Petrology and geochemistry of lava and ash erupted from Volcán Colima, Mexico,during 1998–2005
Ivan P. Savov a,⁎, James F. Luhr b,1, Carlos Navarro-Ochoa c
a School of Earth and Environment, The University of Leeds, Leeds LS2 9JT, United Kingdomb Department of Mineral Sciences, National Museum of Natural Sciences, Smithsonian Institution, PO Box 37012, NHB-119, Washington, D.C. 20013-7012, USAc Observatorio Vulcanológico de la Universidad de Colima, Av. 25 de Julio # 965 Colonia Villa San Sebastián, Colima, Col. C.P. 28045, Mexico
a r t i c l e i n f o
⁎ Corresponding author. School of Earth and EnvironLeeds LS2 9JT, United Kingdom.
Article history:Received 29 November 2007Accepted 27 February 2008Available online 7 March 2008
Lava (n=8) and bulk ash samples (n=6) erupted between July 1999 and June 2005 were investigated to extendtime-series compositional and textural studies of the products erupted from Volcán Colima since 1869. Inparticular, we seek to evaluate the possibility that the current activity will culminate in major explosivePlinian-style event similar to that in 1913. Lava samples continue to show relatively heterogeneous whole-rock compositions with some significant mafic spikes (1999, 2001) as have prevailed since 1976. GroundmassSiO2 contents continue trends to lower levels that have prevailed since 1961, in the direction of the still lowergroundmass SiO2 contents found in 1913 scoriae. Importantly, ash samples from investigated Vulcanian-styleexplosive eruptions in 2005 are devoid of particles with micro-vesiculated groundmass textures; suchtextures characterized the 1913 scoriae, signifying expansion of in-situ magmatic gas as the propellant of the1913 eruption. All magmas erupted since 1913 appear to have arrived in the upper volcanic conduit system in adegassed state. The small to moderate Vulcanian-style explosive eruptions, which have been common since1999 (N16,000 events), have blasted ash clouds as high as 11 km a.s.l. and sent pyroclastic flows out todistances of 5 km. These eruptions do not appear to be powered by expansion of in-situ magmatic gas. Newsmall lava domes have been observed in the crater prior to many explosive eruptions. These plugs of degassedlava may temporarily seal the conduit and allow the build-up of magmatic gases streaming upward frombelow ahead of rising and degassing magma. In this interpretation, when gas pressure exceeds the strength ofthe plug seal in the upper conduit, an explosive Vulcanian-style eruption occurs. Alternatively these explosiveeruptions may represent interactions of hot rock and groundwater (phreato-magmatic).
Volcán Colima (103°37′W, 19°30′45′N) was among the 16 Decadevolcanoes selected for special investigation during the last decade of the20th century by the Commission on Mitigation of Volcanic Disasters ofthe International Association for Volcanology and Chemistry of theEarth's Interior (IAVCEI)(Zobin et al., 2002). Research efforts andknowledge gained from the continuous petrological, geochemical andgeophysical monitoring of this highly active volcano with long termeruptive history have proven beneficial not only for understanding thehistory and evolution of volcanism in the entire Trans-Mexican VolcanicBelt (see Allan, 1986; Luhr 1992, 1997; Moore and Carmichael, 1998;Zobin et al., 2002; Atlas et al., 2006; Carmichael et al., 2006), but also instudies of volcanism elsewhere in the world (see Carmichael, 2002;Siebert andSimkin, 2002;Venzkeet al., 2002; LuhrandHaldar, 2006 andreferences therein). In previous studies (Luhr and Carmichael, 1980;1990; Luhr, 2002) the cyclical nature of historical eruptive activity and
ment, The University of Leeds,
l rights reserved.
magma compositions at Volcán Colima were discussed on the basis ofwhole-rock compositions and mineralogy for block-lava samples(∼61 t.% SiO2) erupted during 1869–1880 and 1961–1999, and for scoriaesamples (∼58.5 t.% SiO2) from the last major Plinian-style explosiveeruption in 1913. The primary purpose of this paper is to extend thatevaluation to include lava samples erupted during 1999 to 2005 and ashsamples collected from small to moderate Vulcanian-style explosiveeruptions during 2001–2005. The goal of this petrologic monitoring is toseek evidence for trends toward more mafic (SiO2-poor) magmacompositions, or vesiculated tephra, either of which might herald theend of the current eruptive cycle in a large Plinian-style explosive eventssimilar to that in 1818 and 1913. Our new results throughmid-2005 showcontinued trends toward 1913-type magmas, but no sign of micro-vesiculated groundmass textures as characterize the 1913 scoriae.
Importantly, this work extends the baseline for future comparisons.Our evaluations of groundmass and mineral textures and compositionsfor ash particles from small to moderate Vulcanian eruptions during2005 are the first of their kind at Volcán Colima. Such eruptionsnumbered in the thousands during the late 1800s to 1909, preceding themajor 1913 eruption, but no products of those eruptions are available formodern study. Thus, we have no idea if relatively mafic magma
242 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
compositions or vesiculated scoriae were produced during this interval.All petrologic studies to date for the years 1869–1909 and 1961–1999have focused on block-lava samples or blocks spalled from the fronts ofsuch lava flows (Luhr and Carmichael, 1980; 1990; Luhr, 2002; Moraet al., 2002). Small tomoderate Vulcanian-type explosive eruptionshaveincreased in frequency and intensity since February 1999 (Zobin et al.,2002), and thus an important aim of this investigation is to characterizethe ash samples from recent such eruptions, which appear to be theequivalents of those that preceded themajor explosive eruption of 1913.
2. Geologic setting
Volcán Colima is 3860m tall highly explosive andesitic stratovolcanosituated in the western segment of the Trans-Mexican Volcanic Belt inMexico (Fig. 1). The volcanism at this continuous volcanic arc chain isgoverned by the E–NE subduction of the Rivera and Cocos Plates underthe Pacific and Central American Pacific coasts of Mexico and CentralAmerica. Clusters of seismisity (including strongM=7 earthquakes) andtheir epicenters in the region where the young and hot Rivera Plate iscurrently subducting (at∼2 cm/yr) between theMiddle America Trench(MAT) and Trans-Mexican Volcanic Belt, show that the Wadati-Benioffzone dips at ∼10° in the shallow forearc regions, but than it steepens to∼50° below 40 km (Pardo and Suerez, 1993; Luhr, 1997).
Three large rift zones intersect in western Mexico about 160 km S–SW of Guadalajara to form a structural triple junction: the Tepic-Zacoalco Rift, the Colima Rift and the Chapala Rift (Allan, 1986; Luhr,1997) (Fig. 1). The Colima Rift Zone extends southward from thisstructural triple junction for ∼160 km reaching the Pacific Coast and isdivided into 3 segments — northern, central and southern (Allan,1986; Luhr, 1997). Volcán Colima straddles the intersection of the NE–SW-trending Tamazula Fault and the boundary between the northernand the central parts of the Colima Rift Zone (see Luhr and Carmichael,1980; Allan et al., 1991; Luhr, 1992; Carmichael et al., 2006). In thatsegment the Colima Rift is ∼50 kmwide and the rift flanks are N1 kmelevated against the graben floor.
Fig. 1. General tectonic and geologic map of the Western part of the Mexican Volcanic BeltCarmichael et al. (2006).
3. Patterns of historical eruptive activity
The activity levels of different volcanoes can be compared in anumberofways. Data available from theSmithsonian's Global VolcanismProgram (Siebert and Simkin, 2002) allows volcano activity levels to becompared based on the number of years with at least 1 eruption overvarious spans of time. Based on the number of years with eruptiveactivity since 1900, Volcán Colima (n=52) is the most active volcano inMexico (followed by Popocatépetl: n=24), and ties with Pavlof (AlaskaPeninsula, alson=52) as the twomost active volcanoes inNorthAmerica.Thehistorical activityof VolcánColima, recently reviewedbyBretónet al.(2002), has involved three main types of eruptions:
(1) Vulcanian-style explosive events of small to moderate size andshort duration that can send plumes from a few hundredmeters to as high as 10 km above the volcano, generate wind-blown ash clouds and minor ash falls, blast ballistic bombs outto distances of 5 km, and be associated with pyroclastic flowsextending out for up to ~5 km;
(2) growth of block-lava domes in the summit crater (or in 1869 atthe upper-flank vent Volcancito); small domes are commonlyblasted out by subsequent Vulcanian eruptions, whereas larger-volume domes can grow to fill the crater and spill over thecrater rim to feed block-lava flows (~61 wt.% SiO2) that extend1–4 km down the flanks of the cone; pyroclastic flows oftenaccompany these eruptions and reach distances up to 5 km,generated as blocks spall off the flow front and from thesummit lava dome; and major Plinian-style explosive erup-tions, rated VEI 4 (Volcano Explosivity Index: Newhall and Self,1982), such as occurred in 1818 and 1913; these send sustainederuptive columns well in excess of 10 km high, producesignificant tephra deposits dominated by vesiculated scoriae(~58.5 wt.% SiO2) and ash, and generate voluminous scoria-richpyroclastic-flow deposits that reach distances up to 15 km fromthe vent (Saucedo-Girón, 1997; Bretón et al., 2002).
showing the location of Volcán Colima. MAT=Middle American Trench. Modified from
Table 1Significant eruptive events at Volcán Colima during July 1999 to June 2005
Date and comment
19995 July Degassing events: two degassing events occurred, one of which sent an ash-bearingcloud to 2.5 km above the summit (~6.3 km a.s.l.); ashfall extended 10–20 km to the W.17 July Explosive eruption: large explosive eruption sends plume to ~11 km a.s.l. Somenear-source ejecta falls were incandescent. Associated pyroclastic flows reached 4.5 kmfrom the crater. Evacuations of La Yerbabuena, Juan Barragan, and other S-flankcommunities take place after the explosive eruption.12 Oct. Explosive eruption: ground reports of an eruption that sent an ash cloud to~6 km a.s.l.
