03.04.2013 Views

An Investigation into the Nature of the Magmatic Plumbing System at ...

An Investigation into the Nature of the Magmatic Plumbing System at ...

An Investigation into the Nature of the Magmatic Plumbing System at ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 PAGES 2187^2220 2011 doi:10.1093/petrology/egr044<br />

<strong>An</strong> <strong>Investig<strong>at</strong>ion</strong> <strong>into</strong> <strong>the</strong> <strong>N<strong>at</strong>ure</strong> <strong>of</strong> <strong>the</strong> <strong>Magm<strong>at</strong>ic</strong><br />

<strong>Plumbing</strong> <strong>System</strong> <strong>at</strong> ParicutinVolcano, Mexico<br />

MICHAEL C. ROWE 1,2 *, DAVID W. PEATE 1 AND<br />

INGRID UKSTINS PEATE 1<br />

1 DEPARTMENT OF GEOSCIENCE, UNIVERSITY OF IOWA, 121 TROWBRIDGE HALL, IOWA CITY, IA 52242, USA<br />

2 SCHOOL OF EARTH AND ENVIRONMENTAL SCIENCES, WASHINGTON STATE UNIVERSITY, PULLMAN, WA 99164, USA<br />

RECEIVED JUNE 23, 2010; ACCEPTED SEPTEMBER 1, 2011<br />

The temporal evolution <strong>of</strong> erupted magma compositions <strong>at</strong> Paricutin<br />

Volcano (Mexico) is <strong>of</strong>ten cited as a classic example <strong>of</strong> assimil<strong>at</strong>ion^fractional<br />

crystalliz<strong>at</strong>ion processes with significant progressive<br />

changes in major element, trace element, and isotopic compositions<br />

occurring over <strong>the</strong> rel<strong>at</strong>ively short 9 year lifespan <strong>of</strong> <strong>the</strong> volcano. In<br />

this study, major and trace element compositions <strong>of</strong> olivine- and<br />

orthopyroxene-hosted melt inclusions are integr<strong>at</strong>ed with new trace<br />

element analyses <strong>of</strong> <strong>the</strong> erupted lavas and d<strong>at</strong>a for entrained xenoliths<br />

and xenolith glasses to provide a more comprehensive evalu<strong>at</strong>ion <strong>of</strong><br />

<strong>the</strong> evolution <strong>of</strong> Paricutin Volcano th<strong>at</strong> questions this view. Melt inclusion<br />

compositions are bimodal with an undegassed, low-Si popul<strong>at</strong>ion<br />

(Type I) similar in composition to <strong>the</strong> whole-rock samples<br />

and a degassed, high-Si popul<strong>at</strong>ion (Type II) recording l<strong>at</strong>e-stage<br />

degassing and crystalliz<strong>at</strong>ion <strong>of</strong> <strong>the</strong> magma. Despite <strong>the</strong> rapid<br />

changes in lava composition, melt inclusions hosted in both olivine<br />

and orthopyroxene do not record any progressive contamin<strong>at</strong>ion or<br />

mixing <strong>of</strong> magmas. Homogeneity <strong>of</strong> Type I melt inclusions within<br />

single lava samples implies significant contamin<strong>at</strong>ion prior to crystalliz<strong>at</strong>ion<br />

and potentially a decoupling <strong>of</strong> assimil<strong>at</strong>ion^fractional<br />

crystalliz<strong>at</strong>ion processes. Pre-existing models <strong>of</strong> magma evolution <strong>at</strong><br />

Paricutin Volcano are not consistent with <strong>the</strong> melt inclusion results<br />

or new trace element whole-rock d<strong>at</strong>a.Whole-rock and melt inclusion<br />

trace element analyses corrobor<strong>at</strong>e previous studies, which have suggested<br />

th<strong>at</strong> <strong>the</strong> early erupted m<strong>at</strong>erial (Phase 1; February^July<br />

1943) was <strong>of</strong> a compositionally distinct magma compared with <strong>the</strong><br />

bulk <strong>of</strong> <strong>the</strong> erupted m<strong>at</strong>erial during Phase 2 (July 1943^1946).<br />

There is a second compositional transition between <strong>the</strong> Phase 2 and<br />

Phase 3 (1947^1952) lavas, marked by a sudden change in Zr/Nb<br />

despite similar MgO values, th<strong>at</strong> is consistent with <strong>the</strong> arrival <strong>of</strong> a<br />

new magma b<strong>at</strong>ch. This transition occurs prior to <strong>the</strong> major<br />

*Corresponding author.Telephone: (509) 335-6770.<br />

E-mail: mcrowe@wsu.edu<br />

compositional change from basaltic andesite to andesite magmas in<br />

<strong>the</strong> waning stages <strong>of</strong> <strong>the</strong> eruption th<strong>at</strong> is consistent with progressive<br />

crustal assimil<strong>at</strong>ion within this l<strong>at</strong>est magma b<strong>at</strong>ch. These d<strong>at</strong>a<br />

demonstr<strong>at</strong>e th<strong>at</strong> <strong>the</strong> petrogenetic evolution <strong>of</strong> magmas <strong>at</strong> Paricutin<br />

is more complex than simple progressive assimil<strong>at</strong>ion and fractional<br />

crystalliz<strong>at</strong>ion and requires <strong>the</strong> presence <strong>of</strong> three compositionally distinct<br />

magma b<strong>at</strong>ches <strong>at</strong> shallow levels.<br />

KEY WORDS: trace element; magma chamber; melt inclusion; crustal<br />

contamin<strong>at</strong>ion; crystalliz<strong>at</strong>ion<br />

INTRODUCTION<br />

Paricutin volcano in Mexico is a rel<strong>at</strong>ively short-lived<br />

(1943^1952) volcanic center (cinder cone þ associ<strong>at</strong>ed lava<br />

and tephra) th<strong>at</strong> is <strong>of</strong>ten cited as <strong>the</strong> ‘classic’ example <strong>of</strong><br />

an assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion (AFC) process.<br />

The composition <strong>of</strong> <strong>the</strong> Paricutin lavas and tephras<br />

evolved over <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption from basaltic andesite<br />

to andesite ( 55^60 wt % SiO 2: Fig. 1a; e.g. Wilcox,<br />

1954), accompanied by an increase in 87 Sr/ 86 Sr (Fig. 1b)<br />

and d 18 O th<strong>at</strong> have been <strong>at</strong>tributed to progressive crustal<br />

assimil<strong>at</strong>ion <strong>of</strong> a single magma b<strong>at</strong>ch (McBirney et al.,<br />

1987). Published models to explain <strong>the</strong> compositional variability<br />

in Paricutin lavas have required significant<br />

amounts <strong>of</strong> crustal assimil<strong>at</strong>ion accompanying crystalliz<strong>at</strong>ion<br />

<strong>of</strong> olivine and plagioclase (Wilcox, 1954; McBirney<br />

et al., 1987; Cebria¤ et al., 2011).<br />

ß The Author 2011. Published by Oxford University Press. All<br />

rights reserved. For Permissions, please e-mail: journals.permissions@<br />

oup.com<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 1. (a) Eruption d<strong>at</strong>e vs SiO 2 (whole-rock lava and tephra d<strong>at</strong>a); (b) inset plot <strong>of</strong> SiO 2 vs 87 Sr/ 86 Sr. D<strong>at</strong>a sources: Wilcox (1954); McBirney<br />

et al. (1987);Luhr(2001);Cebria¤ et al. (2011).<br />

Assimil<strong>at</strong>ion and crystalliz<strong>at</strong>ion are <strong>of</strong>ten inferred to be<br />

intim<strong>at</strong>ely linked, with <strong>the</strong> l<strong>at</strong>ent he<strong>at</strong> <strong>of</strong> crystalliz<strong>at</strong>ion<br />

providing <strong>the</strong> <strong>the</strong>rmal driving force for continued crustal<br />

assimil<strong>at</strong>ion (e.g. Bowen, 1928; DePaolo, 1981; Davidson &<br />

Wilson, 1989; Kuritani et al., 2005). Most investig<strong>at</strong>ions<br />

have been based largely on whole-rock geochemistry; however,<br />

analysis <strong>of</strong> melt inclusions trapped in <strong>the</strong> crystallizing<br />

mineral phases could potentially provide a direct means<br />

to test models <strong>of</strong> coupled assimil<strong>at</strong>ion^crystalliz<strong>at</strong>ion. A<br />

comparison <strong>of</strong> compositional vari<strong>at</strong>ions in primary melt<br />

inclusions (trapped during crystalliz<strong>at</strong>ion) with <strong>the</strong>ir host<br />

mineral composition should allow <strong>the</strong> compositional evolution<br />

<strong>of</strong> <strong>the</strong> magma during crystalliz<strong>at</strong>ion to be monitored,<br />

ra<strong>the</strong>r than relying on final homogenized bulk-rock<br />

compositions. If assimil<strong>at</strong>ion and crystalliz<strong>at</strong>ion are a<br />

coupled process, temporally and sp<strong>at</strong>ially rel<strong>at</strong>ed, it is expected<br />

th<strong>at</strong> as crystalliz<strong>at</strong>ion proceeds and <strong>the</strong> host mineral<br />

composition evolves <strong>the</strong> melt inclusion compositions<br />

will likewise change and record increased crustal contamin<strong>at</strong>ion,<br />

as demonstr<strong>at</strong>ed by Kent et al. (2002) for Yemen<br />

flood basalts. Numerous studies have evalu<strong>at</strong>ed <strong>the</strong>rmal<br />

2188<br />

and physical models for coupled assimil<strong>at</strong>ion and fractional<br />

crystalliz<strong>at</strong>ion in magma chambers, rel<strong>at</strong>ing <strong>the</strong>se<br />

pre-eruptive physical processes to compositional vari<strong>at</strong>ions<br />

in <strong>the</strong> magma (e.g. McBirney et al., 1985; Spera &<br />

Bohrson, 2001; Kaneko & Koyaguchi, 2004; Leitch, 2004;<br />

Spera & Bohrson, 2004; Kuritani et al., 2005, 2007). O<strong>the</strong>r<br />

studies have suggested th<strong>at</strong> assimil<strong>at</strong>ion can occur as a<br />

rapid, l<strong>at</strong>e-stage process, potentially during <strong>the</strong> course <strong>of</strong><br />

an eruption (e.g. Dungan, 2005; Erlund et al., 2010) or<br />

even th<strong>at</strong> bulk lava compositions can record a different<br />

petrogenetic history from melt inclusions, with lavas<br />

recording deeper crystalliz<strong>at</strong>ion whereas <strong>the</strong> melt inclusions<br />

record shallow degassing-induced crystalliz<strong>at</strong>ion and<br />

assimil<strong>at</strong>ion (e.g. Johnson et al., 2008).<br />

The objective <strong>of</strong> this study is to test models <strong>of</strong> magma<br />

evolution <strong>at</strong> Paricutin Volcano so as to develop a better<br />

understanding <strong>of</strong> <strong>the</strong> progressive development <strong>of</strong> <strong>the</strong> magm<strong>at</strong>ic<br />

plumbing system and <strong>the</strong> rel<strong>at</strong>ive timing <strong>of</strong> crustal<br />

assimil<strong>at</strong>ion and crystalliz<strong>at</strong>ion in evolving magma systems.<br />

We present compositional d<strong>at</strong>a for host minerals and<br />

melt inclusions from a well-characterized suite <strong>of</strong> lavas<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

from throughout <strong>the</strong> eruptive history <strong>of</strong> Paricutin Volcano<br />

(Wilcox, 1954; McBirney et al., 1987). In addition, we integr<strong>at</strong>e<br />

new trace element analyses <strong>of</strong> <strong>the</strong> lavas and entrained<br />

crustal xenoliths and xenolith glasses with <strong>the</strong><br />

melt inclusion and crystal chemistry and liter<strong>at</strong>ure d<strong>at</strong>a to<br />

provide a more comprehensive view <strong>of</strong> <strong>the</strong> evolution <strong>of</strong><br />

<strong>the</strong> Paricutin magm<strong>at</strong>ic system.<br />

ERUPTIVE HISTORY OF<br />

PARICUTIN VOLCANO<br />

Paricutin lies within <strong>the</strong> Michoaca¤n^Guanaju<strong>at</strong>o volcanic<br />

field (MGVF), which contains over 1000 small eruptive<br />

centers (Appendix Fig. A1) over an 40 000 km 2 region<br />

(Hasenaka,1994); <strong>the</strong>re is no evidence for any previous volcanism<br />

<strong>at</strong> <strong>the</strong> site <strong>of</strong> <strong>the</strong> volcano. The 9 year eruption <strong>of</strong><br />

Paricutin began on 20 February 1943, following several<br />

weeks <strong>of</strong> intensifying seismicity, and <strong>the</strong> total erupted<br />

volume <strong>of</strong> basaltic andesite and andesite magma is estim<strong>at</strong>ed<br />

<strong>at</strong> 1·38 km 3 .<br />

Previous studies have divided <strong>the</strong> eruption <strong>into</strong> phases<br />

based on changes in eruptive behavior and magma composition<br />

(McBirney et al., 1987; Luhr, 2001; Pioli et al.,<br />

2008). In <strong>the</strong> present study, we use a division <strong>into</strong> four<br />

eruptive phases (1, 2, 3a, 3b), based predominantly on lava<br />

compositions; <strong>the</strong>se are a minor modific<strong>at</strong>ion <strong>of</strong> <strong>the</strong><br />

McBirney et al. (1987) and Pioli et al. (2008) subdivisions.<br />

Phase 1 includes m<strong>at</strong>erial erupted from early to mid-1943;<br />

this short period <strong>at</strong> <strong>the</strong> initi<strong>at</strong>ion <strong>of</strong> <strong>the</strong> eruption is<br />

marked by a more primitive whole-rock composition and<br />

lower K 2O content in both lavas and tephra erupted prior<br />

to July 1943 (Luhr, 2001; Pioli et al., 2008). Phase 2 extends<br />

from July 1943 to 1946, a period during which <strong>the</strong> major<br />

element and isotopic vari<strong>at</strong>ions <strong>of</strong> <strong>the</strong> lavas were rel<strong>at</strong>ively<br />

restricted ( 55 wt % SiO 2; 87 Sr/ 86 Sr 0·7038; McBirney<br />

et al., 1987). Toge<strong>the</strong>r, Phases 1 and 2 make up most <strong>of</strong> <strong>the</strong><br />

erupted volume <strong>at</strong> Paricutin ( 75 vol. %; McBirney et al.,<br />

1987). Crustal xenoliths were predominantly recovered<br />

during <strong>the</strong> first 3 years <strong>of</strong> <strong>the</strong> eruption (during Phases 1<br />

and 2), and comprise a variety <strong>of</strong> felsic igneous crustal<br />

lithologies, variably altered and partially melted, consisting<br />

<strong>of</strong> feldspar and quartz crystals with varying proportions<br />

(up to 90 vol. %) <strong>of</strong> vesicular glass (McBirney et al.,<br />

1987). Phase 3a (‘middle stage’ <strong>of</strong> McBirney et al., 1987) incorpor<strong>at</strong>es<br />

a rapid compositional shift from basaltic andesite<br />

to andesite during 1947 and early 1948 th<strong>at</strong> accounts<br />

for only 8^10% <strong>of</strong> <strong>the</strong> erupted volume. The final eruptive<br />

phase (Phase 3b) extends from approxim<strong>at</strong>ely August<br />

1948 to <strong>the</strong> end <strong>of</strong> <strong>the</strong> eruption in 1952 and is domin<strong>at</strong>ed<br />

by <strong>the</strong> extrusion <strong>of</strong> compositionally similar andesitic lavas<br />

( 60 wt % SiO 2; 87 Sr/ 86 Sr 0·7042; McBirney et al., 1987).<br />

Crustal m<strong>at</strong>erial, ei<strong>the</strong>r as xenoliths or xenocrysts, is very<br />

rare in <strong>the</strong> Phase 3 eruptive m<strong>at</strong>erial.<br />

2189<br />

SAMPLE DETAILS AND<br />

ANALYTICAL METHODS<br />

Sample selection for melt inclusion study<br />

For melt inclusion studies, tephra samples are <strong>of</strong>ten easier<br />

to process because <strong>of</strong> <strong>the</strong> presence <strong>of</strong> n<strong>at</strong>urally glassy melt<br />

inclusions th<strong>at</strong> can be analyzed without any he<strong>at</strong>ing and<br />

rehomogeniz<strong>at</strong>ion (see below). Luhr (2001) analyzed<br />

glassy melt inclusions in tephra samples from Paricutin,<br />

but he was not able to cover <strong>the</strong> full compositional or temporal<br />

range observed in <strong>the</strong> erupted lavas because <strong>of</strong> <strong>the</strong><br />

restriction to tephra samples from <strong>the</strong> collections <strong>at</strong> <strong>the</strong><br />

Smithsonian N<strong>at</strong>ional Museum <strong>of</strong> N<strong>at</strong>ural History.<br />

