach 7·5–8·5 wt % Al2O3 (Fig. 9). Oscillatory zoning (10–20 µm wide with amplitude up to ±1 wt % Al2O3 in some phenocrysts) is superimposed on the rimward zoning. The fine-scale oscillations are characterized by the same strong correlations between all major elements: the zoning profiles of Al2O3, FeOtot, TiO2, Na2O, and K2O are nearly parallel, whereas the profiles of MgO, MnO, and SiO2 are mirror images. This striking interdependence indicates that the same coupled substitutions are responsible for both the oscillations and the core-to-rim trends.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 9. Electron microprobe traverse (core to rim) across one large hornblende phenocryst in the Pagosa Peak Dacite (Bfc 196a; partially crossed polars). The crystal has been truncated at the edge of the thin section. The area between 200 and 600 µm probably represents the core composition of this crystal. The profiles of the oxides shown here are either sub-parallel (Al2O3–FeOtot and MgO–SiO2) or are equally good mirror images (e.g. Al2O3–SiO2 and Fe2O3–SiO2). The Na2O + K2O, and TiO2 profiles are sub-parallel to the Al2O3 profile, whereas MnO is its mirror image. Thus, it would appear that the same coupled substitutions dominate throughout the profiles and that both the long-wavelength trends and fine-scale oscillations were caused by the same factors.
|
|
The thermobarometric potential of Al content in hornblendes in near-solidus silicic magmas has been assessed in a number of studies (e.g. Hammarstrom & Zen, 1986; Hollister et al., 1987; Blundy & Holland, 1990; Vyhnal et al., 1991; Schmidt, 1992; Holland & Blundy, 1994; Anderson & Smith, 1995; Anderson, 1996; Ague, 1997), and notably on Fish Canyon material by Johnson & Rutherford (1989b) in the first experimental calibration of the Al-in-hornblende barometer. The Al zoning recorded by microprobe traverses across Fish Canyon hornblendes is explored in more detail in a companion paper (Bachmann & Dungan, 2002), but the main characteristics of the profiles are summarized here. The dominant substitution controlling Al variations is the edenite exchange [SiT + 
A 
AlT + (Na + K)A], but it is coupled with a less important Ti-Tschermak exchange (SiT + MnM1-M3 AlT + TiM1-M3); both are inferred to be temperature sensitive (Czamanske & Wones, 1973; Spear, 1981; Blundy & Holland, 1990). In contrast, the pressure sensitivity of the Al-Tschermak substitution (SiT + MgM1-M3 AlT + AlM1-M3), demonstrated using Fish Canyon material by Johnson & Rutherford (1989b), does not play a role in increasing the Al content of natural Fish Canyon amphibole, as AlM1-M3 is <0·2 atom p.f.u. and does not correlate significantly with other components. Moreover, if Altot content in Fish Canyon hornblende were primarily controlled by pressure, an increase toward crystal rims would translate dynamically into a foundering magma chamber. The minimum depth implied by hornblende with 6 wt % Al2O3 would have been shallower than 3–5 km (<1 kbar at 760°C and <1·7 kbar at 700°C), using the recent recalibration of the Al-in-hornblende barometer of Anderson & Smith (1995). In contrast to this geologically unrealistic scenario, temperatures calculated using the geothermometric algorithm of Holland & Blundy (1994; edenite–richterite thermometer) at 2·25 kbar pressure increase from 708 ± 10°C for an Al2O3 content of 6 wt % and An28 plagioclase, the typical values of inner parts of hornblende and plagioclase phenocrysts, to 756 ± 15°C for outermost rim compositions (7·7–8·0 wt % Al2O3) with An33 plagioclase. These calculated temperatures are realistic, and the temperature obtained from the rim composition is indistinguishable from independent determinations using the Fe–Ti oxide thermometer (760 ± 30°C; Johnson & Rutherford, 1989a) or quartz–magnetite oxygen thermometry (762 ± 10°C, I. N. Bindeman, unpublished data, 2001). These zoning profiles are, therefore, consistent with hornblende crystallization during nearly isobaric reheating. Increasing fH2O may also have played a role in this rimward increase in Al (Scaillet & Evans, 1999).
Quartz
As noted previously by Lipman et al. (1978) and Whitney & Stormer (1985), quartz occurs as large (up to 5–6 mm in diameter) ameboid grains, deeply dissected by melt channels and pools (Fig. 10a). All three Fish Canyon units display these anhedral quartz textures, but they are particularly obvious in the Pagosa Peak Dacite and the Nutras Creek Dacite, wherein crystal shattering was less pervasive.