200010 Nov. Explosive eruption: ash cloud to 6 km a.s.l., not visible on satellite imagery.
200122 Feb. Explosive eruption: ash cloud to ~2 km above summit, incandescent ballistics to3 km distance, small pyroclastic flows to SW, ash falls in San Marcos (~14 km SE) andTonila (~13 km SSE), new crater with volume of ~190,000 m2.26 May New cryptodome in summit crater: bulge volume estimated at ~150,000 m2.3 Nov. Spine observed on dome: light-colored spine, 40 m tall and 40 m wide at basewith smooth vertical walls.5 Dec. Spine assimilated by growing dome: Spine no longer visible; dome volumeestimated as ~285,000 m2.
200213 Jan. New dome in summit crater: a new dome was photographed, with an estimateddiameter of 170 m and height of 46 m.4 Feb. Dome landslide leaves crater: a landslide-rock-fall from the growing dome in thesouth part of the crater sent material to b1 km towards the S and SSW.5 Feb. La Yerbabuena evacuated: nearest village to the SW of the volcano (8 km).6 Feb. Start of new lava eruption: lava flows and associated rock-falls and pyroclasticflows moved down the S, SW, W, and E flanks with the latter two types typicallyextending 2–3 km from crater. Lava-flow fronts ultimately reached ~550 m laterallyfrom the crater on the SW flank, 2 km on the W flank, and about 0.5 km on the E flank.Lava is thought to have ceased erupting by late Feb. 2003. Several small explosive eventsoccurred toward the end of the lava emission.
20036 May Explosive eruption: aviation authorities report ash cloud to ~6 km a.s.l.27 May Explosive eruption: aviation authorities report ash cloud to ~6 km a.s.l.15 July Explosive eruption: aviation authorities report ash cloud to ~9.1 km a.s.l., visibleon satellite imagery.17 July Explosive eruption and pyroclastic flows: small to moderate explosive eruptionthat drifted to ~6.9 km a.s.l. and spawned pyroclastic flows to distances of ~2 km.2 Aug. Explosive eruption: aviation authorities report ash cloud to ~7.6 km a.s.l., visibleon satellite imagery.28 Aug. Explosive eruption and pyroclastic flows: ash cloud to ~6.9 km a.s.l., pyroclasticflows 2 km to the S, and ballistic blocks. Ash fall in Grullo (~60 km NW).6 Sept. Explosive eruption: aviation authorities report ash cloud to ~6.7 km a.s.l., notvisible on satellite imagery.9–10 Oct. Explosive eruption: aviation authorities report ash clouds to ~8.9 km a.s.l.,visible on satellite imagery.18 Oct. Explosive eruption: aviation authorities report ash cloud to ~7.3 km a.s.l., notvisible on satellite imagery.30 Oct. explosive eruption: steam plume with some ash rose to ~7.3 km.1–2 Dec. Explosive eruptions: aviation authorities report ash cloud to ~7 km a.s.l.19 Dec. Explosive eruption: aviation authorities report ash clouds to ~8.5 km a.s.l.,visible on satellite imagery.30 Dec. Explosive eruption: aviation authorities report ash clouds to ~10.4 km a.s.l.,visible on satellite imagery.
20049 Feb. Explosive eruption: aviation authorities report ash cloud to ~7.5 km a.s.l., visibleon satellite imagery.28 Sept. New dome sighted in summit crater: follows 10 days without explosions and 3days of vigorous fumarolic emissions.30 Sept. Rockfalls down flanks begin: rockfalls (block-and-ash flows) begin as the newdome spills over the W, WNW, and N crater rims. Thru 22 Nov. Lava flows, pyroclasticflows, and small explosions: Block-lava flows traveled down the flanks of Colima,ultimately reaching 2200 m from the crater on the N flank and 600 m on the NW flank.Several small explosions occurred daily, but these generated plumes below 6 km a.s.l.30 Dec. Explosive eruption: aviation authorities report ash cloud to ~7 km a.s.l., visibleon satellite imagery.
20056 Jan. Explosive eruption: aviation authorities report ash cloud to ~7 km a.s.l., visible onsatellite imagery.9 Jan. Explosive eruption: aviation authorities report ash cloud to ~7 km a.s.l., visible onsatellite imagery.
(continued on next page)
(continued)
Date and comment
200517 Jan. Explosive eruption: aviation authorities report ash cloud to ~6.7 km a.s.l., visibleon satellite imagery.26 Jan. Explosive eruption: aviation authorities report ash cloud to ~6.4 km a.s.l., visibleon satellite imagery.12 Feb. Explosive eruption: associated pyroclastic flow traveled ~2.5 km down theMontegrande and San Antonio ravines.10 Mar. Explosive eruption: aviation authorities report ash cloud to 9.1 km a.s.l., visible onsatellite imagery; associatedpyroclasticflows traveled~3 kmdowntheMontegranderavine.13 Mar. Explosive eruption: aviation authorities report ash clouds to ~8.9 km a.s.l.,visible on satellite imagery, as pyroclastic flows down several ravines (El Murerto,Montegrande, San Antonio, and El Cordoban. Ash fall in Mazos (~12.5 km NE).29 Mar. Explosive eruption: aviation authorities report several explosions with ashclouds to a maximum of ~7.6 km a.s.l.1 Apr. Explosive eruption: aviation authorities report several explosions with ash cloudsto a maximum of ~6.7 km a.s.l.7 Apr. Explosive eruption: aviation authorities report several explosions with ash cloudsto a maximum of ~9.1 km a.s.l.14 Apr. Explosive eruption: aviation authorities report several explosions with ashclouds to a maximum of ~6.7 km a.s.l.20 Apr. Explosive eruption: aviation authorities report several explosions with ashclouds to a maximum of ~6.1 km a.s.l.8 May Explosive eruption: small to moderate explosive eruption with pyroclastic flowsdown S-flank ravines.10May Explosive eruption: aviation authorities report ash cloud to ~7.6 km a.s.l., visibleon satellite imagery.11–17 May Pyroclastic flows: on three occasions during this interval explosionsgenerated pyroclastic flows that moved down all flanks.15 May Explosive eruption: crater incandescence preceded the explosive eruption;aviation authorities report ash cloud to ~7.6 km a.s.l.16 May Explosive eruption: small tomoderate explosive eruptionwith pyroclastic flowsdown S-flank ravines.23May Explosive eruption: small to moderate explosive eruptionwith pyroclastic flowsdown S-flank ravines.30 May Explosive eruption: aviation authorities report ash cloud to ~8.5 km a.s.l.,visible on satellite imagery.1 June Explosive eruption: aviation authorities report ash cloud to ~6.1 km a.s.l., visibleon satellite imagery.2 June Explosive eruption: aviation authorities report ash cloud to ~6.1 km a.s.l., visibleon satellite imagery.5 June Explosive eruption: aviation authorities report ash cloud to ~8.9 km a.s.l., visibleon satellite imagery; accompanied by pyroclastic flows.6 June Explosive eruption: aviation authorities report ash cloud to ~8.5 km a.s.l., visibleon satellite imagery.9 June Explosive eruption: the loudest explosion in the past 10 years had an audibleradius of about 50 km.11 June New dome in summit crater.
Note: Eruption plumes are only listed if they are thought to have risen more than 2 kmabove the summit (N6 km a.s.l.).
Table 1 (continued)
243I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
Luhr and Carmichael (1980, 1990) and Luhr (2002) have discussedsimilarities between the 1818 and 1913 Plinian-style eruptions and thepatterns of eruptive activity that followed both of these major explosiveevents. They regarded both eruptions as having terminated historicaleruptive cycles of approximately 100 year duration. Following the 1818eruption, 51 years passed before the next eruption to expel materialbeyond the crater rim in 1869. From 1869 to 1909, there were two majorblock-lava eruptions and thousands (n=2910 between 1893 and 1905) ofsmall tomoderateexplosiveVulcanian-style eruptions (Díaz,1906),whichpeaked in frequency and intensity between 1900 and 1903 (Navarro andCortes 1993; Bretón et al., 2002). Following the 1913 eruption, 48 yearspassed before the next eruption to expelmaterial beyond the crater rim in1961–62. From 1961 tomid-2005, there have been eightmajor block-lavaeruptions and thousands of small to moderate explosive Vulcanian-styleeruptions (nN16,000 events); the latter increased in frequency andintensity first in February 1999 and February 2001, and then to evenhigher levels during 2004 and 2005, continuing as this manuscript waswritten (Bretón et al., 2002; Smithsonian Institution, 1999, 2000, 2001,2002, 2003, 2004, 2005; Venzke et al., 2002).