Samples for <strong>the</strong> present study were selected from a more<br />

represent<strong>at</strong>ive suite <strong>of</strong> lavas <strong>of</strong> known eruption age th<strong>at</strong><br />

were <strong>the</strong> focus <strong>of</strong> previous studies (from <strong>the</strong> Smithsonian<br />

collections; Wilcox, 1954; McBirney et al., 1987), thus providing<br />

a thorough chronological, geochemical and petrological<br />

context as <strong>the</strong> basis for <strong>the</strong> detailed melt inclusion<br />

study. Selected lava samples were specifically chosen to<br />

cover all <strong>of</strong> <strong>the</strong> eruptive phases as previously defined (see<br />

details inTable 1), although this meant th<strong>at</strong> any melt inclusions<br />

were likely to be crystallized and would have to be<br />

rehomogenized to a glass prior to microbeam analysis.<br />

Olivine is <strong>the</strong> dominant phenocryst phase in all lava<br />

samples erupted prior to 1947. The early 1943 lava<br />

(116293-7) contains 5 vol. % olivine phenocrysts up to<br />

1·5 mm in length, with rare plagioclase microphenocrysts,<br />

and <strong>the</strong> groundmass is domin<strong>at</strong>ed by plagioclase4orthopyroxene.<br />

McBirney et al. (1987) noted th<strong>at</strong><br />

plagioclase is present in lavas from 1943 to 1944, but it appears<br />

as a microphenocryst only after mid-1944, consistent<br />

with our observ<strong>at</strong>ions. Lava samples 116295-23 and<br />

116289-8 contain varying amounts (generally less than<br />

4 vol. %) <strong>of</strong> euhedral olivine phenocrysts up to 1mm in<br />

length, and a groundmass domin<strong>at</strong>ed by plagioclase with<br />

minor orthopyroxene and olivine. Lava samples from<br />

early 1947 (116289-9) to l<strong>at</strong>e 1947 (116289-12) have decreasing<br />

olivine abundances, from 5 vol. % to 1^2 vol. %,<br />

with a groundmass <strong>of</strong> plagioclase with minor olivine.<br />

Orthopyroxene is <strong>the</strong> predominant phenocryst phase<br />

after 1947 ( 2^3 vol. %; up to 0·9 mm), as <strong>the</strong> host<br />

magma changed to an andesitic composition, although<br />

experimental d<strong>at</strong>a demonstr<strong>at</strong>e th<strong>at</strong> olivine is a potential<br />

stable phenocryst phase <strong>at</strong> low pressure (51 kbar;<br />

Eggler, 1972). Euhedral olivine is present as a phenocryst<br />

phase in Phase 3 lavas, but it is rimmed with orthopyroxene<br />

(up to 10 mm thick). Luhr (2001) argued th<strong>at</strong> such rims<br />

<strong>of</strong> orthopyroxene must have formed after eruption while<br />

<strong>the</strong> lavas slowly cooled, as <strong>the</strong>se rims are not observed on<br />

olivine grains in rapidly quenched tephra. A few spinel<br />

crystals were observed in <strong>the</strong> early lavas but are volumetrically<br />

insignificant (McBirney et al., 1987; Bannister et al.,<br />

1998).<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Table 1: Paricutin lava and xenolith samples<br />

Smithsonian Collection M<strong>at</strong>erial Eruption Eruption Melt Rehomogeniz<strong>at</strong>ion<br />

no. (NMNH) no. type d<strong>at</strong>e stagey inclusionsz temp. (8C)<br />

116293-7 51-W-18 Lava Feb. 1943 1 16 1158<br />

108081 108081 Lava Jan. 8, 1944 2 – –<br />

116295-27 W-47-27 Lava Oct. 1944 2 – –<br />

116295-23 W-47-23 Lava Sept. 1945 2 17 1175<br />

116289-8 W-46-27 Lava Sept. 18, 1946 2 12 1169<br />

116289-9 W-47-9 Lava Apr. 9, 1947 3a 10 1172<br />

116289-12 W-47-30 Lava Nov. 30, 1947 3a 14 1117<br />

116289-13 W-48-5 Lava Aug. 1948 3b – –<br />

116289-15 FP-20-49 Lava Dec. 13, 1949 3b 25 1112<br />

116289-16 FP-20-50 Lava Sept. 1, 1950 3b – –<br />

116289-19 FP-16-52 Lava Feb. 25, 1952 3b 32 1113<br />

108126 108126 Xenolith Unknown Unknown – –<br />

116289-20 51-W-1 Xenolith May 1943 1 – –<br />

116293-4 51-W-6 Xenolith 1943 1–2 – –<br />

116293-5 51-W-7 Xenolith 1944 2 – –<br />

116289-23 51-W-8 Xenolith 1944 2 – –<br />

*Sample identific<strong>at</strong>ion and eruption d<strong>at</strong>es from <strong>the</strong> Smithsonian N<strong>at</strong>ional Museum <strong>of</strong> N<strong>at</strong>ural History d<strong>at</strong>abase (NMNH<br />

sample prefix removed).<br />

yEruption stages follow those defined in <strong>the</strong> text.<br />

zNumber <strong>of</strong> melt inclusions analyzed for major elements.<br />

Whole-rock lava and crustal xenolith<br />

analyses<br />

Details <strong>of</strong> <strong>the</strong> studied lava and crustal xenolith samples are<br />

summarized in Table 1. Whole-rock trace element compositions<br />

were analyzed by inductively coupled plasma mass<br />

spectrometry (ICP-MS) <strong>at</strong> Washington St<strong>at</strong>e University,<br />

using <strong>the</strong> methods described by Knaack et al. (1994), on aliquots<br />

<strong>of</strong> powder from 11 lava and five crustal xenoliths on<br />

loan from <strong>the</strong> Smithsonian N<strong>at</strong>ional Museum <strong>of</strong> N<strong>at</strong>ural<br />

History. The new trace element d<strong>at</strong>a are presented in<br />

Table 2, toge<strong>the</strong>r with accompanying whole-rock major<br />

element d<strong>at</strong>a for all <strong>the</strong> samples from McBirney et al.<br />

(1987).<br />

Sample prepar<strong>at</strong>ion for melt<br />

inclusion study<br />

Paricutin lava samples were hand crushed and sieved, and<br />

olivine, pyroxene, and plagioclase grains were handpicked<br />

from <strong>the</strong> 4250 mm fragments under a binocular microscope.<br />

Microscope observ<strong>at</strong>ions showed th<strong>at</strong> <strong>the</strong> melt inclusions<br />

were mostly to completely crystalline and<br />

<strong>the</strong>refore had to be rehomogenized to a glass prior to analysis.<br />

Mineral phases in <strong>the</strong> melt inclusions could not be<br />

identified petrographically prior to rehomogeniz<strong>at</strong>ion.<br />

Groundmass-free mineral grains were rehe<strong>at</strong>ed in a 1<strong>at</strong>m<br />

2190<br />

Deltech vertical tube furnace <strong>at</strong> <strong>the</strong> University <strong>of</strong> Iowa.<br />

Furnace temper<strong>at</strong>ures were estim<strong>at</strong>ed from whole-rock<br />

compositions, based on liquidus calcul<strong>at</strong>ions from both<br />

COMAGMAT (Ariskin et al., 1993) and MELTS<br />

(Ghiorso & Sack, 1995; Asimow & Ghiorso, 1998). Oxygen<br />

fugacity in <strong>the</strong> sealed tube was maintained slightly below<br />

<strong>the</strong> QFM (quartz^fayalite^magnetite) oxygen buffer with<br />

a CO 2^H 2 gas mixture. Total time <strong>of</strong> he<strong>at</strong>ing above<br />

10008Cwaskeptto 15 min, with 10 min <strong>at</strong> run temper<strong>at</strong>ure,<br />

based on <strong>the</strong> methods and r<strong>at</strong>ionale discussed by<br />

Rowe et al. (2006, 2007). Samples were <strong>the</strong>n rapidly<br />

quenched, which resulted in glassy melt inclusions (Fig. 2).<br />

Quenched grains were separ<strong>at</strong>ely mounted and polished<br />

to expose <strong>the</strong> melt inclusions. We examined melt inclusions<br />

in both olivine and orthopyroxene, but we were unable to<br />

recover inclusions for analysis from plagioclase crystals.<br />

Electron microprobe and secondary ion<br />

mass spectrometry analytical methods<br />

Major element compositions <strong>of</strong> melt inclusions (including<br />

S and Cl), host olivine and orthopyroxene grains, and<br />

xenolith glasses, were measured by electron microprobe<br />

analysis (EMPA) <strong>at</strong> Oregon St<strong>at</strong>e University on a<br />

Cameca SX100 instrument. Detailed analytical methods<br />

for analysis <strong>of</strong> melt inclusions and olivine grains have<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/<br />

by guest on April 3, 2013


Table 2: Major and trace element compositions <strong>of</strong> Paricutin lavas and xenoliths<br />

Sample (NMNH): 116293-7 108081 116295-27 116295-23 116289-8 116289-9 116289-12 116289-13<br />

Major elements (McBirney et al., 1987; wt %)<br />

SiO 2 54·59 55·39 55·71 55·79 56·13 57·05 58·39 59·09<br />

TiO 2 0·99 0·94 1·01 0·90 1·02 0·89 0·86 0·78<br />

Al 2O 3 17·83 17·64 17·24 17·48 17·34 17·27 17·78 17·55<br />

Fe 2O 3 2·01 2·16 2·06 1·83 1·74 1·42 1·87 2·04<br />

FeO 5·43 5·46 5·48 5·30 5·42 5·21 4·51 4·27<br />

MnO 0·12 0·13 0·13 0·12 0·12 0·12 0·12 0·11<br />

MgO 5·44 5·43 5·61 5·75 5·58 5·64 4·03 4·03<br />

CaO 7·25 7·18 6·98 6·81 6·99 6·94 6·75 6·46<br />

Na 2O 3·95 3·98 3·99 3·81 3·79 3·71 3·86 3·92<br />

K 2O 0·91 1·15 1·18 1·19 1·30 1·23 1·30 1·50<br />

P 2O 5 0·27 0·35 0·33 0·30 0·36 0·29 0·30 0·08<br />

H 2O þ<br />

ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

0·16 0·09 0·20 0·20 0·20 0·17 0·11 0·03<br />

H2O 0·04 0·05 0·06 0·10 0·06 0·02 0·01 0·30<br />

Total 98·99 99·95 99·98 99·58 100·05 99·96 99·89 100·16<br />

Ni (ppm) 116 103 126 127 122 126 71 73<br />

Cr (ppm) 144·7 176 155·2 170 152·7 162·4 78·5 88<br />

Trace elements (Washington St<strong>at</strong>e University, ICP-MS, ppm)<br />

La 13·26 17·70 18·88 18·01 18·57 16·64 18·00 19·46<br />

Ce 28·58 37·47 39·72 37·57 38·79 34·49 37·30 40·17<br />

Pr 3·86 4·93 5·16 4·89 5·05 4·48 4·89 5·21<br />

Nd 16·61 20·39 21·30 20·09 20·75 18·42 20·08 21·10<br />

Sm 3·82 4·54 4·68 4·33 4·58 4·07 4·32 4·63<br />

Eu 1·30 1·52 1·52 1·42 1·47 1·31 1·37 1·41<br />

Gd 3·71 4·31 4·44 4·12 4·30 3·83 4·00 4·10<br />

Tb 0·59 0·68 0·70 0·64 0·67 0·58 0·62 0·63<br />

Dy 3·45 3·99 4·07 3·82 4·01 3·41 3·61 3·75<br />

Ho 0·70 0·79 0·81 0·76 0·79 0·69 0·71 0·75<br />

Er 1·87 2·13 2·18 2·03 2·09 1·79 1·90 1·98<br />

Tm 0·27 0·30 0·31 0·29 0·30 0·25 0·27 0·28<br />

Yb 1·67 1·87 1·92 1·79 1·84 1·61 1·65 1·74<br />

Lu 0·26 0·30 0·31 0·28 0·29 0·26 0·27 0·29<br />

Ba 311 372 388 400 398 416 472 506<br />

Th 1·01 1·57 1·72 1·72 1·81 1·65 1·64 1·82<br />

Nb 5·25 8·87 9·62 8·30 9·09 7·08 6·90 7·19<br />

Y 17·70 20·12 20·92 19·16 20·14 17·49 18·18 18·83<br />

Hf 2·95 3·76 3·98 3·75 3·88 3·53 3·81 4·08<br />

Ta 0·34 0·59 0·63 0·55 0·60 0·47 0·47 0·49<br />

U 0·37 0·53 0·56 0·56 0·56 0·52 0·54 0·58<br />

Pb 4·95 6·03 5·97 6·15 6·31 6·28 6·98 7·45<br />

Rb 10·4 15·3 16·6 17·0 17·3 19·0 20·6 22·4<br />

Cs 0·35 0·43 0·42 0·44 0·45 0·53 0·60 0·58<br />

Sr 603 588 594 582 576 565 562 547<br />

Sc 18·1 16·6 17·7 17·2 18·0 17·3 14·5 14·2<br />

Zr 109 146 157 146 152 135 145 155<br />

2191<br />

(continued)<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


Table 2: Continued<br />

Sample (NMNH): 116289-15 116289-16 116289-19 108126 116289-20 116293-4 116293-5 116289-23 Av. xeno<br />

Major elements (McBirney et al., 1987; wt %)<br />

SiO 2 59·77 60·24 60·07 71·93 70·88 72·61 71·00 75·95 72·47<br />

TiO 2 0·83 0·80 0·81 0·23 0·36 0·17 0·18 0·04 0·20<br />

Al 2O 3 17·29 17·3 17·28 14·98 14·27 14·98 14·83 13·51 14·51<br />

Fe 2O 3 1·21 1·19 1·37 0·80 1·52 0·55 0·64 0·25 0·75<br />

FeO 4·95 4·59 4·39 1·25 1·53 1·51 1·43 0·27 1·20<br />

MnO 0·11 0·10 0·10 0·06 0·05 0·06 0·06 0·03 0·05<br />

MgO 3·72 3·55 3·73 0·55 1·17 0·32 0·53 0·05 0·52<br />

CaO 6·28 6·14 6·16 2·79 1·65 2·96 3·13 1·05 2·32<br />

Na 2O 3·74 4·01 4·00 4·67 4·18 4·75 4·13 3·90 4·33<br />

K 2O 1·67 1·66 1·67 2·15 3·64 1·63 2·38 4·74 2·91<br />

P 2O 5 0·12 0·04 0·03 0·19 0·11 0·39 0·47 0·13 0·14<br />

H 2O þ<br />

JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

0·00 0·04 0·05 0·03 0·05 0·07 0·11 0·01 0·26<br />

H2O 0·31 0·29 0·28 0·27 0·08 0·16 0·19 0·02 0·05<br />

Total 100 99·95 99·94 99·9 99·49 100·16 99·08 99·95 99·72<br />

Ni (ppm) 57 44 63 11 18 15<br />

Cr (ppm) 67·8 66·5 76 4·7 18 15 37 10 17<br />

Trace elements (Washington St<strong>at</strong>e University, ICP-MS, ppm)<br />

La 20·26 20·32 20·24 14·13 21·94 14·65 13·74 5·87 14·06<br />

Ce 41·48 41·60 41·05 26·11 41·65 26·56 25·19 13·61 26·62<br />

Pr 5·33 5·32 5·19 2·95 4·70 2·97 2·88 1·97 3·10<br />

Nd 21·46 21·37 20·86 10·30 16·46 10·46 9·98 8·97 11·23<br />

Sm 4·54 4·51 4·37 1·85 3·30 1·82 1·78 2·90 2·33<br />

Eu 1·39 1·37 1·31 0·54 0·68 0·53 0·52 0·18 0·49<br />

Gd 4·16 4·09 3·90 1·49 3·08 1·51 1·48 3·26 2·16<br />

Tb 0·64 0·63 0·60 0·23 0·50 0·23 0·23 0·59 0·36<br />

Dy 3·72 3·71 3·56 1·37 3·09 1·37 1·34 3·75 2·18<br />

Ho 0·74 0·73 0·69 0·27 0·62 0·28 0·27 0·77 0·44<br />

Er 1·93 1·94 1·88 0·77 1·75 0·75 0·74 2·11 1·22<br />

Tm 0·28 0·28 0·27 0·12 0·27 0·12 0·11 0·32 0·19<br />

Yb 1·71 1·74 1·64 0·80 1·73 0·78 0·77 2·01 1·22<br />

Lu 0·27 0·28 0·26 0·14 0·28 0·14 0·13 0·31 0·20<br />

Ba 540 552 554 327 498 339 676 124 393<br />

Th 1·96 2·02 2·14 1·80 21·32 1·76 1·71 8·26 6·97<br />

Nb 7·51 7·53 8·00 3·87 5·14 3·98 3·76 3·85 4·12<br />

Y 18·79 18·75 17·84 7·63 17·08 7·70 7·58 23·16 12·63<br />

Hf 4·25 4·22 4·11 3·00 4·75 2·96 2·86 2·73 3·26<br />

Ta 0·51 0·52 0·53 0·34 0·56 0·34 0·32 0·43 0·40<br />

U 0·62 0·63 0·67 0·45 4·93 0·46 0·44 3·25 1·91<br />

Pb 7·95 7·96 8·06 5·85 7·04 7·58 8·82 26·18 11·09<br />

Rb 24·7 26·0 27·2 44·8157·3 34·0 62·2 149·9 89·7<br />

Cs 0·64 0·67 0·72 1·17 2·30 1·14 1·23 2·82 1·73<br />

Sr 531 528 534 395 50 441 458 52 279<br />

Sc 14·5 15·0 14·4 2·6 8·5 2·9 2·8 1·9 3·72<br />

Zr 161 161 158 106 153 107 102 45 102<br />

2192<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Fig. 2. Reflected light images <strong>of</strong> Type I and Type II melt<br />

inclusions. Numbering represents <strong>the</strong> SiO2 content <strong>of</strong> melt inclusions<br />

and <strong>the</strong> adjacent olivine forsterite composition. (Note <strong>the</strong> presence <strong>of</strong><br />