Plagioclase
Plagioclase displays diverse textures, showing both internal and marginal resorption as well as oscillatory-zoned euhedral overgrowths on volumetrically minor partially resorbed cores (Fig. 10c). Plagioclase phenocrysts have compositions ranging mainly from An25 to An35, although a few volumetrically minor more calcic cores have compositions up to An75 (Table 4). Despite the fact that high-anorthite cores were not analyzed in the Nutras Creek Dacite, because of the relative paucity of plagioclase analyses in this unit, plagioclase compositions are similar in all three units of the Fish Canyon magmatic system. Oscillatory-zoned rims on plagioclase phenocrysts have two important chemical characteristics that were not noted previously (Fig. 11). First, the background trend progressively rises to more calcic average compositions, from An27–28 in grain interiors to An32-33 at rims. Second, narrow (10 µm wide) excursions to more calcic compositions (up to An40) following dissolution surfaces periodically interrupt the typical background trend.
Sanidine
Texturally, sanidine in the Fish Canyon magma is very similar to quartz (Fig. 10b). It also occurs dominantly as large (2–5 mm) ameboid grains in all three units of the Fish Canyon magmatic system, although efficient crystal shattering and dispersal related to magma fragmentation and emplacement has rendered primary grain morphologies obscure in the Fish Canyon Tuff. In contrast to quartz, many of the extensively resorbed sanidine grains contain inclusions of other mineral phases. Plagioclase is by far the most abundant, but hornblende, biotite, Fe–Ti oxides, apatite, and sphene are also present in minor quantities. Plagioclase inclusions in these poikilitic sanidines are of two types (Fig. 12), both having the same compositions as the euhedral overgrowths on plagioclase phenocrysts (An30 ± 3). The first consists of irregularly shaped grains, which are in optical continuity despite being physically isolated (Fig. 12a–c). These inclusions range in size from a few microns to 1 mm, they are generally subequant with irregular outlines (Fig. 12a–c), and they are preferentially concentrated near the rims of large sanidine crystals (Fig. 12b). Inclusions of the second type are subhedral–euhedral laths up to 1 mm long, characterized by thin melt zones along contacts with sanidine (Fig. 12c and d).
In addition to the marginal and internal resorption textures and multiple types of mineral and glass inclusions, many Fish Canyon alkali feldspars are conspicuously zoned, mainly in the outer parts of crystals (Fig. 13). Optical zoning is manifested as bands of 0·1–1 mm width with internally variable and slightly different extinction angles from the unzoned cores of the crystals. Whereas these bands are generally subparallel to anhedral grain boundaries, their morphology is more irregular than euhedral growth zoning (i.e. planar crystal faces). Where multiple bands are present, an undulatory surface commonly truncates earlier bands, suggesting dissolution (Fig. 13b and d). Celsian (Cn) increases by a factor of 2 mol % absolute, immediately outside the zoning boundaries, balanced mainly by decreases in Or and increases in Ab (Fig. 14). Within each zone, Cn then decreases progressively to its background value (2 mol %), accompanied by a decrease in Or and an increase in Ab (Fig. 14a, Table 5). In phenocrysts containing multiple juxtaposed zones, Cn variations define sawtooth profiles, with abrupt re-establishments of high Cn content across zoning boundaries, followed by linear decreases (Fig. 14b; fig. 5 of Lipman et al., 1997). Sanidines characterized by variable Cn content are invariably poikilitic, containing swarms of isolated plagioclase inclusions in optical continuity [Figs 12b and c, 13, 14a, and 20b (below)], but such inclusions are absent from adjacent unzoned cores.
View larger version (75K):
[in this window]
[in a new window]
|
Fig. 14. (a) Electron microprobe traverse across a zoning boundary in a large Pagosa Peak Dacite sanidine (Bfc 171, same as in Fig. 12b), showing a sharp increase in celsian (Cn) (Ba) just outboard the undulating resorption surface, correlated with a decrease in Or and increase in Ab. The trends then define a progressive increase in Ab, correlated with a decrease in Cn and Or away from this surface. This Ba-rich outer zone is thought to represent a second generation of sanidine, which grew from a melt enriched in Ba. The origin of this Ba enrichment in the melt is thought to result directly from the development of Rapakivi textures [see Wark & Stimac (1992) and text for details], as Ba released during sanidine dissolution is not incorporated in the plagioclase structure. Limited addition of Ba from a more mafic magma may also have played a role in the development of this zoning. (b) Electron microprobe traverse across a large sanidine (PCB1; crossed polars) showing abrupt increases in Cn, by up to 1 mol %, across optical zoning boundaries (numbered 1–4). The profile displayed here was slightly smoothed using a running average.