Table 2Whole-rock Volcán Colima lava analysis by XRF and ICP-MS
244 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
4. Samples studied from 1999–2005 activity
All significant lava or explosive eruptions for July 1999 to mid-2005 are listed in Table 1. For explosive eruptions we include eventsthat generated ash columns greater than 2 km above the summit(N6 km a.s.l.), as well as several eruptions with smaller ash clouds butsignificant pyroclastic flows.
4.1. Lava samples
The 8 analyzed lava samples were collected by personnel of theColima Volcano Observatory as described below:
Col-99E: fragmentof aballistic bombcollected in thePlayón (northerncaldera floor) following the explosive eruption of 17 July 1999.Col-01A: fragment of a ballistic bomb collected by Santiago Hydynin the NE Playón, 1.5 km from the vent, about 1 week following theexplosive eruption of 22 February 2001.Col-01C: sample of the short-lived spine on the summit-craterdome collected by Juan Carlos Gavilanes and Nick Varley on 25November 2001, provided to us by Peter Schaaf.Col-01B: fragment of a ballistic bomb collected in the Playón inDecember 2001, following an unspecified eruption that month.Col-02A: lava block collected on 20 February 2002 at rock-falldeposit accumulating below the front of the active block-lava flow.Col-02B: lava block collected on 25 May 2002 at rock-fall depositaccumulating below the front of the activeW-flank block-lava flow,~1.5 km from the summit.Col-02C: fragment of a ballistic bomb collected on the W flank,~1.5 km from the summit on 25 May 2002. It was ejectedsometime between 16 April 2002 and 25 May 2002.Col-04A: lavablock collectedbyNickVarleyon3October 2004at rock-fall deposit accumulating below the front of the active block-lava flowmoving down the north flank of the cone onto the floor of the Playón.
4.2. Ash samples
The 6 studied ash samples were collected by personnel of theColima Volcano Observatory as described below:
13-03-05: bulk ash collected in LosMazos,12.5 kmNEof the summit,following the 13 March 2005 eruption. Maximum diameter forroughly equant grains is 0.5 mm.13-04-05: bulk ash collected at Cerro Alto, 10 km ENE of thesummit, on 13 April 2005.Maximum diameter for roughly equant grains is 0.6 mm.10-05-05: bulk ash collected in El Fresnito, 15 km NE of thesummit, following the 10 May 2005 eruption. Maximum diameterfor roughly equant grains is 0.6 mm.VF10: bulk ash from one or both of the eruptions on 1 and 2 June2005, collected in Yerbabuena, ~8 km SWof the summit.Maximumdiameter for roughly equant grains is 0.6 mm.VF04: bulk ash from the 6 June 2005 eruption, that fell on plasticsheets 2.2 km NW of the summit, collected by Nick Varley on 11June 2005. Maximumdiameter for roughly equant grains is 1.0mm.VF05: bulk ash from the 6 June 2005 eruption, collected by Navarroat his house in Colima, 32 km S of the summit. Maximum diameterfor roughly equant grains is 0.3 mm.
5. Sample preparation and analytical techniques
Polished 30 μm-thick sections of 8 lava samples we collected fromeruptions between 1999 and 2004 were prepared for petrographic
study, electron microprobe analysis, and electron-imaging. Thewhole-rock compositions of the same lava samples were determinedby X-Ray Fluorescence Spectroscopy (XRF) at the Department ofMineral Sciences, National Museum of Natural Sciences, SmithsonianInstitution (Washington DC, U.S.A.). Trace element compositions forthese same sample powders were determined via Inductively CoupledPlasma-Mass Spectrometry (ICP-MS) at the Washington Universityand are listed in Table 2. Analytical techniques and statements ofprecision and accuracy can be found in Luhr (2002). The whole-rockmajor element compositions of volcanic dome samples from theflanks of Colima were determined by Direct Current Plasma-AtomicEmission Spectroscopy (DCP-AES) at the Department of Geology,University of South Florida (Tampa, FL., U.S.A.). Sample digestions formajor and some trace elements (Cr, Ni, V, Zn, Cu, Mn) in the domesamples were performed after LiBO2 fluxed fusions following theprocedures described in Savov et al. (2005). The low abundance traceelement analysis of the dome samples were performed in the Boston
Fig. 2. Photomicrographs of ash particles from sample VF04, erupted 6 June 2005.(A) Plane-polarized light image at low magnification showing ash particles with avariety of groundmass textures. The three particles labeled a–c with dark, fine-grainedgroundmasses are the type we considered “juvenile.” Particle “a” is enlarged below in(B) and (C) and shown at further enlargement in the backscattered-electron images ofpanel I and J on Fig. 3. The other ash particles in (A) have coarser groundmasses typicalof lava samples. None of the ash particles that we studied contained strongly micro-vesiculated groundmass textures.
245I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
University ICP-MS Facility (Boston, MA, U.S.A.) and are listed inAppendix A. The precision and accuracy for the dome samples is in thesame range as in the main Colima edifice samples reported in Luhr(2002). Further details on the analytical techniques used for the domesamples can be found in Kelley et al. (2003).
For bulk ash samples, splits were placed in beakers and repeatedlyrinsedwith distilled water and decanted to remove fine particles. Thuswe focused our efforts on the coarsest ash particles in each sample. Theresultant coarse splits were dried, mounted in epoxy, and prepared as30 μm-thick polished sections. Maximum grain sizes for each sample(0.3 to 1.0 mm diameter) are listed in the preceding section.
For 2004 lava sample Col-04A and 2005 bulk-ash samples 10-05-05 and VF04, we determined mineral-rim and groundmass composi-tions using JEOL JXA-8900R SuperProbe at the National Museum ofNatural History (NMNH), Smithsonian Institution. Microprobe run-ning conditions were an accelerating voltage of 15 kV, a 20 nA current,and a spot size of 5 micrometers for wavelength dispersive (WDS)analysis. In order to analyze numerous patches of fine-grained toglassy groundmass we used a moving defocused electron beam,accelerating voltage of 15 kV, and 10 nA current (for details— see Luhr,2002).
6. Results
6.1. Groundmass textures
The six ash samples erupted in 2005 that we studied bypetrographic microscope contain a variety of particle types. In themore distal and finer-grained ash samples, it is common to findparticles formed entirely by single crystals or by only groundmass.Most particles, however, are composites of groundmass domains with1 or more attached crystals. Among the groundmass-bearing ashparticles, most abundant among all ash samples investigated are thosewith microlite-rich, glass-poor groundmass textures typical of Colimalavas (Fig. 2A); they likely represent lavas that solidified in the conduitor as part of the summit-crater dome and were subsequently blastedout by Vulcanian-style explosive eruptions. Another minor compo-nent to all six samples is ash particles showing mild to stronghydrothermal alteration, none of which were investigated in thisstudy.
We focused our attention on ash particles that showed the finest-grained and least crystalline groundmass textures, which we thoughtwould represent the most likely candidates for juvenile magmacomponents. At low magnification, these “juvenile” groundmassesappear dark brown–gray in color with abundant small microlites (seeFig. 2A, grains a–c). At higher magnification in plane-polarized light,the plagioclase microlites can be seen more clearly (Fig. 2B), and inreflected light at the same magnification pyroxene and titanomagne-tite microlites are also distinguishable (Fig. 2C). As best as can beevaluated using these petrographic microscope techniques, ashparticles with such “juvenile” groundmass textures are present in all6 of the ash samples investigated.
In Fig. 3 we present 12 representative backscattered-electronimages of groundmass textures in various eruptive products fromVolcán Colima. We start with some textural end-members. First is therapidly quenched scoriae VF95-06α from the 1913 fall deposit (Fig. 3Aand B); its texture is considered to be typical of all scoriae from the1913 fall deposits. Groundmasses in the 1913 scoriae are dominated bymicro-vesiculated glass (glass estimated roughly at 50 vol.%), studdedwith microlites of mafic minerals (hornblende and/or pyroxene),plagioclase, and titanomagnetite up to 10 μm across.
In contrast with the above 1913 scoriae-fall sample, the next pair(Fig. 3C andD) show the groundmass texture of a 1913 scoriae clast fromthe pyroclastic-flow deposit in the Playón, which may also berepresentative of all scoriae from deeper parts of the 1913 pyroclastic-flowdepositswhereheatwas retained. This scoriae clast, about20-cm in
diameter, probably came out of the vent with the same micro-vesiculated, glass-rich groundmass texture recorded in Fig. 3A and B,but it thenwas engulfed in a pyroclastic flow. It was collected at a depthof about 3-m from the top of the 1913 deposit, where it appears to havebeen hot enough to allow the glass to continue crystallizing (annealing).Plagioclase crystals appear to have nucleated much easier than otherminerals. As a consequence, iron and other elements that normallywould be found in pyroxene or titanomagnetite microlites, are mostlypresent in the glass (roughly estimated at 20 vol.%), producing a strongcontrast in the backscattered-electron colors of plagioclase (dark gray)and the Fe-rich glass (medium gray), reflecting a strong difference intheir mean atomic numbers (also note the muchmore subtle differencein the colors of plagioclase and glass in Fig. 2B, F, and J, where microlitesof pyroxene and titanomagnetite are abundant).