Type I and Type II inclusions in compositionally similar olivine.)<br />

2193<br />

been provided by Rowe et al. (2011). Orthopyroxene grains<br />

were analyzed using <strong>the</strong> same procedure and calibr<strong>at</strong>ion<br />

as olivine. Rhyolitic glasses were analyzed using <strong>the</strong> same<br />

beam conditions as basaltic glass ( 7 mm spot, 30 nA<br />

beam current, 15 keV acceler<strong>at</strong>ing voltage) with a linear<br />

correction applied to recalcul<strong>at</strong>e Na and Si counts to a<br />

zero time intercept. Rhyolitic glass standard (USNM<br />

72854 VG-568) was repe<strong>at</strong>edly analyzed as an<br />

intra-labor<strong>at</strong>ory standard. KE-12 Obsidian Glass (Devine<br />

et al., 1995) and Macusani Glass (Pichavant et al., 1987)<br />

were also analyzed as secondary rhyolite standards. Major<br />

element abundances, accuracy, and precision <strong>of</strong> basaltic<br />

and rhyolitic glass standards are presented inTable 3.<br />

Melt inclusion and xenolith glass trace element abundances<br />

were determined by secondary ion mass spectrometry<br />

(SIMS) <strong>at</strong> Arizona St<strong>at</strong>e University. Melt inclusion<br />

trace element concentr<strong>at</strong>ions ( 88 Sr, 89 Y, 90 Zr, 93 Nb, 138 Ba,<br />

139 La, 140 Ce, 144 Nd, 147 Sm, 151 Eu, 158 Gd, 162 Dy, 174 Yb) were<br />

analyzed on a Cameca 3f ion microprobe, toge<strong>the</strong>r with<br />

30 Si and 42 Ca for calibr<strong>at</strong>ion and to allow correction for<br />

potential olivine overlap. For trace element analysis we<br />

used a 16 O primary beam (0·8^1·2 nA) focused to<br />

20^25 mm in diameter. Positive secondary ions were<br />

acceler<strong>at</strong>ed to 4·5 keV and, following conventional energy<br />

filtering techniques (Shimizu et al., 1978), ions with a<br />

75 20 eV excess kinetic energy were allowed <strong>into</strong> <strong>the</strong><br />

mass spectrometer. Trace elements were analyzed in two<br />

blocks (masses 88 Sr to 139 La and 140 Ce to 174 Yb) with 30 Si<br />

measured before and after each block for normaliz<strong>at</strong>ion.<br />

Count times were 10 s ( 30 Si), 20 s ( 42 Ca, 88 Sr), 30 s (masses<br />

from 89 Yto 144 Nd, plus 158 Gd and 162 Dy) or 40 s ( 147 Sm,<br />

151 Eu, 166 Er, 174 Yb). Measured r<strong>at</strong>ios (M þ / 30 Si) were corrected<br />

for interfering oxides using rare earth element<br />

(REE) oxide production values (MO þ /M þ ) from Zinner<br />

& Crozaz (1986) and a 135 Ba 16 O/ 135 Ba oxide production<br />

r<strong>at</strong>io <strong>of</strong> 0·054 (R. Hervig, personal communic<strong>at</strong>ion). For<br />

small inclusions, CaO concentr<strong>at</strong>ions were calcul<strong>at</strong>ed<br />

from 42 Ca/ 30 Si r<strong>at</strong>ios and compared with CaO concentr<strong>at</strong>ions<br />

determined by EMPA. Where analyses are<br />

determined to have overlapped onto <strong>the</strong> olivine or orthopyroxene<br />

host (lower calcul<strong>at</strong>ed CaO wt % rel<strong>at</strong>ive to<br />

EMPA concentr<strong>at</strong>ions), concentr<strong>at</strong>ions were corrected<br />

assuming an essentially linear dilution <strong>of</strong> <strong>the</strong> melt composition.<br />

Basalt glass BHVO-2 G was used as a calibr<strong>at</strong>ion<br />

standard, and accuracy and precision were based on<br />

repe<strong>at</strong> analysis <strong>of</strong> BCR-2G run as an unknown. Precision<br />

is generally better than 5% for masses lighter than 144 Nd<br />

and 6^9% for masses from 147 Sm to 174 Yb. Accuracy,<br />

rel<strong>at</strong>ive to preferred values for BCR-2G (GEOREM:<br />

http://georem.mpch-mainz.gwdg.de), is better than precision<br />

for all elements except 158 Gd (þ11%) and 174 Yb<br />

( 13%). Additional details for accuracy and precision for<br />

basaltic trace element analysis by SIMS are presented in<br />

Table 4.<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Table 3: Repe<strong>at</strong> analysis <strong>of</strong> secondary electron microprobe glass standards, with calcul<strong>at</strong>ed precision and accuracy, run with<br />

analysis <strong>of</strong> melt inclusions and crustal xenolith glasses<br />

Sample: BHVO-2G BCR-2G Lo-02-04ii KE-12 Obsidianz Macusani Glassz<br />

Av. Prec. Acc. Av. Prec. Acc. Av. Prec. Acc. Av. Prec. Acc. Av. Prec. Acc.<br />

(32) (%)* (%)y (28) (%) (%) (12) (%) (%) (5) (%)* (%)y (5) (%)* (%)y<br />

Major elements (electron microprobe analysis, wt %)<br />

SiO2 49·68 0·7 0·8 54·26 1·1 1·2 48·04 0·7 2·6 70·20 0·3 0·1 71·85 0·4 0·6<br />

TiO2 2·76 1·3 1·1 2·33 1·0 2·1 2·44 1·7 8·2 0·30 14·0 8·6 0·05 71·9 12·6<br />

Al2O3 13·68 0·5 0·6 13·87 0·3 1·2 12·37 0·9 1·6 8·03 0·4 6·0 16·15 0·5 1·9<br />

FeO* 10·85 0·8 4·2 12·57 1·0 1·3 10·85 1·2 0·0 8·78 11·6 4·8 0·55 3·7 9·2<br />

MnO 0·16 10·0 3·2 0·21 11·7 25·5 0·16 18·8 2·1 0·27 8·1 4·8 0·06 23·7 2·0<br />

MgO 7·25 0·5 1·6 3·67 0·7 1·1 9·04 1·6 0·1 0·02 21·7 8·5 0·02 4·6 24·5<br />

CaO 11·52 1·1 1·0 7·37 0·9 3·1 11·02 1·8 4·5 0·36 2·9 3·0 0·22 5·2 6·0<br />

Na2O 2·14 2·2 12·2 2·88 3·0 7·6 2·31 2·9 7·5 7·28 1·5 0·0 4·22 1·3 2·2<br />

K2O 0·50 4·3 1·6 1·77 2·0 1·2 0·53 4·1 11·9 – – – – – –<br />

P2O5 0·28 5·8 3·7 0·37 4·8 8·1 0·28 3·5 8·4 4·14 1·0 3·2 3·64 1·2 0·1<br />

S 0·00 – – 0·00 – – 0·13 5·5 11·9 0·02 – – 0·00 – –<br />

Cl 0·01 – – 0·01 – – 0·14 1·8 0·9 0·34 1·5 2·1 0·05 8·3 15·4<br />

F 0·01 – – 0·02 – – 0·01 – – – – – – – –<br />

Total 98·90 99·34 – – 97·40 – – 99·82 96·84<br />

*Precision (%) calcul<strong>at</strong>ed as standard devi<strong>at</strong>ion/average 100.<br />

yAccuracy (%) reported as <strong>the</strong> devi<strong>at</strong>ion from reported values. BCR-2G and BHVO-2G accepted values from GEOREM<br />

(georem.mpch-mainz.gwdg.de/). LO-02-04ii is a n<strong>at</strong>ural glass (more variable major elements) with reported values from<br />

Kent et al. (1999) and S and Cl from Rowe et al. (2006).<br />

zReported values for accuracy calcul<strong>at</strong>ions are from Pichavant et al. (1987) and Devine et al. (1995).<br />

Trace element abundances in xenolith rhyolitic glasses<br />

were determined on a Cameca 6f ion microprobe <strong>at</strong><br />

Arizona St<strong>at</strong>e University. In addition to elements analyzed<br />

during melt inclusion analysis, 47 Ti and 232 Th were also<br />

counted. We used a 16 O primary beam (5 nA) focused to<br />

20^25 mm in diameter. Positive secondary ions were<br />

acceler<strong>at</strong>ed to 10 keV. Energy filtering similar to th<strong>at</strong><br />

described for melt inclusion analysis was applied to <strong>the</strong><br />

rhyolite procedure. After an initial pre-sputter time <strong>of</strong><br />

180 s, ions with a mass less than 144 Nd were counted for 1s<br />

( 93 Nb, 139 La, 144 Nd counted for 4, 2, and 4 s, respectively),<br />

and ions with a mass gre<strong>at</strong>er than 147 Sm were counted for<br />

5 s for each measurement cycle (25 cycles per analysis).<br />

Measured M þ / 30 Si r<strong>at</strong>ios were corrected for interfering<br />

oxides as described above. NIST 612 glass was used for calibr<strong>at</strong>ion<br />

whereas NIST 610 glass was analyzed as an unknown<br />

before and after <strong>the</strong> analytical session. Precision <strong>of</strong><br />

rhyolitic glass trace element analysis (based on NIST 610<br />

analyses; Table 4) is better than 5% for all elements<br />

except Ba, Ce, Eu, and Th (510% precision), and accuracy<br />

is better than 5% for all elements except Ti (12%),<br />

Sr (7%), Nb (15%), Ba (10%), Ce (10%) and Dy (8%).<br />

2194<br />

RESULTS<br />

Melt inclusion screening and corrections<br />

Care must be taken when interpreting ei<strong>the</strong>r n<strong>at</strong>urally<br />

quenched or rehomogenized melt inclusion compositions.<br />

Post-entrapment modific<strong>at</strong>ion (predominantly host^melt<br />

re-equilibr<strong>at</strong>ion, and w<strong>at</strong>er loss) can dram<strong>at</strong>ically alter<br />

<strong>the</strong> composition <strong>of</strong> <strong>the</strong> inclusions (e.g. Danyushevksy<br />

et al., 2000; Hauri, 2002). Corrections for <strong>the</strong>se ‘n<strong>at</strong>ural’<br />

processes and for <strong>the</strong> effects <strong>of</strong> rehomogeniz<strong>at</strong>ion can<br />

result in significant modific<strong>at</strong>ions <strong>of</strong> <strong>the</strong> measured melt<br />

compositions. In <strong>the</strong> following section we detail assumptions<br />

and corrections applied to melt compositions to<br />

provide <strong>the</strong> clearest estim<strong>at</strong>e <strong>of</strong> <strong>the</strong> trapped melt<br />

compositions; we also provide both measured and corrected<br />

inclusion compositions as Supplementary D<strong>at</strong>a<br />

(available for downloading <strong>at</strong> http://www.petrology<br />

.oxfordjournals.org/). Rehomogeniz<strong>at</strong>ion in a 1<strong>at</strong>m furnace<br />

requires <strong>the</strong> assumption <strong>of</strong> mineral^melt equilibrium<br />

<strong>at</strong> <strong>the</strong> time <strong>of</strong> inclusion trapping. Melt inclusions in both<br />

olivine and orthopyroxene were <strong>the</strong>refore recalcul<strong>at</strong>ed to<br />

be in Fe^Mg equilibrium with <strong>the</strong>ir respective hosts. A<br />

constant Fe^Mg distribution coefficient (K D) <strong>of</strong> 0·30 is<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/<br />

by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Table 4: Repe<strong>at</strong> analysis <strong>of</strong> secondary ion mass spectrometry standards, with calcul<strong>at</strong>ed precision and accuracy, run with<br />

analysis <strong>of</strong> melt inclusions and crustal xenolith glasses<br />

Sample:* BHVO-2G BCR-2G NIST 610<br />

Av. Prec. Acc. Av. Prec. Acc. Av. Prec. Acc.<br />

(15) (%)y (%)z (10) (%) (%) (3) (%) (%)<br />

Trace elements (secondary ion mass spectrometry; ppm)z<br />

Ti – – – – – – 495·1 6·1 12·3<br />

Sr 396·0 5·0 – 347·6 3·3 1·6 539·3 4·7 7·8<br />

Y 26·0 6·3 – 34·9 4·1 0·2 462·1 0·1 2·6<br />

Zr 170·0 3·6 – 184·4 1·8 0·2 426·9 4·1 3·0<br />

Nb 18·3 3·6 – 12·8 3·6 2·3 498·6 2·1 15·9<br />

Ba 131·0 3·9 – 686·7 1·9 0·5 476·2 10·1 10·9<br />

La 15·2 5·1 – 24·4 3·3 1·3 157·1 2·3 0·1<br />

Ce 37·6 4·3 – 51·5 3·4 3·6 500·2 7·4 10·5<br />

Nd 24·5 3·1 – 27·8 4·1 4·0 431·9 4·6 0·3<br />

Sm 6·1 8·5 – 6·4 8·0 3·4 456·2 0·4 1·2<br />

Eu 2·1 10·8 – 1·9 24·7 1·4 467·8 6·6 1·4<br />

Gd 6·2 13·2 – 7·5 8·7 10·8436·1 1·9 3·8<br />

Dy 5·3 11·8 – 6·5 6·4 0·4 449·3 1·7 5·1<br />

Yb 2·0 12·4 – 3·0 8·7 13·0466·3 3·1 1·0<br />

Th – – – – – – 460·2 8·4 0·7<br />

*BHVO-2G was used as a calibr<strong>at</strong>ion standard with BCR-2G run as an unknown during melt inclusion analysis. NIST 612<br />

was run as a secondary standard during analysis <strong>of</strong> <strong>the</strong> crustal xenolith glasses.<br />

yPrecision (%) calcul<strong>at</strong>ed as standard devi<strong>at</strong>ion/average 100.<br />

zAccuracy (%) reported as <strong>the</strong> devi<strong>at</strong>ion from reported values with reported values from GEOREM (georem.mpch-mainz<br />

.gwdg.de/; BHVO-2G and BCR-2G) and Pearce et al. (1997) (NIST 610).<br />

typically assumed for olivine^melt equilibria (Roedder &<br />

Emslie, 1970) and for basaltic melt inclusion corrections<br />

(e.g. Rowe et al., 2009, 2011). Erlund et al. (2010) calcul<strong>at</strong>ed<br />

a K D value 0·34 0·02 based on olivine core compositions<br />

and bulk tephra Mg-number using <strong>the</strong> method described<br />

by Toplis (2005). However, a K D value <strong>of</strong> 0·32 0·01 best<br />

fits <strong>the</strong> n<strong>at</strong>urally quenched glass and olivine compositions<br />

presented by Erlund et al. (2010) and for this reason we<br />

have corrected <strong>the</strong> rehomogenized inclusion compositions<br />

using an olivine^melt KD Fe^Mg <strong>of</strong> 0·32. Orthopyroxene^<br />

melt K D Fe^Mg (0·245) is calcul<strong>at</strong>ed after von Seckendorff<br />

& O’Neill (1993) assuming an average orthopyroxene<br />

Mg-number <strong>of</strong> 78. This Fe^Mg distribution coefficient is<br />

similar to th<strong>at</strong> derived by Roedder & Emslie (1970) <strong>of</strong><br />

0·23. Fe speci<strong>at</strong>ion is based on whole-rock ferric/ferrous<br />

determin<strong>at</strong>ions by McBirney et al. (1987).<br />

Melt^host re-equilibr<strong>at</strong>ion (Fe loss; Danyushevsky et al.,<br />

2000) in olivine-hosted melt inclusions is monitored by<br />

comparing measured melt FeO concentr<strong>at</strong>ions with<br />

observed K D Fe^Mg , as this process should produce a neg<strong>at</strong>ive<br />

correl<strong>at</strong>ion between <strong>the</strong>se two parameters in a suite <strong>of</strong><br />

2195<br />

inclusions (Rowe et al., 2011). Low-Si melt inclusions (see<br />

below) from samples 116289-15 and 116289-19 both display<br />

evidence <strong>of</strong> host re-equilibr<strong>at</strong>ion and were corrected using<br />

<strong>the</strong> s<strong>of</strong>tware provided by Danyushevsky et al. (2000). For<br />

both <strong>of</strong> <strong>the</strong>se samples, <strong>the</strong> final melt FeO* was chosen to<br />

be equal to <strong>the</strong> whole-rock FeO, with Fe^Mg K d values as<br />

described above and Fe 3þ /Fe total from <strong>the</strong> whole-rock analysis<br />

(McBirney et al., 1987). Although <strong>the</strong> basic choice <strong>of</strong><br />

an FeO concentr<strong>at</strong>ion equivalent to <strong>the</strong> whole-rock FeO<br />

may be an oversimplific<strong>at</strong>ion, given <strong>the</strong> potential variability<br />

in melt compositions, excluding Fe and Mg, potential<br />

errors induced by this procedure are rel<strong>at</strong>ively small. To illustr<strong>at</strong>e<br />

this, <strong>the</strong> agreement between measured and corrected<br />

SiO2 and TiO2 concentr<strong>at</strong>ions are shown in Fig. 3.<br />

Regardless <strong>of</strong> <strong>the</strong> correction technique applied, MgO and<br />