|
|
View larger version (196K):
[in this window]
[in a new window]
|
Fig. 20. Photomicrographs of granodioritic fragments and hybrid andesite inclusions entrained in the intracaldera Fish Canyon Tuff. (a) Fish Canyon mineral assemblage in a granodioritic xenolith (Bfc 187; partially crossed polars). Mineral compositions in this granodiorite overlap extensively with those in Fish Canyon magma (Fig. 21). (b) Large poikilitic K-feldspar, containing plagioclase inclusions in optical continuity and Ba-rich zones (Bfc 187; crossed polars) (c) Macroscopic view of a hybrid andesite inclusion enclosed in the intracaldera Fish Canyon Tuff, showing a band of mingled crystal-rich Fish Canyon magma marked by a high concentration of xenocrysts (MLX QMI). (d) The only occurrence of pyroxene and euhedral sanidine in the Fish Canyon magmatic system, which are both present in the andesitic inclusions. The clinopyroxenes are also found free in the matrix of the inclusions and, like hornblendes, always show oxidized rims (QMI 2; plane-polarized light). The fine-grained texture should be noted, showing K-rich sanidine (Or72–80) and plagioclase microlites more calcic than average Fish Canyon plagioclases (An40-60 instead of An27–33).
|
|
Fine-scale granophyric intergrowths of quartz and K-feldspar with planar terminations against adjacent melt occur as overgrowths around pre-existing feldspar crystals in late-erupted northern and central intracaldera Fish Canyon Tuff, and in the Nutras Creek Dacite (Lipman et al., 1997). Granophyre is a late magmatic feature, as it has faceted crystal faces and it fills open fractures in broken host crystals, thought to have been partly disrupted by expansion of melt inclusions during rapid decompression related to the early Fish Canyon eruptions (Fig. 15; Best & Christiansen, 1997; Lipman et al., 1997). An important feature, not mentioned by Lipman et al. (1997), is that granophyric overgrowths appear to have preferentially nucleated on sanidines displaying bands of variable Cn content riddled with plagioclase inclusions in optical continuity.
Accessory phases
Accessory phases in the Fish Canyon magma (sphene, Fe–Ti oxides, apatite, zircon, pyrrhotite) are generally euhedral. Sphene, Fe–Ti oxides, apatite, and zircons are all found as inclusions in the major phases and in the interstitial glass. Pyrrhotite is rare and uniquely present as inclusions, mainly in magnetite (Whitney & Stormer, 1985). The largest sphene and Fe–Ti oxide grains can reach 0·5 mm in size, whereas the other phases are generally smaller than 50 µm. We have not undertaken mineral chemistry analyses of these phases.
|
PETROLOGICAL IMPLICATIONS
|
|---|
Reverse zoning in plagioclase overgrowths
The major element composition of plagioclase is controlled by
pressure, temperature, and melt composition, including a major
dependence on H
2O content (
Rutherford & Devine, 1988;
Housh & Luhr, 1991).
The reverse compositional zoning trend over
hundreds of microns in the plagioclase rims requires a progressive
change in one or more of these parameters. As for rimward increases
of Al in hornblende, heating could have conceivably played a
role in the development of these zoning profiles, but changes
in melt chemistry (increased Ca and/or H
2O as a result of interactions
with less evolved melt) may have also contributed to the overall
reverse trend and calcic spikes. Calcic cores in plagioclase
probably represent relics from an earlier magmatic stage.
Plagioclase inclusions in optical continuity associated with Ba zoning in sanidine: Rapakivi textures
One process potentially capable of generating oriented plagioclase inclusions in K-feldspar is phase unmixing at the solvus (exsolution). However, the diversity of sizes and the irregular subequant shapes of the inclusions set them apart from any exsolution textures described in the literature (e.g. Smith, 1974). An alternative hypothesis, proposed by Stimac & Wark (1992) to explain similar textures in silicic lavas of Clear Lake, California, is epitaxial nucleation and growth of plagioclase at many points on the periphery of sanidine, followed by regrowth of sanidine in the interstices between plagioclase grains. The Clear Lake silicic lavas show multiple examples of well-developed plagioclase mantles on sanidine (i.e. Rapakivi textures; Sederholm, 1891), into which these poikilitic zones grade, and the inference that these inclusions are related to plagioclase mantling is straightforward. Although complete plagioclase mantles are rare in Fish Canyon magma, multiple grains of plagioclase with a common optical orientation attached to the rims of large sanidine crystals are preserved in a few samples (Fig. 16), hinting at remnants, and/or incipient formation, of discontinuous plagioclase mantles.
The spa
TOP
ABSTRACT
INTRODUCTION