246 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
The next pair of groundmass photos (Fig. 3E and F) are for Col-04A,the lava block spalled from the lava-flow front in October 2004. Itsgroundmass is rich in glass (roughly estimated at 50 vol.%) that isvirtually free of vesicles. The lack of vesicles shows that the magmawas degassed upon eruption and did not micro-vesiculate. Theabundance of glass reflects the relatively rapid cooling of the blockonce it spalled off the front of the lava flow and rolled down the slope.Its groundmass texture should not be viewed as typical of Colima lavaflows, but rather as the glass-rich end of a spectrum in which most
Fig. 3. Backscattered-electron images of groundmass textures in various eruptive products froat 500× (see 25-μm scale bar in lower right corner of each). All images on the right show the m10-μm scale bar in lower right corner of each). Phase abbreviations are: hbd=hornblendet=titanomagnetite microlite; px=pyroxene microlite. (A) Sample VF95-06α, a rapidly q103.5865°W, along the road leading to the main N entrance to Nevado de Colima and the “Ca15, a scoriae clast collected from the intra-caldera 1913 pyroclastic-flowdeposit exposed by lablock spalled from the flow front in 2004. (F) Enlarged view of a part of Fig. 2E. (G) Sample 1Fig. 2G. (I) Sample VF04, proximal bulk ash from the 6 June 2005 eruption. (J) Enlarged vie
lava flows have a much higher proportion of microlites, as seen invarious particles in Fig. 2A. Observe (in Fig. 3E and F) the subtle gray-scale difference between plagioclase and glass, indicating thesimilarities of their mean atomic numbers, with glass again somewhathigher (and brighter).
The last two pairs of groundmass images are for single “juvenile” ashparticles from two bulk ash samples collected after moderate explosiveeruptions in early 2005. Thus, from the thousands of potential ashparticles from the 6 evaluated samples, we chose only two particles for
mV. Colima shown in pairs at differentmagnifications. All images on the left were takenagnified portion of the left image indicated by the white rectangle, taken at 1500× (seephenocryst; v=vesicle; g=glass; p=plagioclase microlite; pl=plagioclase phenocryst;uenched scoriae sample from the 1913 scoriae-fall deposit, collected at 19.5980°N,seta”where visitors are registered. (B) Enlarged view of a part of Fig. 2A. (C) Sample Col-ter erosion in the Playón. (D) Enlarged view of a part of Fig. 2C. (E) Sample Col-04A, a lava0-05-05, a distal bulk ash from the 10 May 2005 eruption. (H) Enlarged view of a part ofw of a part of Fig. 2.
Fig. 3 (continued).
247I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
electron imaging, both of which were composites of dark, fine-grainedgroundmass and an attached plagioclase phenocryst. The groundmasstextures of these two ash particles, thought to be similar based onexamination by petrographic microscope, proved to be quite differentwhen examined through backscattered electron imaging.
The groundmass texture of the selected ash particle from sample10-05-05 (Fig. 3G and H), erupted on 10 May 2005, is actually ex-tremely rich in microlites, whose combined optical characteristicsapparently produce the darkness seen in both plane-light andcrossed-polarized light imaging of 30-μm-thick sections by petro-graphic microscope. Glass (roughly estimated at 10 vol.%), shows asmall number of vesicles (up to 20 μmacross) and has been stripped ofFe and other elements of high atomic number by crystallization ofpyroxene and titanomagnetite microlites. As a result the glass meanatomic number is below that of plagioclase; the plagioclase istherefore brighter than the glass in backscattered-electron imaging,the only such example among the six particles shown in Fig. 3. Thisash particle appears to represent fresh magma that cooled relativelyslowly, perhaps in the upper conduit or a dome. An importantquestion is: how representative of the 10 May 2005 eruption is the“juvenile” ash particle that we studied? We imaged 4 ash particlesusing backscattered electrons, and all textures were the same, but weconsider this important question to be currently unresolved. The ashparticle from sample VF04 (Fig. 3I and J), erupted on 6 June 2005, is thesame one shown in Fig. 2A–C. Its groundmass displays a few smallvesicles and is otherwise most similar to the 2004 lava block (Fig. 3Eand F) with regard to glass abundance (glass roughly estimated at 60%)and to glass and plagioclasemean atomic numbers and backscattered-electron gray-scale colors. We investigated only 4 ash particlesthought to be “juvenile” from VF04 based on petrographic analysis.One other ash particle had a similar groundmass texture, but the other
two ash particles had glass-poor groundmass textures similar to thatdisplayed by the 10 May 2005 particle (Fig. 3G and H).
6.2. Hornblende-groundmass interfaces
Hornblende is a water-bearingmineral, with about 2 wt.% H2O in itsstructure. Laboratoryexperimentshave shown thathornblende canonlybe stable in silicatemelts that contain3–4wt.%H2O (Rutherford andHill,1993; Moore and Carmichael, 1998, Carmichael, 2002 and referencestherein). An andesiticmagmawith3–4%water in themelt arriving in theupper-crustal conduit systemwould begin to explosively vesiculate anderupt to form vesiculated scoriae (or pumice) and ash, resembling theproducts of the 1818 and 1913 VEI-4 explosive eruptions. Hornblendecrystals in rapidly quenched scoriae and pumice samples show a cleaninterface between the hornblende and the micro-vesiculated glassygroundmass (Fig. 3A and B), demonstrating that hornblende remained astablemineral up to the pointof explosive quenching.Whenhornblendeis found in lava flows, loss of magmatic water prior to eruption is clearlyrecorded in the disequilibrium rim textures of the hornblende crystals(see Fig. 5B in Luhr, 2002). In thin section, dark reaction rims arecharacteristic of hornblende crystals in lava samples, representingconversion of the water-bearing hornblende crystal to a fine-grainedassemblage of water-free minerals. Recent experimental studies haveattempted to quantify the rate of reaction-rim growth as functions ofmelt composition, temperature, and pressure to evaluate ascent rates ofmagmas once they leave the hornblende stability field (Rutherford andHill, 1993; see also Carmichael, 2002). No hornblende crystals wereobserved either among the “juvenile” or non-juvenile ash particlesstudied from the 2005 explosive eruptions, consistent with the paucityof hornblende in all products erupted after the 1961–1962 period (Luhrand Carmichael, 1980, 1990; Luhr, 2002).
Fig. 4. Whole-rock compositional plots versus eruption year: (A) SiO2; (B) TiO2; (C) MgO; (D) Ba and (E) Rb. Where shown the vertical bars of the crosses in a corner of each plotindicate the ±1σ estimates of analytical precision discussed in Luhr (2002). With white squares we indicated the bulk rock composition from volcanic domes extruded in 1810 and1869 (Volcancito dome) (Savov, Varley and Watts—work in progress). Data for the domes and for all volcanics erupted between 1999 and 2005 are from this study (Table 2) (see alsoAppendix A). Data for previous years are from Luhr and Carmichael (1980, 1990) and Luhr (2002).
248 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
6.2.1. Whole-rock major element compositionThe Loss on Ignition (LOI) values in the newly analyzed Colima
volcanic rocks range from −0.04 to 0.4%; average LOI=0.13% (in theColima dome samples the range is similar: −0.04 to 0.36%; averageLOI=0.19% — see Appendix A). The measured LOI values show nosystematic variations with increasing SiO2 K2O, Na2O, MgO or Al2O3.The 8 new whole-rock lava analyses presented in Table 2 for theperiod mid-1999 to late-2004 and the newly analyzed 12 whole-rocksamples from volcanic domes extruded between 1818 and 1869 (seeAppendix A and B) are combinedwith earlier published data (Luhr and
Carmichael, 1980, 1990; Luhr, 2002) on time-series plots in Fig. 4. Theparameters for these plots were chosen to represent different types ofchemical behavior through time in whole-rock Colima productserupted since the 1818 eruption i.e. the start of our ability to relateeruptive products on the ground with known eruption dates. SiO2 isshown in Fig. 4A, because it is the most abundant major element,provides a measure of magma viscosity, and allows the more mafic(lower SiO2) scoriae erupted in 1913 to be distinguished from all lavasamples that preceded or followed it. Lavas erupted in 1869, 1880,1961–62, and 1975were individually quite homogeneous in SiO2, with
Fig. 5. Whole-rock low abundance trace element concentrations in Volcán Colima scoriae, lava and dome samples. Note the position of low SiO2 and high MgO sample Col 1B:A) Average chondrite-normalized Coryell-Masuda diagram showing the REE concentrations for lava samples erupted/extruded between 1810 and 2005. Chondrite REE abundancesare after Nakamura (1974). Note the identical REE patterns between 1913 and the rest of the samples, indicating similar melting/crystallization and/or assimilation histories.B) Whole-rock chondrite-normalized [La/Sm] ratios versus eruption year. Chondrite La and Sm abundances are after Nakamura (1974). Note the similarity in the [La/Sm], indicatingsimilar magmatic source; C) Whole-rock Ba/La ratios versus eruption year. Ba behaves as a fluid-mobile element in respect to the light rare earth element La (Savov et al., 2005). Thelarge span and the higher Ba/La ratios in the 1818 and 1913 scoriae in comparisonwith the recent volcanic products reflects the higher volatile contents of these magmas upon arrivalin the Colima shallow conduit system (and thus the higher tephra volumes and spread of the 1818 and the 1913 ash associated with these eruptions). D) Whole-rock Ba/Th ratiosversus eruption year. Th is usually highly enriched in the continental crust and is fluid-immobile element (Savov et al., 2005). Note the very large variation but overall higher Ba/Thratios in the 1818 and 1913 samples and the trend toward increasing Ba/Th ratios associatedwith 1961–2005 samples that have higherMgO and SiO2 shown on Fig. 4 (sample Col-1A).