FeO have <strong>the</strong> highest potential for error as <strong>the</strong>se components<br />

are heavily leveraged by <strong>the</strong> host mineral compositions.<br />

If olivine-hosted melt inclusions are instead simply<br />

corrected to be in equilibrium with <strong>the</strong>ir host olivine, incomp<strong>at</strong>ible<br />

elements (and SiO 2) vary by less than 5%,<br />

whereas MgO and FeO vary by up to 30%. Only <strong>the</strong><br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 3. Measured vs corrected inclusion compositions for TiO2 and<br />

SiO2.<br />

corrected melt inclusion compositions (Table 5) are discussed<br />

in <strong>the</strong> following sections and figures; <strong>the</strong> measured<br />

melt compositions are provided as Supplementary D<strong>at</strong>a.<br />

Sulfur provides an excellent monitor for melt inclusion<br />

leakage as it is quickly lost by degassing under low or <strong>at</strong>mospheric<br />

pressure. Melt inclusions with sulfur concentr<strong>at</strong>ions<br />

below <strong>the</strong> sulfur detection limit (70 ppm S) have<br />

been removed from <strong>the</strong> present study as <strong>the</strong>se potentially<br />

represent breached inclusions th<strong>at</strong> have completely<br />

degassed during ei<strong>the</strong>r eruption or <strong>the</strong> rehomogeniz<strong>at</strong>ion<br />

process and <strong>the</strong>refore may have suffered secondary alter<strong>at</strong>ion<br />

or significant diffusion along fractures (Nielsen<br />

et al., 1998). Although recent studies have demonstr<strong>at</strong>ed<br />

<strong>the</strong> feasibility <strong>of</strong> diffusion or re-equilibr<strong>at</strong>ion <strong>of</strong> melt<br />

inclusions with <strong>the</strong> host melt through phenocrysts, this process<br />

is more difficult to monitor when inclusion trace<br />

element compositions are not anomalous, and <strong>the</strong>refore<br />

re-equilibr<strong>at</strong>ion may be extremely minor (e.g. Spandler<br />

et al., 2007). We <strong>the</strong>refore consider th<strong>at</strong> trace and<br />

minor element abundances in unbreached melt inclusions<br />

reflect <strong>the</strong> actual compositions from <strong>the</strong> time <strong>of</strong> melt<br />

entrapment.<br />

2196<br />

Corrected major element compositions <strong>of</strong><br />

melt inclusions<br />

Each <strong>of</strong> <strong>the</strong> sampled Paricutin lavas records two compositionally<br />

distinct popul<strong>at</strong>ions <strong>of</strong> olivine-hosted melt inclusions<br />

(Figs 4 and 5; Table 5): one with lower SiO 2<br />

(Type I) and one with higher SiO 2 (Type II), with a typical<br />

average gap <strong>of</strong> 4^8 wt % SiO 2 between <strong>the</strong> two<br />

types. As is demonstr<strong>at</strong>ed below, vari<strong>at</strong>ions in sulfur content<br />

also serve as a distinguishing characteristic <strong>of</strong> <strong>the</strong> two<br />

groups (Fig. 5). The compositional gap is present in both<br />

<strong>the</strong> corrected and uncorrected d<strong>at</strong>asets, indic<strong>at</strong>ing this is<br />

not an artifact <strong>of</strong> <strong>the</strong> correction process. There is no rel<strong>at</strong>ionship<br />

between <strong>the</strong> size <strong>of</strong> <strong>the</strong> melt inclusions and <strong>the</strong>ir<br />

compositional type (Type I, 5^40 mm; Type II, 5^45 mm;<br />

Fig. 5), and <strong>the</strong> same olivine grain can host both Type I<br />

and Type II inclusions, although on average <strong>the</strong> Type II<br />

inclusion host compositions have a lower Mg-number<br />

(Fig. 4; Table 5). The absence <strong>of</strong> a correl<strong>at</strong>ion between inclusion<br />

size and composition and <strong>the</strong> fact th<strong>at</strong> both inclusion<br />

types can be found in <strong>the</strong> same grain reduces <strong>the</strong><br />

likelihood th<strong>at</strong> compositional differences are an artifact <strong>of</strong><br />

re-equilibr<strong>at</strong>ion and diffusion (e.g. Spandler et al., 2007).<br />

Similarly, olivine^melt re-equilibr<strong>at</strong>ion cannot produce<br />

<strong>the</strong> observed vari<strong>at</strong>ions in <strong>the</strong> melt compositions, in particular<br />

<strong>the</strong> bimodal popul<strong>at</strong>ions, based on <strong>the</strong> absence <strong>of</strong> a<br />

compositional correl<strong>at</strong>ion to inclusion size (Danyushevsky<br />

et al., 2002) and <strong>the</strong> requirement <strong>of</strong> a multi-phase crystallizing<br />

assemblage (see below). A similar bimodal popul<strong>at</strong>ion<br />

was documented from a small volume basaltic eruption <strong>at</strong><br />

Dotsero Volcano, Colorado, and was interpreted to record<br />

l<strong>at</strong>e-stage crustal assimil<strong>at</strong>ion and episodic crystalliz<strong>at</strong>ion<br />

(Rowe et al., 2011).<br />

Average Type I melt inclusions have major element compositions<br />

th<strong>at</strong> are equivalent to or slightly elev<strong>at</strong>ed rel<strong>at</strong>ive<br />

to whole-rock compositions, with <strong>the</strong> noted exception <strong>of</strong><br />

SiO 2 and Al 2O 3. Type I inclusions have SiO 2 concentr<strong>at</strong>ions<br />

equivalent to whole-rock SiO 2 concentr<strong>at</strong>ions in<br />

Phase 1 lavas, but SiO 2 is lower rel<strong>at</strong>ive to whole-rock compositions<br />

by up to 4 wt % in eruptive Phases 2 and 3<br />

(Fig. 4). Al 2O 3 and CaO concentr<strong>at</strong>ions are proportionally<br />

enriched in all inclusions rel<strong>at</strong>ive to whole-rock abundances,<br />

from 0·3 to2·1 wt % and from 0·3 to0·7 wt%,respectively<br />

(Fig. 6). Generally, TiO 2 and P 2O 5 contents in<br />

<strong>the</strong> melt inclusions decrease over <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption,<br />

similar to <strong>the</strong> whole-rock trends (Fig. 6). Al 2O 3 concentr<strong>at</strong>ions<br />

do not vary system<strong>at</strong>ically and remain<br />

rel<strong>at</strong>ively constant throughout Phase 3. K 2O and SiO 2 are<br />

<strong>the</strong> only major elements to increase in <strong>the</strong>Type I inclusions<br />

over <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption, with average K 2O and<br />

SiO2 in inclusions varying in <strong>the</strong> range 1·2^2·0 wt % and<br />

54·2^57·6 wt %, respectively, from <strong>the</strong> beginning to<br />

<strong>the</strong> end <strong>of</strong> <strong>the</strong> eruption. SiO 2 concentr<strong>at</strong>ions in Type I inclusions<br />

are lowest in Phase 2 lavas with average SiO 2 as<br />

low as 52·5 wt %. From Phase 2 to Phase 3, SiO 2 increases<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

by 4 wt %, similar to <strong>the</strong> increase in <strong>the</strong> whole-rock<br />

compositions over <strong>the</strong> same time period but <strong>of</strong>fset to system<strong>at</strong>ically<br />

lower concentr<strong>at</strong>ions (Fig. 4; Tables 2 and 5:<br />

McBirney et al., 1987).<br />

Type II melt inclusion compositions are generally more<br />

variable than Type I, both overall and within a given<br />

sample. Average Type II melt inclusion compositions are<br />

enriched in SiO 2,TiO 2,K 2O, and P 2O 5, rel<strong>at</strong>ive to both<br />

Table 5: Average corrected melt inclusion major and trace element compositions<br />

<strong>the</strong> whole-rock and Type I compositions. Phase 1 Type II<br />

compositions are significantly enriched rel<strong>at</strong>ive to Phase 2<br />

(e.g. 2·2 wt % average TiO 2 vs 1·2 wt % average TiO 2 in<br />

Phase 2). The average concentr<strong>at</strong>ions <strong>of</strong> both CaO and<br />

Al 2O 3 increase through Phase 1 and 2 and <strong>the</strong>n decrease<br />

significantly in Phase 3 lavas (Table 5). However, <strong>at</strong><br />

all times average CaO and Al 2O 3 contents in Type II<br />

inclusions are lower than those <strong>of</strong> <strong>the</strong> whole-rock and<br />

Sample: 116293-7 116293-7 116295-23 116295-23 116289-8<br />

Comment: Type I Type II Type I Type II Type I<br />

Host: Olivine Olivine Olivine Olivine Olivine<br />

Av. SD Av. SD Av. SD Av. SD Av.<br />

Corrected inclusion composition (wt %)<br />

SiO2 54·22 1·09 61·93 0·88 52·71 2·95 56·52 1·24 52·51<br />

TiO2 1·24 0·11 2·17 0·49 1·06 0·38 1·18 0·29 1·11<br />

Al2O3 20·29 0·73 15·47 1·61 17·79 1·66 16·82 2·20 18·41<br />

FeO 4·60 0·69 4·39 0·97 6·91 1·30 5·97 1·34 6·65<br />

Fe2O3 1·97 0·23 1·89 0·34 2·41 0·48 2·14 0·38 2·08<br />

MnO 0·11 0·02 0·12 0·04 0·14 0·04 0·14 0·02 0·14<br />

MgO 3·48 0·53 2·77 0·70 6·14 1·35 4·87 0·92 5·68<br />

CaO 7·67 0·77 5·03 0·58 7·18 0·72 6·53 0·72 7·32<br />

Na2O 4·61 0·32 3·24 0·25 3·79 0·46 3·77 0·46 4·05<br />

K2O 1·20 0·10 2·04 0·24 1·20 0·35 1·47 0·26 1·36<br />

P2O5 0·39 0·04 0·79 0·41 0·34 0·17 0·40 0·11 0·37<br />

S 0·046 0·008 0·013 0·00 0·069 0·021 0·020 0·01 0·070<br />

Cl 0·116 0·015 0·099 0·03 0·084 0·021 0·055 0·03 0·093<br />

F 0·02 0·00 0·03 0·02 0·02 0·01 0·02 0·01 0·02<br />

Total 100·00 0·00 100·00 0·00 100·00 0·00 100·00 0·00 100·00<br />

Host Mg-no. 80·80 2·26 77·53 3·39 83·20 0·68 81·73 1·56 81·80<br />

X (host) wt % 4·46 1·46 3·80 2·26 0·26 2·82 0·96 2·47 0·06<br />

Dilution 1·04 1·04 0·99 1·03 1·01<br />

Trace elements (ppm)<br />

Sr 668 51 440 153 580 0·4 595 171 507<br />

Y 18·5 1·2 24·9 12·3 18·9 0·4 17·5 8·7 24·1<br />

Zr 121 7 186 89 146 1·1 135 76 165<br />

Nb 6·3 0·5 9·8 3·6 8·9 0·6 8·6 4·4 11·2<br />

Ba 351 41 465 30 395 5 405 91 555<br />

La 13·3 0·9 17·3 7·6 16·7 1·1 17·0 7·4 21·5<br />

Ce 30·1 2·7 37·0 13·3 37·0 1·2 35·1 15·3 47·2<br />

Nd 17·7 1·9 23·7 10·6 20·8 1·6 19·4 9·1 25·8<br />

Sm 4·0 0·6 5·3 2·0 4·8 0·1 4·5 1·7 5·7<br />

Eu 0·9 0·2 1·1 0·2 1·1 0·2 1·3 0·6 1·7<br />

Gd 4·0 0·3 4·0 1·0 3·7 0·6 4·0 1·5 5·8<br />

Dy 3·3 1·2 3·9 0·2 3·4 0·5 3·6 1·9 4·2<br />

Yb 1·9 0·3 2·0 0·6 2·2 0·3 1·9 0·8 2·2<br />

2197<br />

(continued)<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


Table 5: Continued<br />

JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Sample: 116289-8 116289-9 116289-9 116289-12 116289-12 116289-15<br />

Comment: Type II Type I Type II Type I Type II Type I<br />

Host: Olivine Olivine Olivine Olivine Olivine Olivine*<br />

SD Av. SD Av. SD Av. SD Av. SD Av. SD<br />

Corrected inclusion composition (wt %)<br />

SiO2 1·92 53·06 0·43 58·69 0·70 54·33 1·26 62·23 1·35 56·74 0·69<br />

TiO2 0·19 0·97 0·04 1·41 0·15 1·02 0·19 1·46 0·10 0·93 0·07<br />

Al2O3 0·98 18·60 0·45 15·33 0·72 19·16 2·38 15·45 0·62 19·30 0·46<br />

FeO 0·65 6·64 0·62 6·43 0·63 5·60 1·49 4·38 0·26 4·94 0·01<br />

Fe2O3 0·26 1·79 0·16 1·78 0·13 2·29 0·33 1·78 0·10 1·21 0·00<br />

MnO 0·04 0·12 0·04 0·12 0·03 0·12 0·04 0·10 0·02 0·14 0·04<br />

MgO 0·52 6·03 0·47 4·95 0·27 4·06 1·00 3·09 0·19 3·26 0·10<br />

CaO 0·44 7·42 0·30 5·42 0·34 7·14 0·80 4·62 0·50 6·98 0·19<br />

Na2O 0·27 3·65 0·38 3·70 0·25 4·23 0·45 3·74 0·32 4·26 0·36<br />

K2O 0·13 1·25 0·12 1·62 0·09 1·52 0·17 2·47 0·15 1·87 0·23<br />

P2O5 0·06 0·31 0·02 0·42 0·06 0·36 0·05 0·52 0·06 0·38 0·09<br />

S 0·01 0·056 0·028 0·026 0·01 0·055 0·014 0·021 0·01 0·043 0·010<br />

Cl 0·02 0·064 0·036 0·069 0·03 0·101 0·014 0·122 0·02 0·088 0·009<br />

F 0·01 0·02 0·00 0·02 0·00 0·02 0·00 0·03 0·00 0·02 0·01<br />

Total 0·00 100·00 0·00 100·00 0·00 100·00 0·00 100·00 0·00 100·00 0·00<br />

Host Mg-no. 0·99 83·47 0·97 81·15 0·66 80·23 0·39 79·66 0·51 78·63 0·55<br />

X (host) wt % 1·41 1·23 1·79 0·15 1·04 0·03 3·38 0·25 0·78 – –<br />

Dilution – 0·98 – 0·98 – 0·99 – 0·99 – 1·03 –<br />

Trace elements (ppm)<br />

Sr 36 585 44 389 37 593 30 287 26 542 35<br />

Y 2·7 20·1 4·2 28·6 3·6 20·9 3·5 29·0 2·9 19·0 0·7<br />

Zr 33 154 26 214 36 168 33 253 16 167 7·0<br />

Nb 2·0 8·7 1·6 14·1 2·2 8·9 1·8 13·4 0·8 8·5 0·5<br />

Ba 35 454 63 752 52 515 75 812 83 565 11<br />

La 2·9 19·1 3·7 28·0 4·0 19·8 3·8 26·9 1·7 19·1 0·9<br />

Ce 4·1 40·0 8·6 58·4 7·6 41·0 7·7 60·2 6·9 41·0 1·2<br />

Nd 2·5 21·2 4·4 32·4 3·1 22·5 4·8 32·1 2·2 22·3 1·6<br />

Sm 0·8 4·8 1·4 7·0 1·0 5·5 0·9 6·4 0·7 4·9 0·5<br />

Eu 0·2 1·2 0·5 1·9 0·4 1·4 0·3 2·0 0·3 0·9 0·4<br />

Gd 0·9 4·1 1·3 7·3 0·8 4·3 1·7 7·9 0·6 4·1 0·6<br />

Dy 0·9 3·9 1·0 5·1 1·3 3·7 0·7 5·6 0·3 3·3 0·5<br />

Yb 0·3 2·3 0·4 2·5 0·1 2·2 0·3 2·5 0·2 1·9 0·2<br />

Type I inclusions. Type II melt inclusions are compositionally<br />

similar to <strong>the</strong> m<strong>at</strong>rix glass <strong>of</strong> <strong>the</strong> contemporaneous<br />

tephra deposits (Luhr, 2001; Erlund et al., 2010; Fig. 6).<br />

Although Luhr (2001) expressed doubts about <strong>the</strong> ability<br />

<strong>of</strong> orthopyroxene to preserve undegassed melt inclusions,<br />

given <strong>the</strong> strong cleavage <strong>of</strong> <strong>the</strong> mineral, we have successfully<br />

collected major and vol<strong>at</strong>ile element d<strong>at</strong>a from this<br />

2198<br />

(continued)<br />

mineral phase, which is present only in <strong>the</strong> Phase 3 lavas.<br />

Orthopyroxene-hosted melt inclusions from Phase 3 lavas<br />

tend to be smaller (515 mm) than <strong>the</strong> olivine-hosted inclusions<br />

(Fig. 5), which limited our ability to obtain trace<br />

element analyses on <strong>the</strong>m. They are compositionally similar<br />

to <strong>the</strong> olivine-hosted inclusions and can be similarly<br />

subdivided <strong>into</strong> two popul<strong>at</strong>ions based on <strong>the</strong>ir SiO 2<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


Table 5: Continued<br />

ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Sample: 116289-15 116289-15 116289-15 116289-19 116289-19 116289-19 116289-19<br />

Comment: Type II Type I Type II Type I Type II Type I Type II<br />

Host: Olivine Opx Opx Olivine* Olivine Opx Opx<br />

Av. SD Av. SD Av. SD Av. SD Av. SD Av. SD Av. SD<br />

Corrected inclusion composition (wt %)<br />

SiO2 63·48 1·19 58·88 0·68 63·28 – 57·24 0·69 63·09 2·38 57·62 2·60 64·47 –<br />

TiO2 1·92 0·15 0·84 0·03 1·48 – 0·99 0·14 1·33 0·26 0·85 0·12 1·34 –<br />

Al2O3 15·27 0·96 17·37 0·39 13·62 – 19·39 0·68 16·07 1·62 16·89 1·08 14·38 –<br />

FeO 5·17 0·07 5·70 0·64 5·92 – 4·39 0·02 4·26 0·54 6·77 1·47 4·66 –<br />

Fe2O3 1·17 0·01 1·46 0·14 1·33 – 1·37 0·01 1·24 0·12 2·20 0·40 1·56 –<br />

MnO 0·07 0·02 0·12 0·03 0·08 – 0·14 0·04 0·06 0·01 0·13 0·03 0·13 –<br />

MgO 3·33 0·00 3·29 0·34 3·59 – 3·18 0·06 2·97 0·41 3·62 1·05 2·25 –<br />

CaO 3·85 0·09 6·68 0·25 4·45 – 6·83 0·26 4·28 0·96 6·26 0·59 4·43 –<br />

Na2O 2·47 0·39 3·58 0·24 2·83 – 4·18 0·23 3·20 0·82 3·54 0·33 3·31 –<br />

K2O 2·54 0·02 1·60 0·14 2·58 – 1·96 0·24 2·89 0·39 1·70 0·40 2·86 –<br />

P2O5 0·58 0·00 0·31 0·03 0·63 – 0·35 0·05 0·45 0·10 0·28 0·03 0·52 –<br />

S 0·019 0·00 0·046 0·011 0·011 – 0·039 0·010 0·019 0·01 0·048 0·021 0·007 –<br />

Cl 0·120 0·03 0·086 0·015 0·170 – 0·092 0·011 0·110 0·02 0·078 0·013 0·078 –<br />

F 0·03 0·00 0·01 0·00 0·03 – 0·02 0·00 0·02 0·01 0·01 0·00 0·03 –<br />

Total 100·00 0·00 100·00 0·00 100·00 – 100·00 0·00 100·00 0·00 100·00 0·00 100·00 –<br />

Host Mg-no. 78·17 0·26 80·77 1·70 81·57 – 80·09 0·27 79·41 0·84 79·11 3·27 77·91 –<br />

X (host) wt % 1·70 0·14 2·24 1·70 3·40 – 1·52 1·01 1·85 3·52 2·40 –<br />

Dilution 1·00 1·04 0·99 1·04 1·01 1·02 1·03 –<br />

Trace elements (ppm)<br />

Sr – – – – – – 523 32 396 137 – – – –<br />

Y – – – – – – 17·9 0·1 23·6 4·9 – – – –<br />

Zr – – – – – – 163 6 209 73 – – – –<br />

Nb – – – – – – 9·0 0·3 13·5 5·8 – – – –<br />

Ba – – – – – – 589 26 1093 164 – – – –<br />

La – – – – – – 19·2 1·3 28·6 5·3 – – – –<br />

Ce – – – – – – 41·2 2·8 58·7 10·7 – – – –<br />

Nd – – – – – – 21·6 1·4 28·2 6·2 – – – –<br />

Sm – – – – – – 4·2 0·5 6·1 1·0 – – – –<br />

Eu – – – – – – 1·0 0·1 1·5 0·6 – – – –<br />

Gd – – – – – – 3·5 0·5 5·7 3·0 – – – –<br />

Dy – – – – – – 3·3 0·7 4·9 1·5 – – – –<br />

Yb – – – – – – 1·7 0·2 2·0 0·4 – – – –<br />

*Compositions corrected using <strong>the</strong> spreadsheet and procedure described by Danyushevksy et al. (2000).<br />

contents, although <strong>the</strong> low-SiO 2 Type I popul<strong>at</strong>ion is dominant<br />

(Fig. 4, Table 5). The only significant difference between<br />

orthopyroxene- and olivine-hosted melt inclusions<br />

is th<strong>at</strong> Type I opx-hosted inclusions have SiO 2 concentr<strong>at</strong>ions<br />

more similar to those <strong>of</strong> <strong>the</strong> whole-rock and up to<br />

2 wt % gre<strong>at</strong>er than those <strong>of</strong> <strong>the</strong> Type I olivine-hosted<br />

inclusions. MgO concentr<strong>at</strong>ions are also comparable in<br />

2199<br />

Type I opx-hosted and olivine-hosted inclusions (Table 5).<br />

Higher FeO concentr<strong>at</strong>ions inType I opx-hosted inclusions<br />

(up to 1·5 wt %) rel<strong>at</strong>ive to olivine-hosted inclusions may<br />

be a function <strong>of</strong> <strong>the</strong> Fe-loss corrections ra<strong>the</strong>r than a<br />

geologically significant difference. Type I opx-hosted inclusions<br />

more closely approxim<strong>at</strong>e <strong>the</strong> whole-rock compositions<br />

for Phase 3 lavas than <strong>the</strong> Type I olivine-hosted<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 4. Major and trace element compositions, and host Mg-number <strong>of</strong> Type I (open symbols) and Type II (filled symbols) melt inclusions vs<br />

eruption d<strong>at</strong>e. Whole-rock major element d<strong>at</strong>a are from McBirney et al. (1987) (black symbols and continuous line) and Luhr (2001). Dashed<br />

and dotted lines, respectively, represent <strong>the</strong> average compositions for each group <strong>of</strong> olivine-hosted and orthopyroxene-hosted melt inclusions.<br />

Luhr (2001) melt inclusion compositions are plotted for comparison (open circles).<br />

2200<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Fig. 5. (a) SiO 2 vs S for melt inclusions, divided <strong>into</strong> <strong>the</strong> three main<br />

phases and inclusion Types I and II. (b) Histogram <strong>of</strong> olivine- and<br />

orthopyroxene-hosted melt inclusion sizes for Type I (low-Si) and<br />

Type II (high-Si) inclusion popul<strong>at</strong>ions.<br />

inclusions. Type II opx-hosted inclusions are compositionally<br />

similar to Type II olivine-hosted inclusions in a given<br />

sample.<br />

Trace element compositions <strong>of</strong> lavas and<br />

melt inclusions<br />

New trace element d<strong>at</strong>a for 11 whole-rock lava samples are<br />

reported in Table 2. The trace element characteristics <strong>of</strong><br />

<strong>the</strong> Paricutin samples are summarized on a primitivemantle-normalized<br />

diagram (Fig. 7). They show<br />

2201<br />

Fig. 6. TiO2, P2O5 and Al2O3 vs SiO2 in melt inclusions and<br />

whole-rock samples demonstr<strong>at</strong>ing <strong>the</strong> overall trend <strong>of</strong> Type I melt inclusion<br />

and whole-rock compositions toward xenolith (filled circles)<br />

and xenolith glass (open circles) compositions. Type II melt inclusions<br />

follow fractional crystalliz<strong>at</strong>ion trends (f.c.) similar to <strong>the</strong> Luhr<br />

(2001) and Erlund et al. (2010) tephra glass compositions.<br />

enrichments in <strong>the</strong> large ion lithophile elements (LILE;<br />

Cs,Rb,Ba,U,K)andTh,withneg<strong>at</strong>iveanomalies<strong>at</strong><br />

Nb, Ta, P, and Ti; <strong>the</strong>se fe<strong>at</strong>ures are typical for magmas<br />

derived from subduction-modified mantle (e.g. Pearce &<br />

Pe<strong>at</strong>e, 1995). The REE p<strong>at</strong>terns show strong light REE<br />

(LREE) enrichments (La/Sm N ¼ 2·63 0·23), elev<strong>at</strong>ed<br />

middle REE (MREE) to heavy REE (HREE) levels<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 7. Primitive mantle normalized whole-rock trace element p<strong>at</strong>terns <strong>of</strong> lavas (black lines) and crustal xenoliths (gray lines). Primitive mantle<br />

composition from McDonough & Sun (1995). D<strong>at</strong>a from this study.<br />

(Dy/Yb N ¼1·43 0·03) and no neg<strong>at</strong>ive Eu anomalies<br />

(Eu/Eu* ¼ 1·01 0·03). The Phase 2 samples are distinctive<br />

in having lower Zr/Nb (16^18) compared with those from<br />

Phase 1 and Phase 3, a fe<strong>at</strong>ure th<strong>at</strong> is also apparent in<br />

o<strong>the</strong>r published d<strong>at</strong>asets (Luhr, 2001; Cebria¤ et al., 2011).<br />

Large ion lithophile elements (Rb, Cs, Ba, Pb, K) show a<br />

progressive increase and Sr shows a progressive decrease<br />

with time from Phase 1 to Phase 2 to Phase 3 samples<br />

(Fig. 4). Zr and <strong>the</strong> LREE show similar abundances in<br />

<strong>the</strong> Phase 2 and Phase 3 samples th<strong>at</strong> are higher than in<br />

<strong>the</strong> Phase 1 samples, whereas Y and <strong>the</strong> HREE show<br />

similar abundances in <strong>the</strong> Phase 1 and Phase 3 samples<br />

th<strong>at</strong> are lower than in <strong>the</strong> Phase 2 samples. Nb contents<br />

increase from <strong>the</strong> Phase 1 to Phase 2 whole-rock<br />

samples and decrease to Phase 3. In general, most incomp<strong>at</strong>ible<br />

trace elements show positive linear trends against<br />

SiO 2, with <strong>the</strong> Phase 2 samples displaced to higher<br />

contents.<br />

For <strong>the</strong> melt inclusions, only <strong>the</strong> olivine-hosted inclusions<br />

were large enough (gre<strong>at</strong>er than 20 mm) for trace<br />

element analysis. Type I melt inclusions generally have<br />

trace element abundances similar to whole-rock compositions<br />

(Fig. 4; Tables 2 and 5). Sr concentr<strong>at</strong>ions in <strong>the</strong><br />

melt inclusions exhibit <strong>the</strong> gre<strong>at</strong>est devi<strong>at</strong>ion from <strong>the</strong><br />

2202<br />

whole-rock compositions, particularly <strong>the</strong> Phase 1 and<br />

Phase 3b samples, with concentr<strong>at</strong>ions decreasing over<br />

<strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption (Fig. 4). Trace element abundances<br />

in Type II melt inclusions have a significantly<br />

gre<strong>at</strong>er devi<strong>at</strong>ion from <strong>the</strong> whole-rock compositions than<br />

<strong>the</strong> Type I inclusions, particularly in Phase 3 lavas, with<br />

concentr<strong>at</strong>ions up to double <strong>the</strong> whole-rock values (e.g.<br />

Ba, Fig. 4). Only Sr concentr<strong>at</strong>ions in Type II melt inclusions<br />

are substantially below <strong>the</strong> whole-rock and Type I inclusion<br />

values, with abundances 50% lower than in <strong>the</strong><br />

whole-rock lavas.<br />

Vol<strong>at</strong>ile compositions <strong>of</strong> melt inclusions<br />

Sulfur concentr<strong>at</strong>ions in undegassed Type I melt inclusions<br />

are variable with average abundances from 400 to<br />

700 ppm and <strong>the</strong> highest concentr<strong>at</strong>ions in Phase 2 and<br />

Phase 3a samples (Fig. 5; Table 5). Phase 1, Type I melt inclusions<br />

have sulfur concentr<strong>at</strong>ions less than 600 ppm (in<br />

Fo 83^76 olivine), significantly lower than observed in Phase<br />

1 tephras by Luhr (2001; 1000^1200 ppm S in Fo84 olivine)<br />

and Johnson et al. (2009; 1320^1750 ppm S in Fo 85^87 olivine).<br />

Type II melt inclusions have consistently lower sulfur<br />

abundances (average concentr<strong>at</strong>ions from 100 to 300 ppm<br />

S), rel<strong>at</strong>ive to Type I inclusions, although concentr<strong>at</strong>ions<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

are gre<strong>at</strong>er than <strong>the</strong> m<strong>at</strong>rix glass S concentr<strong>at</strong>ions in <strong>the</strong><br />

tephra samples ( 40 ppm; Luhr, 2001).<br />

Average chlorine abundances vary from 500 to 1200 ppm<br />

Cl (Table 5) with no system<strong>at</strong>ic differences betweenType I<br />

and Type II inclusions. With <strong>the</strong> exception <strong>of</strong> <strong>the</strong> Phase 1<br />

samples th<strong>at</strong> have anomalously high Cl (average<br />

1150 ppm), average Cl concentr<strong>at</strong>ions in Type I inclusions<br />

remain rel<strong>at</strong>ively constant throughout <strong>the</strong> eruption and<br />

display no system<strong>at</strong>ic vari<strong>at</strong>ions (averages range from 630<br />

to 990 ppm; Fig. 8a). In contrast, average Cl/K r<strong>at</strong>ios<br />

Fig. 8. (a) Cl/K vs chlorine concentr<strong>at</strong>ions for Type I melt inclusions. (See Table 1 for clustering <strong>of</strong> samples <strong>into</strong> eruptive phases.) Bulk (BC),<br />

Lower (LC), Upper (UC) and Middle (MC) crust compositions from Rudnick & Gao (2004). Decreasing Cl and Cl/K represents a degassing<br />

p<strong>at</strong>h, whereas constant Cl and decreasing Cl/K (increasing K) represents crustal contamin<strong>at</strong>ion. (b) Cl/K vs K2O/TiO2, for Type I<br />

olivine-hosted melt inclusions. (Note <strong>the</strong> anomalous Cl/K r<strong>at</strong>ios <strong>of</strong> melt inclusions in <strong>the</strong> Phase 1 lava.)<br />

2203<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

decrease over <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption from 0·12 to 0·06<br />

in Type I inclusions (Fig. 8). The neg<strong>at</strong>ive correl<strong>at</strong>ion between<br />

K 2O/TiO 2 and Cl/K despite rel<strong>at</strong>ively constant Cl<br />

(excluding inclusions which record degassing as illustr<strong>at</strong>ed<br />

in Fig. 8) indic<strong>at</strong>es th<strong>at</strong> Cl/K vari<strong>at</strong>ions in undegassed<br />

melt inclusions are being driven largely by increasing<br />

K 2O concentr<strong>at</strong>ions as a result <strong>of</strong> <strong>the</strong> high K 2O and<br />

low Cl <strong>of</strong> likely crustal contaminants (Fig. 8). Chlorine<br />

degassing as indic<strong>at</strong>ed in Fig. 8 is also supported by<br />

decreasing sulfur contents in <strong>the</strong>se melt inclusions (see<br />

Supplementary D<strong>at</strong>a). Cl/K is also neg<strong>at</strong>ively correl<strong>at</strong>ed<br />

with o<strong>the</strong>r indices <strong>of</strong> crustal contamin<strong>at</strong>ion, such as Ba/<br />

Nb and SiO 2 wt % in melt inclusions. Type II inclusions<br />

have average Cl/K ranging from 0·08 to 0·03, but show significantly<br />

more internal vari<strong>at</strong>ion.<br />

Host mineral compositions<br />

Olivine<br />

Olivine host compositions were measured adjacent to each<br />

melt inclusion, and <strong>the</strong>ir vari<strong>at</strong>ions with eruption d<strong>at</strong>e are<br />

shown in Fig. 4. Average olivine compositions hosting<br />

Type I inclusions range from Fo 80·0 to Fo 83·5; Luhr(2001)<br />

reported a similar range in olivine compositions adjacent<br />

to melt inclusions in tephra samples from 1943 to 1948<br />

(Fo 80·0 to Fo 84·4). Average olivine compositions hosting<br />

Type II inclusions extend to more evolved compositions,<br />

ranging from Fo 82·0 to Fo 77·5. Olivine compositions hosting<br />

Type I and Type II inclusion popul<strong>at</strong>ions have similar temporal<br />

vari<strong>at</strong>ions. Average Phase 1 olivines (Type I ¼ Fo 80·8,<br />

Type II ¼Fo77·5) are more evolved than l<strong>at</strong>er Phase 2 olivines<br />

(Type I ¼ Fo 82·9,TypeII¼Fo 81·6). This result is in contrast<br />

to <strong>the</strong> olivine core compositions in basal tephra<br />

deposits reported by Erlund et al. (2010), which indic<strong>at</strong>e<br />

th<strong>at</strong> <strong>the</strong> early eruptive m<strong>at</strong>erial was particularly mafic<br />

with olivine compositions up to Fo 88·4. We see no evidence<br />

<strong>of</strong> such Mg-rich olivines in <strong>the</strong> early erupted lavas.<br />

Average olivine compositions become progressively more<br />

fayalitic through Phase 3a and <strong>into</strong> Phase 3b before<br />

becoming more forsteritic <strong>at</strong> <strong>the</strong> end <strong>of</strong> <strong>the</strong> eruption<br />