249I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
total variations for those respective eruptions as follows: 1.16, 1.07,0.51, and 0.48 wt.%. Analyzed scoriae from the major Plinian explosiveeruption of 1913 are slightly less homogeneous in SiO2, with a totalvariation of 1.33 wt.%. (it must be noted that Robin et al., 1991, arguedfor a still broader range of 1913 whole-rock compositions, which wehave not been able to confirm in our studies to date).
Beginning with andesitic block-lava erupted in 1975–76, andcarrying through the rest of the observational period until 2005, wefindwider variations trending toward lower whole-rock SiO2 (Fig. 4A).The total SiO2 variation since 1976 is 2.03, which is 1.5 to 4 times thevariations seen in any of the previous historical eruptions. In part thisgreater whole-rock compositional variation must be attributed to themuch more detailed sampling that we have for the current eruptivecycle.
The most important observation from Fig. 4A is that whole-rockSiO2 contents of samples erupted since 1976 have shown widevariability trending toward the low SiO2 values of 1913 scoriae. Someminor overlaps between these two populations are present; the 1976–2004 samples with the lowest SiO2 are Col-30 (1982), Col-99E (1999),and Col-1B (2001). Two of these most recent samples are labeled on
Fig. 4A and are highlighted in discussion of other Fig. 4 plots. Theseand other whole-rock lava samples with low SiO2 contents appear torecord spikes of more-mafic magma injected into the sub-volcanicmagma reservoir system beneath Volcán Colima. Similar widevariations in whole-rock compositions may have occurred duringeruptions in the late 1800s to 1909, but no samples are available toconfirm this speculation.
A similar, and even cleaner separation of the 1913 scoriaecompositions from the lava compositions that preceded and followedis shown by the time-series plot of TiO2 in Fig. 4B; CaO (not shown)exhibits similar behavior as well.
For other elements in time-series plots, patterns aremore complex.Fig. 4C shows the eruption year versus whole-rockMgO. As in the casefor SiO2, compared to the 1976–2004, the 1869, 1880, 1961–62, and1975 lavas, and the 1913 scoriae, show smaller variations (0.4–0.7 wt.%MgO). Much larger variations dominate the period 1976–2004(1.56 wt.% MgO), and these overlap and exceed the highest values ofMgOmeasured for 1913 scoriae; Cr and Cu (not shown) exhibit similarpatterns. The sampleswith the lowest SiO2 contents (Fig. 4A) also havethe highest MgO contents (Fig. 4C).
Table 3Electron-microprobe analyses of pyroxene rims and estimates of pre-eruptivetemperature based on 2 Pyroxene geothermometers
Fe, Mg and Ca values are mol fractions of those cations. Mg #=100×(Mg/Mg+Fe).Two-pyroxene geothermometers list temperatures calculated from the pyroxene-rimcompositions based on formulations by:T-Quilf (Andersen et al., 1993).T-BM (Bertrand and Mercier, 1985) T-W (Wells, 1977).T-WB-Temperature (Wood and Banno, 1973).
Table 4Electron-microprobe analyses of plagioclase rims and groundmass
Melt H2O (wt.%) is calculated from the formulation of Putirka (2005) based on coexistingplagioclase rim and groundmass compositions, assuming the 2-pyroxene temperaturescalculated from the formulation of Bertrand and Mercier (1985) as listed in Table 3.
250 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
6.2.2. Whole-rock trace element compositionsTime-series data for whole-rock large ion lithophile elements Ba
and Rb are shown in Fig. 4D and E. The observed trends for Ba and Rbmimic those for SiO2 and are also representative for K, Sr, Hf, Pb, Th, andU (not shown). Again we see evidence that the tight compositionalgroupings present through 1975 begin to collapse toward widercompositional ranges during 1976–2004. The samples that are poorestin SiO2 and richest inMgO, are also poorest in Ba and Rb (Fig. 4D and E).
Plotted on Coryell-Masuda diagram, the chondrite-normalized rareearth element (REE) abundances in the 1913 (and 1818) and the morerecently erupted samples show nearly identical patterns (Fig. 5A). TheREE abundances of erupted samples are also identical with those of thevolcanic dome samples (Fig. 5A, see also Appendix A). Volcán Colimasamples all have relatively flat heavy rare earth element (HREE) patterns(7.5–10×chondrites) and enriched light rare earth element (LREE)contents (25–40×chondrites), which usually are attributed to subduc-tion-related LREE-enriched melt-modified mantle wedge in originmagmas (Luhr, 1997, 2002; Savov et al., 2005). This is confirmed bythe similar [La/Sm]N ratios in the1913deposits and the samples from the1999–2004 period, all of which show very narrow ranges, usually in therange 2.4 to 2.8 (Fig. 5B). Although all the bulk rock volcanics with highquality REE measurements (n=71) contain feldspar [both phenocrysts
(N0.3 mm) and/or microphenocrysts (b0.3 mm)], large Eu anomalieswere not identified in the studied samples (Fig. 5A). Keyelemental ratiosused in subduction zone studies (La/Sm; Ba/La and Ba/Th) are shown onFig. 5B–D and discussed later in the text.
7. Mineral and groundmass compositions
For the three samples analyzed by electron microprobe, we selectedminerals that showed growth rims against fine-grained to glassygroundmass. We analyzed the compositions of orthopyroxene, clin-opyroxene, and plagioclase rims (outer 10 μm) on numerous crystals,with mean values listed in Tables 3 and 4. We also used a movingdefocused electron beam to analyze various patches of fine-grained toglassy groundmass adjacent to these same crystals, with mean valueslisted inTable 4. The goal of these efforts is to record the compositions ofthis quench-equilibrium phase assemblage so that geothermometersand algorithms to calculate melt-water content can be applied. Thisapproach continues the time line established for Colima since 1869based on samples studied by Luhr (1992, 2002), as indicated by plots ofvarious measured and derived parameters versus time in Figs. 5 and 6.
The mean values of Mg# in orthopyroxene and clinopyroxene rimsare shown in Fig. 6A and B. A problem with this approach for pyrox-enes is that variability in pyroxene-rim compositions for a polishedsection of a single sample (shown by representative error bars inFig. 6) is comparable with the total variations observed for mean rimcompositions among all analyzed samples. Four different 2-pyroxenethermometers were evaluated using the pyroxene-rim compositionsas listed in Table 3. Derived temperatures range up to 40 °C for a singlesample. We focus on the temperatures estimated from the model ofBertrand and Mercier (1985), because only that model yieldstemperatures b950 °C for most (5 of 6) of the hornblende-richandesites erupted in 1869–1913; phase-equilibrium experiments haveshown that the maximum thermal stability limit for hornblende in
Fig. 6. Average pyroxene-rim compositions (Mg#=100×Mg/(Mg+Fet)) plotted versuseruption year. Error bars show average 1σ estimates for the three samples in Table 3.(A) orthopyroxene rims; (B) clinopyroxene rims. Data for 1999–2005 are from this study(Table 3). Data for previous years are from Luhr (1992, 2002). (C) Rim equilibrationtemperatures calculated from the 2-pyroxene thermometers of Wells (1977), Wood andBanno (1973), Bertrand and Mercier (1985) and from groundmass — Plagioclase rimthermometer of Putirka (2005).
Fig. 7. Average plagioclase rim (An%=100×Ca/(Ca+Na)) and groundmass compositionsplotted versus eruption year. Error bars show average 1σ estimates for the threesamples in Table 4. (A) plagioclase rims; (B) groundmass. Data for 1999–2005 are fromthis study (Table 4). Data for previous years are from Luhr (1992, 2002). (C) Wt.% H2O inthe melt required to match the pyroxene-rim temperatures (Bertrand and Mercier,1985) listed in Table 3 based on the plagioclase-melt model of Putirka (2005).
251I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
andesites is ∼950 °C (Allen et al., 1975). Fig. 6C shows the variouspyroxene rim and plagioclase rim-groundmass temperatures in time-series. These results indicate that increases in whole-rock MgO since1976 (Fig. 4C) have been accompanied by higher pyroxene-rim andplagioclase rim-groundmass temperatures.
The plagioclase rims from the 1913 samples (Fig. 7A) have higherAn contents compared to all other analyzed samples. Between 1975and 1982 a significant increase in mean plagioclase rim An composi-tions occurred (from [47.2–51.0 mol%] it increased to 54.8 mol%). Since1982, which corresponded with the first of the three mafic spikessince 1976, plagioclase rim An contents have remained high butshifted to progressively lower values over time (Fig. 7A). Unfortu-
nately, neither of the two subsequent samples recording mafic whole-rock spikes (Col-99E and Col-01B), discussed above for Fig. 4, wereanalyzed for mineral and groundmass compositions.