(Type I ¼ Fo 80·9, Type II¼Fo 79·4; Fig. 4). As previously<br />

st<strong>at</strong>ed, both Type I and Type II inclusions may be found<br />

within <strong>the</strong> same olivine grain. In <strong>the</strong>se instances, with few<br />

exceptions, <strong>the</strong> olivine host measured adjacent to <strong>the</strong> inclusions<br />

is not appreciably different between <strong>the</strong> two types.<br />

Orthopyroxene<br />

Orthopyroxene grains from Phase 3b lavas are more<br />

variable, in terms <strong>of</strong> Mg-number [molar Mg/<br />

(Mg þ Fe T) ¼ 72^82], than <strong>the</strong> coexisting olivine grains<br />

(Fig. 4). Despite <strong>the</strong> limited number <strong>of</strong> Type II melt inclusions<br />

found in orthopyroxenes, host compositions for both<br />

Type I and Type II inclusions are indistinguishable. The<br />

overall range <strong>of</strong> orthopyroxene compositions is constant<br />

through eruptive Phase 3b; however, <strong>the</strong> average<br />

Mg-number decreases from 81 to 79 from 1949 to 1952, in<br />

2204<br />

contrast to <strong>the</strong> increase in average Fo-number in olivine<br />

grains over <strong>the</strong> same time interval (Fig. 4).<br />

Crustal xenoliths<br />

Crustal xenoliths are predominantly granitic in composition<br />

(70^76 wt % SiO 2), with melting textures ranging<br />

from veinlets cross-cutting intact granitic fragments to<br />

completely pumiceous rhyolitic glass. Samples <strong>of</strong> xenoliths<br />

cover <strong>the</strong> full textural range, thus allowing us to potentially<br />

examine both bulk and partial assimil<strong>at</strong>ion <strong>of</strong> crustal<br />

xenoliths to better constrain <strong>the</strong> n<strong>at</strong>ure and timing <strong>of</strong> assimil<strong>at</strong>ion.<br />

Older trace element analyses <strong>of</strong> xenoliths were<br />

obtained by a combin<strong>at</strong>ion <strong>of</strong> techniques, including X-ray<br />

fluorescence, <strong>at</strong>omic absorption and neutron activ<strong>at</strong>ion<br />

(McBirney et al., 1987). New trace element analyses by<br />

ICP-MS provide a more internally consistent d<strong>at</strong>aset<br />

(Table 2). Bulk xenoliths are predominantly depleted in<br />

high field strength elements (HFSE; Nb, Ta, Zr, Hf, Ti)<br />

and enriched in LILE (e.g. Cs, Rb, K) rel<strong>at</strong>ive to <strong>the</strong><br />

whole-rock lava compositions (Table 2, Fig.7). The xenolith<br />

compositions fall <strong>into</strong> two groups, with one group having<br />

higher Al 2O 3 and Sr and lower K, Rb, Y, HREE, Th and<br />

87 Sr/ 86 Sr compared with <strong>the</strong> o<strong>the</strong>r group. Rel<strong>at</strong>ive to <strong>the</strong><br />

basalts, <strong>the</strong> bulk xenoliths generally have slightly higher<br />

Zr/Nb (12^30), and higher Ba/Nb (32^180), Th/Nb<br />

(0·5^4·1) and Th/Yb (1·8^12·3). By comparison, lower crustal<br />

xenoliths from <strong>the</strong> Valle de Santiago Maar field in <strong>the</strong><br />

nor<strong>the</strong>rn Michoacan^Guanaju<strong>at</strong>o volcanic field have comparable<br />

Sr abundances but are depleted in LREE, Zr,<br />

Rb, and Th and enriched in Yb rel<strong>at</strong>ive to Paricutin xenoliths<br />

(Urrutia-Fucugauchi & Uribe-Cifuentes, 1999).<br />

Glass from three crustal xenoliths with textures varying<br />

from partially melted to frothy and highly vesicular<br />

(Fig. 9) was analyzed to determine <strong>the</strong> variability in<br />

major and trace element abundances (Table 6). Glass compositions<br />

are predominantly high-silica rhyolite (SiO 2<br />

71^79%), with only a few dacite analyses (SiO 2 65^68%).<br />

In each xenolith, <strong>the</strong> glass compositions have higher K 2O,<br />

lower Na2O and much lower Th contents than <strong>the</strong> bulk<br />

composition, but o<strong>the</strong>rwise exhibit similar compositional<br />

fe<strong>at</strong>ures. For example, xenolith 116289-23 has higher K 2O<br />

and lower FeO, MgO, Sr, Ba and LREE contents<br />

compared with <strong>the</strong> o<strong>the</strong>r two xenoliths; <strong>the</strong>se differences<br />

are also seen in <strong>the</strong> glass compositions. In general, <strong>the</strong><br />

dacitic glasses are more trace element enriched<br />

compared with <strong>the</strong> rhyolitic glasses. Sr concentr<strong>at</strong>ions<br />

range from 9 to 41000 ppm whereas Ba ranges from<br />

31 to 691ppm. Ti is similarly variable, with abundances<br />

from 14 to 9500 ppm (Table 6). The glasses show a significantly<br />

gre<strong>at</strong>er range in Ba/Nb (9^6700) than <strong>the</strong> bulk<br />

xenoliths (32^180), but generally have higher values as a<br />

result <strong>of</strong> <strong>the</strong>ir high Ba and very low Nb concentr<strong>at</strong>ions.<br />

K 2O/TiO 2 r<strong>at</strong>ios are similar in <strong>the</strong> glasses and bulk xenoliths,<br />

ranging from 5 to 15, except where TiO 2 contents<br />

are 50·1 wt %, when higher r<strong>at</strong>ios (50^190) are observed.<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Fig. 9. Photomicrographs <strong>of</strong> crustal xenoliths showing varying amounts <strong>of</strong> melting. Textures range from (a) nearly completely crystalline to<br />

(b) glassy groundmass but intact grains to (c, d) mostly glassy and pumiceous. Black scale bar in each image represents 1·25 mm. Sample<br />

116289-20 (a) did not contain any glass for analysis.<br />

Rare earth element concentr<strong>at</strong>ions are generally lower in<br />

<strong>the</strong> glasses compared with <strong>the</strong> bulk xenoliths, although<br />

with significant variability and overlap <strong>of</strong> HREE<br />

abundances.<br />

DISCUSSION<br />

In <strong>the</strong> following discussion, we integr<strong>at</strong>e our new<br />

whole-rock and melt inclusion d<strong>at</strong>a with previously published<br />

d<strong>at</strong>a to develop a consistent model for <strong>the</strong> temporal<br />

and compositional development <strong>of</strong> <strong>the</strong> Paricutin magm<strong>at</strong>ic<br />

system. Earlier studies (e.g. Wilcox, 1954; McBirney et al.,<br />

1987; Luhr, 2001), including our own, were based on <strong>the</strong><br />

sample suite th<strong>at</strong> was collected as <strong>the</strong> Paricutin eruption<br />

progressed, which is archived <strong>at</strong> <strong>the</strong> Smithsonian N<strong>at</strong>ional<br />

Museum <strong>of</strong> N<strong>at</strong>ural History. Cebria¤ et al. (2011) recently<br />

published new elemental and Sr^Nd isotope d<strong>at</strong>a on a<br />

newly collected suite <strong>of</strong> lavas, whose eruptive chronology<br />

was estim<strong>at</strong>ed from s<strong>at</strong>ellite photos and <strong>the</strong> detailed descriptions<br />

<strong>of</strong> <strong>the</strong> eruptive history. O<strong>the</strong>r recent studies<br />

(Pioli et al., 2008; Erlund et al., 2010) considered newly collected<br />

sample pr<strong>of</strong>iles through <strong>the</strong> tephra deposits,<br />

2205<br />

primarily to investig<strong>at</strong>e <strong>the</strong> physical volcanology <strong>of</strong> <strong>the</strong><br />

eruption and its rel<strong>at</strong>ionship to <strong>the</strong> changing composition<br />

<strong>of</strong> <strong>the</strong> erupted magma. The sample str<strong>at</strong>igraphy and compositions<br />

allowed <strong>the</strong> tephra to be correl<strong>at</strong>ed to <strong>the</strong> different<br />

eruptive phases but not to <strong>the</strong> detailed chronology<br />

available for <strong>the</strong> Smithsonian sample collection.<br />

In <strong>the</strong> first part <strong>of</strong> <strong>the</strong> discussion we explore <strong>the</strong> link between<br />

<strong>the</strong> Type II inclusions and <strong>the</strong> tephra m<strong>at</strong>rix glasses.<br />

The remainder <strong>of</strong> <strong>the</strong> discussion focuses on temporal vari<strong>at</strong>ions<br />

in <strong>the</strong> composition <strong>of</strong> <strong>the</strong> Type I melt inclusions<br />

and <strong>the</strong>ir host lava compositions, and how <strong>the</strong>se rel<strong>at</strong>e to<br />

processes <strong>at</strong> a deeper level in <strong>the</strong> magma plumbing system.<br />

Origin <strong>of</strong> Type II high-Si melt inclusions<br />

Type II melt inclusions, identified by <strong>the</strong>ir overall higher<br />

SiO 2 concentr<strong>at</strong>ions, typically record a gre<strong>at</strong>er compositional<br />

range than <strong>the</strong> Type I inclusions within a particular<br />

whole-rock sample. They are displaced to higher TiO 2,<br />

P 2O 5 and K 2O values and lower Al 2O 3 and CaO values<br />

rel<strong>at</strong>ive to <strong>the</strong> Type I inclusions and <strong>the</strong> whole-rock lava<br />

and tephra samples, but overlap in composition with <strong>the</strong><br />

m<strong>at</strong>rix glasses <strong>of</strong> <strong>the</strong> tephra samples (Fig. 6; Luhr, 2001;<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Table 6: Represent<strong>at</strong>ive major and trace element analyses <strong>of</strong> crustal xenolith glasses<br />

Sample: 116289-23 116289-23 116289-23 116289-23 116293-5 116293-5 116293-5 116293-5 116293-4 116293-4 116293-4 116293-4 116293-4<br />

<strong>An</strong>alysis: 23-1-2 23-1-4 23-1-7 23-1-8 5-1-1 5-1-5 5-1-7 5-1-9 4-1-1 4-1-5 4-1-6 4-1-9 4-1-11<br />

Major elements (wt %)<br />

SiO2 71·98 75·33 73·32 68·39 75·80 64·92 73·43 77·30 74·52 77·03 77·10 71·74 64·13<br />

TiO2 0·00 0·07 0·00 0·05 0·04 0·00 0·07 0·03 0·42 0·00 0·00 0·25 1·14<br />

Al2O3 16·04 13·61 14·82 18·32 13·74 21·70 14·84 12·83 12·98 12·95 12·93 14·31 13·94<br />

FeO 0·19 0·34 0·13 0·13 1·13 0·89 1·60 1·18 2·69 1·23 1·12 3·57 7·60<br />

MnO 0·00 0·00 0·00 0·00 0·02 0·00 0·03 0·00 0·14 0·01 0·04 0·18 0·29<br />

MgO 0·07 0·09 0·03 0·03 0·26 0·17 0·41 0·24 0·80 0·45 0·35 0·99 2·24<br />

CaO 0·48 0·38 0·32 0·82 0·54 4·24 0·77 0·58 2·29 1·59 1·50 1·90 2·75<br />

Na2O 4·87 3·97 4·35 5·54 4·54 6·29 4·70 4·07 3·89 4·51 4·55 4·49 4·98<br />

K2O 6·53 6·07 6·43 6·97 3·59 2·23 3·71 3·50 2·22 2·60 2·50 2·56 1·36<br />

S 0·00 0·004 0·008 0·00 0·00 0·005 0·006 0·00 0·002 0·00 0·008 0·004 0·003<br />

Cl 0·012 0·012 0·011 0·009 0·018 0·005 0·010 0·013 0·012 0·011 0·009 0·024 0·015<br />

Total 100·18 99·88 99·43 100·23 99·67 100·45 99·58 99·74 99·96 100·39 100·11 100·02 98·54<br />

Trace elements (ppm)<br />

Ti 15·3 18·4 14·1 17·2343 92·8 253 223 1642 47·9 23·11848 9497<br />

Sr 11·2 11·3 9·1 99·1129 1016 210 135 291 280 467 226 445<br />

Y 6·3 2·9 1·8 5·7 5·1 13·2 5·2 4·8 16·2 16·6 2·8 10·3 81·0<br />

Zr 2·4 0·1 1·1 0·1 31·2 1·7562 30·3 24·8370 5·2 93·3 1003<br />

Nb 0·3 0·1 0·1 0·1 2·1 0·6 1·7 1·9 7·4 0·3 0·1 10·7 75·2<br />

Ba 43·0 53·3 31·7250 420 867 691 408 428 576 670 510 676<br />

La 1·3 1·1 1·5 1·4 6·8151 6·6 5·4 9·1 4·2 3·3 10·9 46·2<br />

Ce 1·4 1·7 2·3 1·9 12·2227 11·9 8·9 16·4 7·7 5·1 17·7 101<br />

Nd 0·6 0·6 0·9 0·5 3·6 62·9 4·0 3·2 5·6 3·0 1·8 5·8 44·1<br />

Sm 0·5 0·2 0·2 0·3 0·8 10·3 0·9 0·5 1·5 0·5 0·3 1·5 11·2<br />

Eu 0·2 0·3 0·1 0·8 2·5 2·9 3·0 2·3 2·1 2·6 2·3 2·9 2·9<br />

Gd 0·6 0·4 0·2 0·5 0·9 7·4 1·2 1·1 2·6 1·2 0·5 2·1 13·2<br />

Dy 1·3 0·6 0·5 1·0 1·4 5·6 1·5 1·0 4·0 2·4 0·4 2·3 18·9<br />

Yb 0·6 0·3 0·2 0·8 0·6 0·7 0·7 0·6 1·6 3·6 0·4 1·0 7·2<br />

Th 0·2 0·3 0·2 0·2 0·7 0·2 0·7 0·6 0·2 0·1 0·4 0·4 0·4<br />

Erlund et al., 2010). Despite <strong>the</strong> rel<strong>at</strong>ively high forsterite<br />

content <strong>of</strong> <strong>the</strong> host olivine (Fo 82·0^77·5) Erlund et al. (2010)<br />

demonstr<strong>at</strong>ed th<strong>at</strong> <strong>the</strong> olivine compositions were in equilibrium<br />

with <strong>the</strong> tephra m<strong>at</strong>rix glass. Luhr (2001) noted<br />

th<strong>at</strong> tie-lines connecting bulk tephra samples with <strong>the</strong>ir<br />

m<strong>at</strong>rix glass compositions were different in orient<strong>at</strong>ion<br />

(e.g. increasing K 2O and TiO 2) from <strong>the</strong> main temporal<br />

trend through <strong>the</strong> bulk lava and tephra samples (e.g.<br />

increasing K 2O and decreasing TiO 2). Luhr (2001) and<br />

Erlund et al. (2010) showed th<strong>at</strong> <strong>the</strong> compositions <strong>of</strong> each<br />

tephra m<strong>at</strong>rix glass could be derived from its bulk tephra<br />

composition through significant fractional crystalliz<strong>at</strong>ion<br />

(up to 40%) <strong>of</strong> plagioclase þ olivine ( orthopyroxene<br />

in <strong>the</strong> Phase 3 samples), phases th<strong>at</strong> form <strong>the</strong> groundmass<br />

<strong>of</strong> <strong>the</strong> Paricutin lavas. The constant K 2O/TiO 2 <strong>of</strong> each<br />

2206<br />

bulk tephra^m<strong>at</strong>rix glass pair indic<strong>at</strong>e th<strong>at</strong> titanomagnetite<br />

is not part <strong>of</strong> this crystallizing assemblage, consistent<br />

with petrographic observ<strong>at</strong>ions and experimental constraints<br />

(Eggler, 1972; McBirney et al., 1987). Similarly, <strong>the</strong><br />

constant P2O5/TiO2 r<strong>at</strong>ios indic<strong>at</strong>e an absence <strong>of</strong> ap<strong>at</strong>ite<br />

crystalliz<strong>at</strong>ion.<br />

Major element vari<strong>at</strong>ions indic<strong>at</strong>e th<strong>at</strong> <strong>the</strong> Type II melt<br />

inclusions were formed through a similar crystalliz<strong>at</strong>ion<br />

process to <strong>the</strong> tephra m<strong>at</strong>rix glasses, and <strong>the</strong> trace element<br />

d<strong>at</strong>a are also consistent with a model domin<strong>at</strong>ed by fractional<br />

crystalliz<strong>at</strong>ion <strong>of</strong> plagioclase þ olivine orthopyroxene.<br />

Trace elements incomp<strong>at</strong>ible in <strong>the</strong>se phases (e.g.<br />

Rb, Ba, Y, Zr, Nb, REE) are all enriched in <strong>the</strong> Type II inclusions<br />

rel<strong>at</strong>ive to <strong>the</strong> Type I inclusions and whole-rock<br />

samples, whereas Sr contents (comp<strong>at</strong>ible in plagioclase)<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

are lower (Fig. 4), consistent with <strong>the</strong> lower CaO and<br />

Al 2O 3 contents (Figs 4 and 6). This indic<strong>at</strong>es th<strong>at</strong> Type II<br />

melt inclusions record a very l<strong>at</strong>e-stage crystalliz<strong>at</strong>ion<br />

event, as feldspar is not a phenocryst phase in most <strong>of</strong><br />

<strong>the</strong>se samples but is present in <strong>the</strong> groundmass.<br />