The time-series trend for the mean groundmass compositions isrepresented by the SiO2 vs. eruption dates plot shown in Fig. 7B. Thetrend from 1869 to 1913, although sparsely populated and showinghigh variability for the two samples analyzed from the 1869 and 1880lavas, displays a significant lowering of groundmass SiO2 over time. In1961, groundmass SiO2 values returned to the same high levels (71–72 wt.%) as seen in one sample each from 1869 and 1880 (Fig. 7B).Since 1961 there has been a steady and significant shift to lowergroundmass SiO2, heading toward the still lower values recorded for1913 scoriae. Groundmass SiO2 strongly reflects the total abundance of
252 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
crystals in the magma, increasing as proportion of crystals increases.Crystal abundance in turn is strongly influenced by both magmatictemperature and melt water, decreasing as these two parametersincrease. Thus, decreases in groundmass SiO2 with time, as seen inFig. 7B, might reflect increases in either magmatic temperature ormelt-water content.
Melt-water contents were calculated from plagioclase rim andgroundmass pairs (Table 4), using the algorithm of Putirka (2005) byadjusting themelt-water content until the calculated plagioclase-melttemperaturematched the value derived from the Bertrand andMercier(1985) pyroxene thermometer (Table 3). We have not attempted toestimate errors for melt-water content. The 1869 lavas and the 1913scoriae have the highest overlapping calculated water contents (4.5–6.0 wt.%). The 1880 lava has lower calculated water contents (2.5–4.5 wt.%), in the same range as the lavas erupted during 1961–2004.
It is important to note that all lavas and scoriae erupted during1869–1913 are very rich in hornblende (2.4–5.3 vol.%) compared to allproducts erupted since 1961–1962 (0 to 1 vol.%). As discussed above,hornblende in andesitic magmas has a maximum thermal stabilitylimit of ∼950 °C. Another constraint on the stability of hornblende isimposed by a minimum required amount of water in the co-existingmelt, thought to be ∼3–4 wt.% (Moore and Carmichael, 1998). Thus,hornblende might be absent or of minor abundance if the magmatictemperature is N950 °C or if the melt-water content is b3 wt.%. Thescarcity of hornblende in the 1961–1975 lavas is difficult to under-stand, because both the calculated temperatures (Fig. 6C) and melt-water contents (Fig. 7C) are similar to those for the hornblende-rich1880 lavas. In lavas erupted after 1975, elevated temperature (Fig. 6C)is the most likely explanation for the scarcity of hornblende.
8. Discussion and implications
8.1. Volatile contents of melt inclusions and fluid-sensitive trace elementratios confirm the wet nature of 1913 magmas
Recent ion-probe measurements of melt inclusions trapped inplagioclase and orthopyroxene crystals from the 1869 and 1880eruptions reveal very low melt H2O contents (0.39–0.72 wt.%) (Atlaset al., 2006). Low H2O contents of melt inclusions trapped inorthopyroxenes (0.2 to 2.3 wt.%) and clinopyroxenes (1.1 to1.6 wt.%)from the 1998–2005 eruption period of Colima were also reported inRuebi and Blundy (2007a,b). Ion-probe measurements of meltinclusions trapped in pyroxenes from alkaline cinder cone volcanicsshowmelt H2O concentrations averaging ∼2 wt.% (i.e. pre-1913 levels;Luhr and Savov — unpublished). In contrast, ion-probe study of meltinclusions trapped in orthopyroxenes from the Plinian VEI 4 1913eruption of Colima reveal consistently very high H2O contents(average=5.7 wt.%, maximum=6.2 wt.%) (Luhr et al., 2006). Finally,study of 2 melt inclusions trapped in clinopyroxenes from the 1913eruption by Atlas et al. (2006) confirms the high H2O contents of the1913 magmas (∼3.4 wt.%).
In addition to the melt inclusion studies, based on severalsignificant spikes in the fluid-mobile/fluid-immobile trace elementratios such as Ba/La and Ba/Th it appears that the trace elementsystematics confirm the wetter nature of the 1818 and 1913 magmasin respect to the more recent (1961–2005) Colima samples (see Fig. 5Cand D where 1913 samples reach the highest levels of Ba/La and Ba/Thamong all post 1818 Colima volcanic products). There is also a con-tinuous trend toward increasing Ba/La and Ba/Th since 1962 andindeed the 2001–2005 samples show the highest Ba/Th and Ba/Laratios among all post-1960 samples (Fig. 5C and D).
8.2. Evaluating the effect of possible (basement) assimilation
Recent study by Ruebi and Blundy (2007a,b) reported micron-to-millimeter in size gabbroic “xenoliths” in some of their thin sections
from Colima. These authors showed that the major element composi-tion of the majority of the bulk rock Colima volcanics (mostlyintermediate in composition) plot on a mixing curve between thecompositions of plagioclase- and pyroxene-hosted melt inclusions (allwith high SiO2) and the composition of their gabbroic “xenoliths”(with low SiO2). This observation, along with the statement thatColima does not erupt basalts, led Ruebi and Blundy (2007a,b) toconclude that the intermediate products (andesites) at Colima (andother volcanoes) represent mixtures between mafic igneous volcanicbasement and high SiO2 magma. This raises the question if variabledegrees of basement assimilation could affect the magma composi-tional trends (including volatile budgets) and possibly explain thedramatically different eruption behavior of Colima in 1818 and 1913?
The purpose of this paper is not to test the hypothesis of Ruebi andBlundy (2007a,b). However, we stress the fact that there exists aplethora of recent literature, which shows the origin of the Colimavolcanics as derived from source regions within the mantle wedgecontaining hornblende and/or phlogopite, and no crustal inputs (seeLuhr, 1997 and Carmichael, 2002). For more detailed discussion onthese issues the reader is directed to Appendix C.
8.3. Magmatic gas retention: evidence from textures
Of the lava, scoriae, and ash samples that we have investigatedfrom eruptions of Volcán Colima during 1869–2005, only the scoriaefrom 1913 show micro-vesiculated groundmass textures and showclean hornblende rims adjacent to the groundmass. We infer that onlythe 1913 magma retained a significant fraction of its original volatilecontent to the point of frothing, fragmentation, and explosive (VEI 4)eruption in the volcanic conduit.
All other studied samples of lava and ash particles show weakly tonon-vesiculated groundmass textures. If hornblende is present itshows a strong reaction corona against the groundmass, indicating aperiod of late-stage mineral growth at magmatic temperature withoutsufficientmagmatic water to sustain hornblende growth.We concludethat these lava and ash samples erupted during 1869–1880 and 1961–2005 represent arrival of degassed magma in the upper volcanicconduit or the summit crater. Absent the ability to power explosiveeruptions driven by expansion of magmatic gas, these degassedmagmas presumably erupted through one of three mechanisms.
Lava domes in the crater and block-lava flows that they feed overthe crater rim and down the flanks of the upper cone must reflectascent of new degassed magma from below that rises buoyantly fromdepth and pushes earlier degassed magma up into the summit craterand down the flanks.
Small to intermediate Vulcanian-style explosive eruptions since1998 have sent vertical ash-and-gas plumes as high as 11 km a.s.l. andsent pyroclastic flows out to distances of 5 km (Table 1). Ash particlesfrom these eruptions show only minor vesiculation and include twofundamentally different types, with transitions. On the one hand areash particles with microlite-rich groundmasses, which we interpret asfragments of lava plugs/domes in the upper conduit, that crystallizedrelatively slowly. On the other hand are ash particles with dark andfine-grained groundmasses, which we interpret as hot degassedmagma, newly arrived in the upper conduit, which was explosivelyfragmented by the Vulcanian-style eruptions. The fluid for theseexplosive eruptions might have come from two sources, one internal,the other external.
The possible internal source is magmatic gas, evolved from deeperdegassing magmas, that rises ahead of the magma from which itemerged. Prior to many recent explosive eruptions at Volcán Colima,new lava domes were observed (May and June 2005) in the summitcrater. These degassed lava domes may temporarily plug the conduitand allow build-up of the magmatic gases streaming upward frombelow. Eventually, the gas pressure may exceed the strength of theplug, resulting in an explosive Vulcanian-style eruption.
253I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
The alternative scenario, involving an external fluid source, is thatthese eruptions are phreato-magmatic, resulting from interaction ofhot degassed magma and surrounding hot solid rock with ground-water. Distinguishing between these two mechanisms for theexplosive eruptions is possible. It requires various types of data: (1)quantitative time-series estimates for gas flux from the summit crater,which should decrease prior to explosive eruptions if conduit pluggingoccurs; (2) quantitative estimates of the SO2 release from explosiveeruptions, which should be higher if the gas is fundamentallymagmatic as compared to groundwater-derived; and (3) quantitativetime-series data for rainfall on the volcano, whose maxima might beexpected to precede the explosive eruptions if they are phreato-magmatic in nature.