Although <strong>the</strong> Type II inclusions typically record a large<br />

compositional range, <strong>the</strong>re is little system<strong>at</strong>ic vari<strong>at</strong>ion in<br />

<strong>the</strong> inclusion compositions th<strong>at</strong> would support progressive<br />

assimil<strong>at</strong>ion in <strong>the</strong> evolution <strong>of</strong> Type II inclusions (e.g.<br />

increasing indices <strong>of</strong> crustal contamin<strong>at</strong>ion with melt evolution<br />

or host phenocryst composition). K2O/TiO2 r<strong>at</strong>ios<br />

are typically considered an excellent indic<strong>at</strong>or <strong>of</strong> contamin<strong>at</strong>ion<br />

in Paricutin lavas (e.g. McBirney et al., 1987; Luhr,<br />

2001; Erlund et al., 2010), as both elements are incomp<strong>at</strong>ible<br />

in <strong>the</strong> crystallizing assemblage, whereas likely crustal<br />

assimilants have significantly higher K 2O/TiO 2 (e.g. typically<br />

6^13 in <strong>the</strong> bulk crustal xenoliths and 10^8000 in<br />

<strong>the</strong> xenolith glasses; Tables 2 and 6) compared with <strong>the</strong><br />

lavas (K 2O/TiO 2 0·8^2·1). Therefore, contamin<strong>at</strong>ed<br />

magmas will have higher K 2O/TiO 2 values, and yet<br />

K 2O/TiO 2 does not significantly increase with SiO 2 in<br />

Type II inclusions within a single sample (Fig. 10).<br />

Although <strong>the</strong>re is no direct correl<strong>at</strong>ion with SiO 2, <strong>the</strong><br />

range in K 2O/TiO 2 values <strong>at</strong> constant SiO 2 observed<br />

in some Type II inclusions may record minor<br />

assimil<strong>at</strong>ion; however, this vari<strong>at</strong>ion may also simply be<br />

<strong>the</strong> result <strong>of</strong> n<strong>at</strong>ural vari<strong>at</strong>ion in very l<strong>at</strong>e-stage groundmass<br />

crystalliz<strong>at</strong>ion (see below). Trace element r<strong>at</strong>ios such<br />

as Ba/Nb th<strong>at</strong> are higher in contamin<strong>at</strong>ed magmas are<br />

likewise similar to whole-rock r<strong>at</strong>ios for single samples,<br />

except for high Ba/Nb in Type II inclusions from<br />

one <strong>of</strong> <strong>the</strong> Phase 3 samples (116289-19) toward <strong>the</strong> end<br />

<strong>of</strong> <strong>the</strong> eruption, from which <strong>the</strong>re is also significant<br />

evidence for crustal assimil<strong>at</strong>ion from whole-rock<br />

compositions. Indices <strong>of</strong> contamin<strong>at</strong>ion <strong>the</strong>refore suggest<br />

th<strong>at</strong> no significant assimil<strong>at</strong>ion occurred during <strong>the</strong><br />

l<strong>at</strong>e-stage crystalliz<strong>at</strong>ion and evolution <strong>of</strong> <strong>the</strong> Type II melt<br />

inclusions.<br />

It was observed <strong>at</strong> <strong>the</strong> time <strong>of</strong> eruption th<strong>at</strong> although<br />

denser and rel<strong>at</strong>ively degassed lavas, including <strong>the</strong> lava<br />

samples from this study, were erupted from a vent site <strong>at</strong><br />

<strong>the</strong> base <strong>of</strong> <strong>the</strong> cone, <strong>the</strong> cr<strong>at</strong>er remained <strong>the</strong> site <strong>of</strong> continued<br />

degassing and explosive eruptions (Krauskopf, 1948).<br />

To explain this essentially simultaneous activity <strong>at</strong> two different<br />

vent sites, more recent models have suggested a shallow<br />

separ<strong>at</strong>ion <strong>of</strong> vol<strong>at</strong>iles <strong>at</strong> very shallow levels, ei<strong>the</strong>r <strong>at</strong><br />

<strong>the</strong> base <strong>of</strong> <strong>the</strong> cone or just below, after which point <strong>the</strong><br />

more degassed magma erupts from <strong>the</strong> base <strong>of</strong> Paricutin<br />

(Krauskopf, 1948; McBirney et al., 1987; Pioli et al., 2008).<br />

This model is supported by <strong>the</strong> low sulfur concentr<strong>at</strong>ions<br />

Fig. 10. K2O/TiO2 vs SiO2 for a represent<strong>at</strong>ive sample <strong>of</strong> Phase 2 (116295-23) and <strong>of</strong> Phase 3b (116289-19) melt inclusions (Type I and Type II<br />

inclusions).<br />

2207<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

in Type II inclusions rel<strong>at</strong>ive to Type I inclusions (Fig. 5),<br />

potentially indic<strong>at</strong>ing low-pressure degassing ei<strong>the</strong>r before<br />

or during inclusion entrapment. Additionally, tephra samples<br />

are notably lacking in Type II melt inclusions, with<br />

<strong>the</strong> exception <strong>of</strong> only a few inclusions in a single sample<br />

(Luhr, 2001). This suggests th<strong>at</strong> entrapment <strong>of</strong> <strong>the</strong> Type II<br />

inclusions probably occurred <strong>at</strong> shallow levels, after separ<strong>at</strong>ion<br />

<strong>of</strong> a gas-rich component in <strong>the</strong> main conduit. This is<br />

supported by sulfur concentr<strong>at</strong>ions in Type II inclusions<br />

above th<strong>at</strong> <strong>of</strong> <strong>the</strong> tephra glasses reported by Luhr (2001),<br />

again suggesting th<strong>at</strong> crystalliz<strong>at</strong>ion <strong>of</strong> Type II inclusions<br />

did in fact occur shortly before complete degassing and<br />

syn- or post-eruptive cooling <strong>of</strong> <strong>the</strong> erupted lavas.<br />

The presence <strong>of</strong> high-Si melt inclusions in rel<strong>at</strong>ively high<br />

forsterite olivine may be explained if <strong>the</strong> inclusions were<br />

trapped within embayments or necks in <strong>the</strong> olivine th<strong>at</strong><br />

were sealed very l<strong>at</strong>e, such th<strong>at</strong> <strong>the</strong> olivine and melt were<br />

not in equilibrium. This would explain <strong>the</strong> similar host forsterite<br />

content betweenType I and Type II inclusions, such<br />

th<strong>at</strong> inclusions are trapped predominantly in <strong>the</strong> higher<br />

forsterite host (except where sealed <strong>of</strong>f). Rowe et al. (2011)<br />

identified a similar high-Si inclusion popul<strong>at</strong>ion from a<br />

lava flow <strong>at</strong> Dotsero Volcano, Colorado. Textural evidence<br />

indic<strong>at</strong>ed th<strong>at</strong> <strong>the</strong> Dotsero high-Si popul<strong>at</strong>ion resulted<br />

from closing <strong>of</strong>f <strong>of</strong> embayments in olivine hosts, resulting<br />

in <strong>the</strong> juxtaposition <strong>of</strong> higher Si, fraction<strong>at</strong>ed melts<br />

with rel<strong>at</strong>ively high forsterite olivine. Given <strong>the</strong> twodimensional<br />

n<strong>at</strong>ure <strong>of</strong> <strong>the</strong> polished melt inclusions, however,<br />

this is difficult to assess. High-Si melt inclusions in<br />

Fig. 2 show no obvious signs <strong>of</strong> elong<strong>at</strong>ion th<strong>at</strong> may support<br />

this model and do not appear distinctive in form from<br />

Type I inclusions. Rehomogeniz<strong>at</strong>ion <strong>of</strong> <strong>the</strong> melt inclusions,<br />

had <strong>the</strong>y not been completely sealed, would have resulted<br />

in complete vol<strong>at</strong>ile loss (S, Cl), so it is likely th<strong>at</strong><br />

<strong>the</strong>se inclusions were completely enclosed, in contrast to<br />

<strong>the</strong> high-Si inclusion popul<strong>at</strong>ion <strong>at</strong> Dotsero Volcano<br />

(Rowe et al., 2011). Regardless <strong>of</strong> <strong>the</strong> method <strong>of</strong> entrapment,<br />

<strong>the</strong> assumption <strong>of</strong> olivine^melt equilibrium in <strong>the</strong> correction<br />

process does not strongly affect <strong>the</strong>se melt compositions<br />

(<strong>the</strong> correction is generally less than addition or<br />

subtraction <strong>of</strong> 4 wt % olivine), as indic<strong>at</strong>ed by comparing<br />

<strong>the</strong> measured and corrected melt compositions for this<br />

popul<strong>at</strong>ion (see Supplementary D<strong>at</strong>a), nor does it affect<br />

our interpret<strong>at</strong>ions and conclusions.<br />

Eruptive Phase 1, a compositionally<br />

distinct initial magma b<strong>at</strong>ch<br />

The earliest magm<strong>at</strong>ism <strong>at</strong> Paricutin (eruptive Phase 1)<br />

was only sparsely sampled during <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption,<br />

despite <strong>the</strong> fact th<strong>at</strong> it makes up 15% <strong>of</strong> <strong>the</strong> total<br />

erupted volume (McBirney et al., 1987); however, coverage<br />

has been improved recently by extensive sampling <strong>of</strong> <strong>the</strong><br />

basal tephras (Pioli et al., 2008; Johnson et al., 2009;<br />

Erlund et al., 2010). These d<strong>at</strong>a confirm th<strong>at</strong> <strong>the</strong>re is a distinct<br />

compositional break between <strong>the</strong> Phase 1 and Phase<br />

2208<br />

2 magmas, highlighted by a clear gap in K 2O contents<br />

(Phase 1 51 wt %; Phase 2 41 wt%;Luhr,2001;Erlund<br />

et al., 2010). Although <strong>the</strong>re is compositional overlap for<br />

o<strong>the</strong>r major elements, <strong>the</strong> whole-rock compositions <strong>of</strong><br />

Phase 1 magmas have comparable MgO and lower SiO 2<br />

compared with <strong>the</strong> Phase 2 lavas, whereas Phase 1 melt inclusions<br />

on average have higher SiO 2 and lower MgO<br />

compared with Phase 2 lavas (Fig. 4; Tables 2 and 5).<br />

Luhr (2001) and Erlund et al. (2010) showed th<strong>at</strong> fractional<br />

crystalliz<strong>at</strong>ion crustal assimil<strong>at</strong>ion could not explain<br />

<strong>the</strong>se minor and major element differences, and argued<br />

th<strong>at</strong> Phases 1 and 2 were characterized by separ<strong>at</strong>e, compositionally<br />

distinct, magma b<strong>at</strong>ches.<br />

Trace element d<strong>at</strong>a confirm this model <strong>of</strong> distinct<br />

magma b<strong>at</strong>ches. Despite having more primitive compositions<br />

(higher MgO, higher Cr and Ni; McBirney et al.,<br />

1987), Phase 2 lavas are enriched in nearly all incomp<strong>at</strong>ible<br />

trace elements (excluding Sr, Ti, and Sc) rel<strong>at</strong>ive to Phase<br />

1 lavas (Table 2). Basal phase 1 tephras, identified by<br />

Johnson et al. (2009) and Erlund et al. (2010), however,<br />

have lower SiO 2 and higher MgO concentr<strong>at</strong>ions and a<br />

more depleted incomp<strong>at</strong>ible trace element sign<strong>at</strong>ure<br />

(excluding Sr, which is comparable with Phase 1 lavas)<br />

than ei<strong>the</strong>r Phase 1 or 2 lavas (Table 2; Johnson et al.,<br />

2009). Fractional crystalliz<strong>at</strong>ion <strong>of</strong> olivine plagioclase<br />

cannot produce <strong>the</strong> observed differences in r<strong>at</strong>ios <strong>of</strong><br />

highly incomp<strong>at</strong>ible elements between Phase 1 (e.g. Zr/Nb<br />

21) and Phase 2 (e.g. Zr/Nb 17) magmas (Fig. 11).<br />

There is also no change in <strong>the</strong> calcul<strong>at</strong>ed Eu anomaly between<br />

Phase 1 and Phase 2 lavas, fur<strong>the</strong>r suggesting th<strong>at</strong><br />

fraction<strong>at</strong>ion <strong>of</strong> plagioclase was not driving <strong>the</strong> changes in<br />

trace element concentr<strong>at</strong>ions. However, within Phase 1,<br />

melt inclusions (Type I and Type II) indic<strong>at</strong>e concurrent<br />

plagioclase and olivine fraction<strong>at</strong>ion, based on an increasing<br />

neg<strong>at</strong>ive Eu anomaly and decreasing Sr concentr<strong>at</strong>ions<br />

(although more sc<strong>at</strong>tered) with decreasing host forsterite<br />

content. Trace element compositions <strong>of</strong> melt inclusions<br />

from Phase 1 (February^July 1943) lavas also support a<br />

model in which <strong>the</strong> Phase 1 lavas are derived from a different<br />

magma b<strong>at</strong>ch compared with <strong>the</strong> subsequent Phase 2<br />

lavas. For r<strong>at</strong>ios <strong>of</strong> highly incomp<strong>at</strong>ible elements, such as<br />

Zr/Nb, <strong>the</strong>re is no compositional overlap <strong>of</strong> Type I inclusions<br />

in magmas <strong>of</strong> <strong>the</strong> two eruptive phases: inclusions in<br />

Phase 1 magmas have Zr/Nb <strong>of</strong> 18^20, whereas inclusions<br />

in Phase 1 magmas have Zr/Nb <strong>of</strong> 15^17, consistent with<br />

<strong>the</strong> observed difference in <strong>the</strong> whole-rock compositions.<br />

Phase 1 magmas have similar Th/Nb whole-rock r<strong>at</strong>ios to<br />

<strong>the</strong> Phase 2 magmas, but higher Ba/Nb r<strong>at</strong>ios (Phase 1<br />

450, Phase 2 550); <strong>the</strong>se differences cannot be explained<br />

by assimil<strong>at</strong>ion <strong>of</strong> crustal m<strong>at</strong>erial similar to <strong>the</strong> analyzed<br />

xenoliths, which have elev<strong>at</strong>ed Ba/Nb (average <strong>of</strong> 95) and<br />

Th/Nb rel<strong>at</strong>ive to <strong>the</strong> lavas (Table 2). Fur<strong>the</strong>r evidence<br />

th<strong>at</strong> shallow crustal assimil<strong>at</strong>ion is not <strong>the</strong> cause for <strong>the</strong><br />

distinct change in composition comes from <strong>the</strong><br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Fig. 11. (a) Whole-rock vari<strong>at</strong>ion <strong>of</strong> Zr/Nb vs K2O/TiO2. Included for comparison are bulk tephra analyses from Luhr (2001) and Cebria¤ et al.<br />

(2011), incorpor<strong>at</strong>ed <strong>into</strong> each eruptive phase based on eruption d<strong>at</strong>e. Continuous curves represent simple mixing lines between <strong>the</strong> average<br />

xenolith composition and xenolith 116293-5 (Table 2) and a Phase 2 and an initial Phase 3 lava. The average xenolith (used in previous studies)<br />

and 116293-5 (best fit to Phase 3 assimil<strong>at</strong>ion by a single xenolith) compositions are utilized to demonstr<strong>at</strong>e potential assimil<strong>at</strong>ion in Phase 3<br />

and to illustr<strong>at</strong>e th<strong>at</strong> assimil<strong>at</strong>ion by Phase 2 lavas cannot produce Phase 3 lavas. Tick marks are 5 wt % increments. The discrepancy between<br />

Phase 1 and Phase 2 lavas should be noted.‘Assimil<strong>at</strong>ion’ and ‘mixing’ trends are schem<strong>at</strong>ic only, representing potential assimil<strong>at</strong>ion <strong>of</strong> average<br />

xenoliths by Phase 1 lavas and mixing between Phase 2 and Phase 3a (discussed in <strong>the</strong> text). Melt inclusions follow similar trends but are<br />

<strong>of</strong>fset probably as a result <strong>of</strong> a calibr<strong>at</strong>ion difference between <strong>the</strong> SIMS and ICP-MS and are <strong>the</strong>refore not plotted. (b) Ba/Nb vs K2O/TiO2<br />

for whole-rock (large symbols) and melt inclusions (small symbols). Also plotted are <strong>the</strong> whole-rock analyses <strong>of</strong> Luhr (2001), Cebria¤ et al.<br />

(2011), and <strong>the</strong> basal tephra melt inclusion analyses from Johnson et al. (2009).<br />

2209<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

indistinguishable 87 Sr/ 86 Sr r<strong>at</strong>ios <strong>of</strong> Phase 1 and Phase 2<br />

lavas despite <strong>the</strong> elev<strong>at</strong>ed 87 Sr/ 86 Sr <strong>of</strong> <strong>the</strong> local crust<br />

(McBirney et al., 1987).<br />

Comparing melt inclusion and host crystal compositions<br />

provides a means to evalu<strong>at</strong>e <strong>the</strong> possible links between<br />

contamin<strong>at</strong>ion and fraction<strong>at</strong>ion during <strong>the</strong> evolution <strong>of</strong><br />