8.4. Significance of results for future eruptive activity
This study aims to provide some baseline data for the discreteexplosive eruptions that have occurred since Feb. 1999, whichappear to be similar to the many explosive eruptions that occurredfrom the late 1880s until 1909, preceding the major 1913 explosiveeruption of relatively mafic (58.5 wt.% SiO2), micro-vesiculatedscoriae. Since 1976, whole-rock lava compositions have beenrelatively heterogeneous and mafic, with notable mafic spikes in1982, 1999, and 2001 that reached MgO contents higher than thosein any 1913 scoriae. A pronounced decrease in groundmass SiO2
content since 1961 is consistent with a trend toward SiO2-poor1913-type magma, but may actually reflect increased magmatictemperature since 1961 (Fig. 6C). Most importantly, no samples
Appendix A
Whole-rock lava analysis of Colima volcanic dome samples by DCP-AES and ICP-MS
erupted since 1913 have shown micro-vesiculated groundmasses;all magmas currently arriving in the upper volcanic conduit systemappear to be degassed. The lavas and ash erupted during 1961–2005are very poor in hornblende, compared to the hornblende-rich lavasof 1869–1880, preceding the 1913 eruption. This likely reflects acombination of highermagmatic temperatures and lowermagmaticwater contents, even prior to ascent-related degassing, for 1961–2005 magmas.
Acknowledgments
We wish to thank Terry Plank for analyzing the Colima domesamples for low abundance trace elements in her ICP-MS lab at BostonUniversity and Jeff Ryan for allowing access to his B-free lab and DCPspectrometer at the University of South Florida. We are grateful toNick Varley, Juan Carlos Gavilanes, Santiago Hidding, (UniversidadColima), Peter Schaaf (UNAM) and Rob Watts (Universidad de PuertoRico) for providing some of the lava, dome and ash samples used inthis study. Tim Gooding, Tim Rose, and Marc Lippella (all from theSmithsonian Institution) provided assistance in sample preparationand instrumental analysis. We thank Bruce Marsh and Zachary Atlasfor thoughtful reviews of ourmanuscript. Thismanuscript waswrittenduring IPS postdoctoral stay at the Department of TerrestrialMagnetism of the Carnegie Institution of Washington DC.
Finally, we are bitter from the early departure of our mentor andfriend Jim Luhr and hope that the approached we discussed in thispaper will be used by the future generations of petrologists interestedin volcano monitoring via ash geochemistry.
254 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
Appendix C. Evaluating the effect of possible (basement) assimilation
Based on the dataset reported here, we believe that the similarcrust-sensitive element concentrations and isotope ratios for allColima samples (Fig. 5A-D and also Table 2) points toward non-crustalsources for volcan Colima magmas. Below we list several argumentsfor lack of volumetrically significant igneous basement assimilation atvolcan Colima:
1) REE and HFSE systematics- due to their immobility in fluids (Savovet al., 2005) and their sensitivity to crustal assimilation (feldspars,garnet and hornblende all possess very distinct signatures of theseelements), the nearly identical HFSE and REE (see Fig. 5A) in all
Colima rocks indicate similar degree of source melting and extentof metasomatic melt-peridotite interactions (see Luhr, 1997). Itwould be hard to explain long-term compositional similarities at asingle volcanic complex if the bulk of its lavas were created bymechanical mixing of diverse basement lithologies (limestones,altered andesites, shales, quartzites) and uprising lavas.
2) Crust sensitive light element concentration: Boron has been shownto be highly sensitive to volcanic basement assimilation processes(Savov and Leeman, 2007). All Colima samples have low Bconcentrations (B~5 ppm vs. continental crust with B≫10 ppm)and low B/Be ratios (B/Be=2–9 vs. Aleutians and Central America arcbasement with B/BeN100; see Hochstaedter et al., 1996), indicatinglittle or no role for volcanic basement in the magmagenesis.
255I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
3) Stable isotope ratios: All volcanic rocks from Colima have non-crustal light element isotope ratios (δ11B+2‰ to +10‰; δ7Li~+5‰)(Savov et al., 2007). In comparison, the continental crust possessesconsistently lower Li isotope ratios (δ7Li~0‰)(see Teng et al.,2004) and crustal in origin high-silica obsidians from theWestern Mexican Volcanic Belt have δ 11B as low as -10‰ (Savovunpublished).
4) Radiogenic isotope ratios: All volcan Colima rocks have non-crustal206Pb/204Pb, 207Pb/204Pb, 87Sr/86Sr (~0.7036), Nd (4.3 to 6.2) and Hf(5–10) (Luhr 1997; Savov et al., 2007). Moreover, the calc-alkalineColima andesites do not plot on mixing lines of any kind as long asthey involve local basement lithologies reported in Luhr (1997).
5) Plagioclase mineral chemistry: Luhr and Haldar (2006) describe inconsiderable detail the potential effect of addition of feldspar-containing xenoliths (in their case troctolites) to arc mafic magmasfrom the Barren Island volcano in the Indian Ocean. These authorspostulate that such mixing would produce large positive Euanomalies and clear mixing relationships between the plagioclasemodes and bulk rock Al2O3 content. The lack of Eu anomalies(Figure 5A) and the lack of plagioclase-bulk rock Al2O3 correlationsin all Colima rocks (not shown) indicate the absence of feldsparassimilation. In addition, it has been shown experimentally thatincrease in melt H2O contents (and Al/Si and Ca/Na) producesplagioclase crystals with elevated anorthite contents (Sisson andGrove, 1993; Kohut and Nielsen, 2003). Interestingly, the highanorthite content of Colima plagioclase crystals from 1913 and1818 samples correspond to the samples with the highestcalculated melt H2O content and the highest measured meltinclusion H2O abundances (Luhr et al., 2006). Thus it appears thatthe Colima plagioclase crystals crystallized in equilibriumwith thesurroundingmelts, arguing against secondary addition of feldsparsduring basement assimilation.
6) Volume issues: The large volumes of the old (Nevado de Colima)and the current (Fuego da Colima) stratocones and thewidespread,and sometimes significant in thickness, Pleistocene to recentvolcanic deposits around Volcan Colima (Saucedo-Girón, 1997;Bretón et al., 2002) indicate that if the eruptive products aremixtures of dacitic magmas and assimilated igneous basement-than both of these source materials should be fairly widespreadnear the volcanic edifice. Currently there exist no geophysicalevidence for the existence of large plutonic body immediatelyunderlaying volcan Colima and the majority of erupted lavas arenot dacitic, once again arguing against assimilation on a largescales.
7) Presence of basalts: Although basalts are rare in thewestern part ofthe Mexican Volcanic Belt, many of the erupted basalt-andesites inthis continental arc setting have high MgO (up to 10 wt %) andcould contain olivine phenocrysts with Fo88–92 i.e. with mantlewedge origin (Carmichael, 2002). We accent on the fact that,although rare, low SiO2 basalt samples are, indeed, present atvolcan Colima (for example see basalt sample SAY 22E reported inLuhr and Carmichael, 1990) and often they have N9 wt% bulk rockMgO contents.
In summary, based on the currently available data, it seems thatvolcan Colima products show only minor differences in their crust-sensitive bulk rock trace element and stable and radiogenic isotopesystematics (especially the REE). However, further studies arenecessary to test the possible relationships between the Cretaceousto Tertiary batholith-size (~160 km long) granodiorite plutonsoutcropping in the East and West side of the Colima Graben (fromPuerto Vallarta in [Jalisco State] to Manzanillo [Colima State]) and theintermediate andesite magmas erupting at volcan Colima. For nowwebelieve that the basement plays no role in the composition of the post1818 volcan Colima rocks and that different volatile contents of 1913versus post-1913 Colima volcanics are controlled by the relative speed
of magma emplacement and the processes of shallow degassing/cooling and conduit crystallization.
References
Allen, J.C., Boettcher, A.L., Marland, G., 1975. Amphiboles in andesite and basalt: I.Stability as a function of P-T-fO2. American Mineralogist 60, 1069–1085.
Allan, J.F., 1986. Geology of the northern Colima and Zacoalco grabens, southwestMexico; Late Cenozoic rifting in the Mexican Volcanic Belt. Geological Society ofAmerica Bulletin 97, 473–485.
Allan, J.F., Nelson, S.A., Luhr, J.F., Carmichael, I.S.E., Wopat, M., Wallace, P.J., 1991.Pliocene–recent rifting in SW Mexico and associated volcanism: an exotic terranein the making. In: Dauphin, J.P., Simoneit, B.A. (Eds.), The Gulf and PeninsularProvince of the Californias. American Association of Petroleum GeologyMemoir, 47,pp. 425–445.
Atlas, Z.D., Dixon, J.E., Sen, G., Finny, M., Martin Del Pozzo, A.L., 2006. Melt inclusionsfrom Volcán Popocatépetl and Volcán de Colima, Mexico: melt evolution due tovapor-saturated crystallization during ascent. Journal of Volcanology and Geother-mal Research 153, 221–240.
Andersen, D.J., Lindsley, D.H., Davidson, P.M., 1993. QUILF: a PASCAL program to assessequilibria among the Fe–Mg–Mn–Ti oxides, pyroxenes, olivine, and quartz.Computers and Geoscience 19, 1333–1350.
Bertrand, P., Mercier, J.C.C., 1985. The mutual solubility of coexisting ortho- andclinopyroxene: toward an absolute geothermometer for the natural system? Earthand Planetary Science Letters 76, 109–122.
Breton, M., Ramirez, J.J., Navarro, C., 2002. Summary of the historical eruptive activity ofVolcán de Colima, México: 1519–2000. Journal of Volcanology and GeothermalResearch 117, 21–46.
Carmichael, I.S.E., Frey, H.M., Lange, R.A., Hall, C.M., 2006. The Pleistocene cinder conessurrounding Volcan Colima, Mexico re-visited: eruption ages and volumes,oxidation states, and sulfur content. Bulletin of Volcanology 68, 407–419.
Carmichael, I.S.E., 2002. The andesite aqueduct: perspectives on the evolution ofintermediate magmatism in west-central (105–99°W) Mexico. Contribuitons ofMineralogy and Petrology 143, 641–663.
Diaz, S., 1906. Efemerides del Volcán de Colima, observaciones diarias desde losObservatorios de Zapotlan y Colima de 1893 a 1905. Sria. De Fomento.
Hochstaedter, A.G., Ryan, J.G., Luhr, J.F., Hasenaka, T., 1996. On B/Be ratios in the MexicanVolcanic Belt. Geochimica Cosmochimica Acta 60, 613–628.
Kelley, K.A., Plank, T., Ludden, J., Staudigel, H., 2003. Composition of altered oceaniccrust at ODP Sites 801 and 1149. Geochemistry Geophysical Geosystem 4, 6 paper2002GC000435.
Luhr, J.F., Carmichael, I.S.E., 1980. The Colima volcanic complex, México: Part I. Post-caldera andesites from Volcán Colima. Contributions to Mineralogy and Petrology71, 343–372.
Luhr, J.F., Carmichael, I.S.E., 1990. Petrological monitoring of cyclical eruptive activity atVolcán Colima, México. Journal of Volcanology and Geothermal Research 42,235–260.
Luhr, J.F., 1992. Slab-derived fluids and partial melting in subduction zones: insightsfrom two contrasting Mexican volcanoes (Colima and Ceboruco). Journal ofVolcanology and Geothermal Research 54, 1–18.
Luhr, J.F., 1997. Extensional tectonics and the diverse primitive volcanic rocks in theWestern Mexican Volcanic Belt. Canadian Mineralogist 35, 473–500.
Luhr, J.F., 2002. Petrology and geochemistry of the 1991 and 1998–1999 lava flows fromVolcán Colima, México. Journal of Volcanology and Geothermal Research 117,169–194.
Luhr, J.F., Haldar, D., 2006. Barren Island Volcano (NE Indian Ocean): island-arc high-alumina basalts produced by troctolite contamination. Journal of Volcanology andGeothermal Research 149, 177–212.
Luhr, J.F., Navarro, C., Connor, C.B., Connor, L., 2006. The 1913 VEI-4 Plinian eruption ofVolcan de Colima (Mexico): tephrochronology, petrology, and plume modeling. EosTransaction AGU 87 (52) Fall Meet. Suppl., Abstract V43B-1786.
Moore, G.M., Carmichael, I.S.E., 1998. The hydrous phase equilibria (to 3 kbar) of anandesite and basaltic andesite from western Mexico: constraints on water contentand conditions of phenocryst growth. Contributions to Mineralogy and Petrology130, 304–319.
Mora, J.C., Macias, J.L., Saucedo, R., Orlando, A., Manetti, P., Vaselli, O., 2002. Petrology ofthe 1998–2000 products of Volcán de Colima, México. Journal of Volcanology andGeothermal Research 117, 195–212.
Nakamura, 1974. Determination of REE, Ba, Fe, Mg, Na and K in carbonaceous andordinary chondrites. Geochimica Cosmochimica Acta 38, 757–775.
Navarro, C., Cortes, A., 1993. Eruptive behavior of Colima volcano previous to Pelean-type eruptions with presence of active domes at the summit during XIX & XXcenturies. An evaluation of its volcano hazard. Abstracts. First internationalworkshop on environmental volcanology, 6, 15–17 November, Morelia Michoacan,Mexico.
Newhall, C.G., Self, S., 1982. The volcanic explosivity index (VEI): an estimate ofexplosive magnitude for historical volcanism. Journal of Geophysical Research 87,1231–1238.
Pardo, M., Suerez, G., 1993. Steep subduction geometry of the Rivera Plate beneath theJalisco Block in western Mexico. Geophysical Research Letters 20, 2391–2394.
Putirka, K.D., 2005. Igneous thermometers and barometers based on plagioclase+ liquidequilibria; tests of some existing models and new calibrations. AmericanMineralogist 90, 336–346.
256 I.P. Savov et al. / Journal of Volcanology and Geothermal Research 174 (2008) 241–256
Robin, C., Camus, G., Gourgaud, A., 1991. Eruptive and magmatic cycles at Fuego deColima volcano (Mexico). Journal of Volcanology and Geothermal Research 45,209–225.
Reubi, O., Blundy, J., 2007a. Volcán de Colima, an andesitic volcano fed by a daciticreservoir. Is that typical of continental arc magmatism? Geochimica CosmochimicaActa 834 Goldschmidt Conference Abstracts, A.
Reubi, O., Blundy, J., 2007b. Formation of Andesites by Infiltration of Mafic Magmas in aCrystallizing Tonalitic Pluton at Volcán de Colima, Mexico. State-of-the-Arc(SOTA)'07 Meeting Volume, pp. 200–203.
Rutherford, M.J., Hill, P.M., 1993. Magma ascent rates from amphibole breakdown: anexperimental study applied to the 1980–1986 Mount St. Helens eruptions. Journalof Geophysical Research 98, 19667–19685.
Saucedo-Giron, R., 1997. Reconstrución de la última erupción explosiva del Volcán deColima in 1913. Master's Thesis, Instituto de Geoísica, Universidad NacionalAutónoma de México, 185 pp.
Savov, I.P., Ryan, J.G., D’Antonio, M., Kelley, K., Mattie, P., 2005. Geochemistry ofserpentinized peridotites from the Mariana Forearc–Conical Seamount, ODP Leg125: implications for the elemental recycling at subduction zones. GeochemistryGeophysics Geosystems 6 (4), Q04J15. doi:10.1029/2004GC000777.
Savov, I.P., Leeman, W., 2007. Boron Isotopic Variations in the NW USA Rhyolites:Yellowstone, Snake River Plain and Eastern Oregon, GSA Annual Meeting, Denver,CO, USA. Abstract # 131295.
Savov, I.P., Cai, M.Y., Steven, B.Shirey, Ryan, J.G., Hauri, E., Goldstein, S.L., Navarro-Ochoa, C.,2007. Is new Plinian eruption imminent at Colima volcano in Mexico? Insights frommineral chemistry, melt inclusion volatiles and bulk rock B–Li–Sr–Nd–Hf–Pb isotopesystematics. Eos Transaction AGU 88 (52) Fall Meet. Suppl., Abstract V31H-06.
Siebert, L., Simkin, T., 2002. Volcanoes of the world: an illustrated catalog of Holocenevolcanoes and their eruptions. Smithsonian Institution, Global Volcanism ProgramDigital Information Series, GVP-3. http://www.volcano.si.edu/gvp/world/.
Sisson, T.W., Grove, T.L., 1993. Temperatures and H2O contents of low-MgO high-alumina basalts. Contributions Mineralogy Petrology 113, 167–184.
Smithsonian Institution, 1999. Bulletin of the Global Volcanism Network, 24–06, -02,-04, -06, - 08.
Smithsonian Institution, 2000. Bulletin of the Global Volcanism Network, 25–06.Smithsonian Institution, 2001. Bulletin of theGlobal VolcanismNetwork, 26-04, -05, -10.Smithsonian Institution, 2002. Bulletin of the Global Volcanism Network, 27–02, -05,
-06, -11.Smithsonian Institution, 2003. Bulletin of the Global Volcanism Network, 28–06, -07,
-08, -11.Smithsonian Institution, 2004. Bulletin of the Global Volcanism Network, 29–05.Smithsonian Institution, 2005. Bulletin of the Global Volcanism Network, 30–01, -03.Teng, F.Z., McDonough, W.F., Rudnick, R.L., Dalpe, C., Tomascak, P.B., Chappell, B.W., Gao,
S., 2004. Lithium isotopic composition and concentration of the upper continentalcrust. Geochimica et Cosmochimica Acta 68, 4167–4178.
Venzke, E., Wunderman, R.W., McClelland, L., Simkin, T., Luhr, J.F., Siebert, L., Mayberry,G., 2002. Global Volcanism, 1968 to the Present. Smithsonian Institution, GlobalVolcanism Program Digital Information Series, GVP-4. . http://www.volcano.si.edu/reports/.
Wells, P.R.A., 1977. Pyroxene thermometry in simple and complex systems. Contribu-tions to Mineralogy and Petrology 62, 129–139.
Wood, B.J., Banno, S., 1973. Garnet-orthopyroxene and orthopyroxene-clinopyroxenerelationships in simple and complex systems. Contributions to Mineralogy andPetrology 42, 109–124.
Zobin, V.M., Luhr, J.F., Taran, Y.A., Breton, M., Cortes, A., De La Cruz-Reyna, S., Dominguez,T., Galindo, I., Gavilanes, J.C., Muniz, J.J., Navarro, C., Ramirez, J.J., Reyes, G.A., Ursua,M., Velasco, J., Alatorre, E., Santiago, H., 2002. Overview of the 1997–2000 activity ofVolcán de Colima, México. Journal of Volcanology and Geothermal Research 117,1–19.