Phase 1 magmas using olivine forsterite content as an<br />

index <strong>of</strong> fraction<strong>at</strong>ion. Indices <strong>of</strong> crustal assimil<strong>at</strong>ion (e.g.<br />

Ba/Nb, K 2O/TiO 2, Sr/Nd) in olivine-hosted melt inclusions<br />

(Type I) show no system<strong>at</strong>ic evidence <strong>of</strong> progressive<br />

contamin<strong>at</strong>ion with melt evolution. K 2O/TiO 2 and Cl/K<br />

r<strong>at</strong>ios remain constant with decreasing forsterite content<br />

(from 83 to 78). It has previously been suggested th<strong>at</strong> Cl/<br />

K may provide a means to identify crustal contamin<strong>at</strong>ion<br />

(Rowe et al., 2009), given <strong>the</strong> rel<strong>at</strong>ively high K 2O content<br />

<strong>of</strong> average upper crustal rocks (Rudnick & Gao, 2004)<br />

and <strong>the</strong> neg<strong>at</strong>ive correl<strong>at</strong>ion between Cl/K and o<strong>the</strong>r indices<br />

<strong>of</strong> contamin<strong>at</strong>ion, such as Ba/Nb, SiO 2 or K 2O/TiO 2<br />

in olivine-hosted melt inclusions (Fig. 8). Ba/Nb is more<br />

variable in Phase 1 melt inclusions, ranging from 53 to 60.<br />

However, <strong>the</strong>se vari<strong>at</strong>ions are non-system<strong>at</strong>ic and may reflect<br />

minor heterogeneity <strong>of</strong> <strong>the</strong> melt.<br />

Type I melt inclusions record significant differences in<br />

vol<strong>at</strong>ile concentr<strong>at</strong>ions between Phase 1 and Phase 2 samples.<br />

Phase 1 magmas from this study have higher Cl<br />

(1380^900 ppm) and lower S contents (360^590 ppm) compared<br />

with <strong>the</strong> subsequent Phase 2 magmas (Cl 290^<br />

950 ppm; S 490^930 ppm). However, based on <strong>the</strong> melt inclusion<br />

d<strong>at</strong>a from Luhr (2001) and Johnson et al. (2009)<br />

Phase 1 melt inclusions from tephra samples have higher S<br />

than Phase 2 melt inclusions. Cl/K r<strong>at</strong>ios <strong>of</strong> Type I inclusions<br />

in <strong>the</strong> Phase 1 and Phase 2 samples are also significantly<br />

different, with Cl/K in Phase 1 samples higher than<br />

l<strong>at</strong>er erupted m<strong>at</strong>erial as a result <strong>of</strong> both lower K 2O and<br />

higher Cl in <strong>the</strong> Phase 1 melt inclusions (Fig. 8). W<strong>at</strong>er contents<br />

in melt inclusions from all Paricutin eruptive phases<br />

range from 1·3 to4·2 wt % (Luhr, 2001; Pioli et al., 2008),<br />

and show no system<strong>at</strong>ic vari<strong>at</strong>ions during <strong>the</strong> eruption.<br />

Several melt inclusions from <strong>the</strong> Phase 1 samples contain<br />

measurable magm<strong>at</strong>ic CO 2 contents th<strong>at</strong> range from 250<br />

to 1000 ppm (Luhr, 2001; Pioli et al., 2008; Johnson et al.,<br />

2009), whereas <strong>the</strong> l<strong>at</strong>er erupted m<strong>at</strong>erial is essentially<br />

devoid <strong>of</strong> CO2. This implies th<strong>at</strong> olivine crystalliz<strong>at</strong>ion<br />

took place over a range <strong>of</strong> pressures from 20 to 400 MPa<br />

(51km to 13 km) in <strong>the</strong> Phase 1 magmas, whereas olivine<br />

crystalliz<strong>at</strong>ion in <strong>the</strong> subsequent Phase 2 and 3 magmas<br />

occurred <strong>at</strong> shallower levels ( 200 MPa; 6·6kmdepth).<br />

Therefore, Phase 1 magmas appear to have been erupted<br />

from a deeper crustal level and possess distinctly different<br />

major and trace element compositions rel<strong>at</strong>ive to l<strong>at</strong>er<br />

erupted m<strong>at</strong>erial. New trace element d<strong>at</strong>a suggest ei<strong>the</strong>r<br />

th<strong>at</strong> <strong>the</strong> Phase 1 and Phase 2 magma b<strong>at</strong>ches came from<br />

compositionally different mantle sources or th<strong>at</strong> <strong>the</strong> incomp<strong>at</strong>ible<br />

element differences result from assimil<strong>at</strong>ion in<br />

2210<br />

<strong>the</strong> lower crust <strong>of</strong> m<strong>at</strong>erial th<strong>at</strong> is distinct from <strong>the</strong> entrained<br />

xenoliths; this would require additional isotopic<br />

d<strong>at</strong>a (Nd^Pb^O) to evalu<strong>at</strong>e fur<strong>the</strong>r.<br />

Origin <strong>of</strong> Type I low-Si melt inclusions<br />

(Eruptive Phases 2 and 3)<br />

As previously documented, Type I inclusions have characteristically<br />

lower SiO 2 concentr<strong>at</strong>ions rel<strong>at</strong>ive to<br />

whole-rock samples but o<strong>the</strong>rwise have major and trace<br />

element temporal vari<strong>at</strong>ions th<strong>at</strong> m<strong>at</strong>ch those <strong>of</strong> <strong>the</strong><br />

whole-rock samples. The system<strong>at</strong>ic <strong>of</strong>fset between <strong>the</strong><br />

Type I inclusions and <strong>the</strong> whole-rock compositions probably<br />

reflects a combin<strong>at</strong>ion <strong>of</strong> crystalliz<strong>at</strong>ion <strong>of</strong> <strong>the</strong> melt,<br />

minor contamin<strong>at</strong>ion, and crystal accumul<strong>at</strong>ion in <strong>the</strong><br />

magma after inclusion entrapment. Erlund et al. (2010)<br />

observed th<strong>at</strong> some Phase 2 melt inclusions (Luhr, 2001),<br />

as well as early erupted bombs (Foshag & Gonzalez-<br />

Reyna, 1956), similarly display system<strong>at</strong>ically low SiO2<br />

concentr<strong>at</strong>ions. Phase 2 melt inclusions from tephra samples<br />

also have Al 2O 3 concentr<strong>at</strong>ions higher than tephra<br />

glass (Luhr, 2001), consistent with <strong>the</strong> observed<br />

high-Al 2O 3 melt inclusions in olivine from <strong>the</strong> lava samples,<br />

but with similar CaO/Al 2O 3 r<strong>at</strong>ios to <strong>the</strong> host<br />

magma.<br />

For <strong>the</strong> remainder <strong>of</strong> <strong>the</strong> discussion on melt evolution<br />

and Type I inclusions we focus on trace element abundances<br />

in both whole-rock samples and melt inclusions.<br />

Homogeneity <strong>of</strong> Eruptive Phase 2<br />

samplesçany melt inclusion evidence<br />

for assimil<strong>at</strong>ion?<br />

Lavas and tephras erupted during Phase 2 (July 1943 to<br />

1946) show limited major element variability, with MgO<br />

<strong>of</strong> 4·7^5·3 wt % and SiO 2 <strong>of</strong> 55·1^56·6 wt %. As noted by<br />

Erlund et al. (2010), <strong>the</strong> Phase 2 tephra are remarkably<br />

homogeneous in terms <strong>of</strong> whole-rock composition, mineral<br />

compositions, and groundmass texture. Elemental r<strong>at</strong>ios<br />

such as K 2O/TiO 2 and Ba/Nb would be insensitive to crystal<br />

fraction<strong>at</strong>ion in <strong>the</strong> Paricutin magmas, but sensitive to<br />

minor inputs <strong>of</strong> crustal m<strong>at</strong>erial similar to <strong>the</strong> local basement<br />

and xenoliths. Compared with <strong>the</strong> preceding Phase<br />

1 magmas, <strong>the</strong> Phase 2 magmas show gre<strong>at</strong>er variability<br />

<strong>of</strong> K 2O/TiO 2 (1·0^1·3) and Ba/Nb (39^51), indic<strong>at</strong>ive <strong>of</strong><br />

minor amounts <strong>of</strong> crustal contamin<strong>at</strong>ion (Fig. 11).<br />

Vari<strong>at</strong>ions in melt inclusion composition provide an<br />

opportunity to evalu<strong>at</strong>e <strong>the</strong> timing and amount <strong>of</strong> contamin<strong>at</strong>ion<br />

rel<strong>at</strong>ive to crystalliz<strong>at</strong>ion <strong>of</strong> <strong>the</strong> magma. For <strong>the</strong><br />

two Phase 2 lava samples analyzed, K 2O/TiO 2 r<strong>at</strong>ios<br />

(1·14 0·15 and 1·23 0·11) in Type I melt inclusions are<br />

rel<strong>at</strong>ively constant and similar to whole-rock values (1·32<br />

and 1·27 respectively). This is consistent with olivinehosted<br />

melt inclusions from <strong>the</strong> tephra with an average<br />

K2O/TiO2 r<strong>at</strong>io <strong>of</strong> 1·18 0·09 compared with <strong>the</strong><br />

whole-rock tephra value <strong>of</strong> 1·27 0·02 (Luhr, 2001).<br />

Similarly, Cl/K r<strong>at</strong>ios are constant (0·08 0·015), despite a<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

range in SiO 2 <strong>of</strong> 52^54 wt %. Only one melt inclusion<br />

appears to record a more contamin<strong>at</strong>ed composition with<br />

aK 2O/TiO 2 <strong>of</strong> 1·47 and Cl/K <strong>of</strong> 0·07. This inclusion also<br />

has higher Ba/Nb (49·7) rel<strong>at</strong>ive to <strong>the</strong> whole-rock and<br />

o<strong>the</strong>r Phase 2 melt inclusions. As indices <strong>of</strong> contamin<strong>at</strong>ion<br />

(K 2O/TiO 2, Ba/Nb) have similar ranges in <strong>the</strong> wholerocks<br />

and melt inclusions, this would suggest th<strong>at</strong> any significant<br />

contamin<strong>at</strong>ion <strong>of</strong> <strong>the</strong> Phase 2 magma occurred<br />

prior to olivine crystalliz<strong>at</strong>ion and entrapment <strong>of</strong> <strong>the</strong> Type<br />

I melt inclusions. This is consistent with <strong>the</strong> generally low<br />

crystalliz<strong>at</strong>ion pressures estim<strong>at</strong>ed for <strong>the</strong>se magmas<br />

based on H 2O contents and a general lack <strong>of</strong> CO 2 in<br />

olivine-hosted melt inclusions (Luhr, 2001; Pioli et al.,<br />

2008). Erlund et al. (2010) proposed a model by which <strong>the</strong><br />

Phase 2 magmas crystallize during ascent, with early<br />

erupted magmas retaining some CO 2 whereas l<strong>at</strong>er<br />

erupted Phase 2 lavas record low crystalliz<strong>at</strong>ion pressures<br />

(5100 MPa) and <strong>the</strong> establishment <strong>of</strong> a shallow magma<br />

storage region.<br />

<strong>N<strong>at</strong>ure</strong> <strong>of</strong> <strong>the</strong> transition from Phase 2 to<br />

Phase 3<br />

The most distinctive compositional change in <strong>the</strong> Paricutin<br />

lavas occurred <strong>at</strong> <strong>the</strong> end <strong>of</strong> 1946 and through 1947, with<br />

a shift from basaltic andesite ( 56 wt % SiO2) to andesite<br />

( 60 wt % SiO 2) compositions (Wilcox, 1954) th<strong>at</strong> was<br />

accompanied by significant increases in 87 Sr/ 86 Sr and<br />

d 18 O, indic<strong>at</strong>ive <strong>of</strong> increased crustal assimil<strong>at</strong>ion<br />

(McBirney et al., 1987; Cebria¤ et al., 2011). Several studies<br />

have used <strong>the</strong> onset <strong>of</strong> this change to mark <strong>the</strong> transition<br />

between <strong>the</strong> Phase 2 and Phase 3 eruptive stages (e.g.<br />

Pioli et al., 2008; Erlund et al., 2010). However, our new<br />

trace element analyses, combined with recent liter<strong>at</strong>ure<br />

d<strong>at</strong>a (Luhr, 2001; Erlund et al., 2010; Cebria¤ et al., 2011), indic<strong>at</strong>e<br />

th<strong>at</strong>, in many respects, <strong>the</strong>re was a more fundamental<br />

compositional change almost a year earlier, sometime<br />

in early 1946, which is used here to mark <strong>the</strong> onset <strong>of</strong> <strong>the</strong><br />

Phase 3 stage. This change is most clearly defined by a<br />

shift in Zr/Nb from 15·4^18·9 in <strong>the</strong> Phase 2 magmas to<br />

higher values (19·0^22·3) in <strong>the</strong> Phase 3 magmas th<strong>at</strong> are<br />

similar to those <strong>of</strong> <strong>the</strong> Phase 1 magmas (Fig. 11). Although<br />

some <strong>of</strong> <strong>the</strong> sc<strong>at</strong>ter in Zr/Nb values is a result <strong>of</strong> slight<br />

interlabor<strong>at</strong>ory calibr<strong>at</strong>ion differences, <strong>the</strong> observ<strong>at</strong>ion <strong>of</strong><br />

a change in Zr/Nb <strong>at</strong> this time is a robust fe<strong>at</strong>ure <strong>of</strong> <strong>the</strong><br />

d<strong>at</strong>a as it is apparent in three independent studies th<strong>at</strong><br />

have analyses <strong>of</strong> both Phase 2 and Phase 3 samples (this<br />

study; Luhr, 2001; Cebria¤ et al., 2011).<br />

A striking fe<strong>at</strong>ure <strong>of</strong> <strong>the</strong> lavas erupted between mid-1943<br />

and mid-1947 th<strong>at</strong> span this shift in Zr/Nb r<strong>at</strong>ios is <strong>the</strong>ir<br />

rel<strong>at</strong>ively constant MgO content (5·2^5·8 wt %), despite<br />

small progressive increases in SiO 2, K 2O/TiO 2, and<br />

87 Sr/ 86 Sr (Fig. 12). Compared with <strong>the</strong> earlier low Zr/Nb<br />

samples, <strong>the</strong> subsequent high Zr/Nb samples have higher<br />

SiO 2 (55·4^56·6 wt % vs 56·7^57·8 wt %), K 2O/TiO 2<br />

(1·1^1·3 vs1·3^1·6), Ba/Nb (39^48 vs 52^65), Ba/La (21^22<br />

2211<br />

vs 24^26) and Th/Nb (0·18^0·21 vs 0·22^0·26), and lower<br />

Nb contents (7^8 ppm vs 8^10 ppm). Although <strong>the</strong> higher<br />

SiO 2,K 2O/TiO 2, Ba/Nb, Ba/La and Th/Nb are suggestive<br />

<strong>of</strong> an increased input <strong>of</strong> crustal m<strong>at</strong>erial, <strong>the</strong> constant<br />

MgO contents, toge<strong>the</strong>r with constant Ni (88^127 ppm),<br />

Cr (153^227 ppm) and CaO/Al 2O 3 (0·39^0·41), make it difficult<br />

to explain such vari<strong>at</strong>ions with a progressive assimil<strong>at</strong>ion<br />

and fractional crystalliz<strong>at</strong>ion model as proposed by<br />

Wilcox (1954), McBirney et al. (1987) and Cebria¤ et al.<br />

(2011). The local crustal lithologies (bulk xenoliths, xenolith<br />

glasses, basement outcrops) have average Zr/Nb in<br />

<strong>the</strong> range 20^30, which does not give sufficient leverage to<br />

account for <strong>the</strong> change in Zr/Nb in <strong>the</strong>se lavas <strong>at</strong> similar<br />

MgO values. Instead, <strong>the</strong>se d<strong>at</strong>a suggest th<strong>at</strong> <strong>the</strong> shift in<br />

Zr/Nb values represents <strong>the</strong> involvement <strong>of</strong> a new magma<br />

b<strong>at</strong>ch, compositionally distinct from <strong>the</strong> Phase 2 magma.<br />

The minor vari<strong>at</strong>ions in K 2O/TiO 2 and Ba/Nb, with limited<br />

vari<strong>at</strong>ion in SiO 2, seen in <strong>the</strong> l<strong>at</strong>er Phase 2 magmas<br />

might be better explained by mixing with this new<br />

magma b<strong>at</strong>ch ra<strong>the</strong>r than small extents <strong>of</strong> crustal<br />

assimil<strong>at</strong>ion.<br />

The exact timing <strong>of</strong> <strong>the</strong> input <strong>of</strong> this new high Zr/Nb<br />

magma b<strong>at</strong>ch <strong>into</strong> <strong>the</strong> eruptive system <strong>at</strong> Paricutin is somewh<strong>at</strong><br />

uncertain, as <strong>the</strong> Smithsonian collection has only a<br />

few samples from <strong>the</strong> critical time period from 1945 to<br />

1947. The recent study <strong>of</strong> Cebria¤ et al. (2011) improved<br />

sample coverage in this interval, but <strong>the</strong>ir positions within<br />

<strong>the</strong> eruptive chronology are known only to within<br />

4^5 months. Samples Par-2 and Par-4 from Cebria¤ et al.<br />

(2011) have high Zr/Nb and have inferred eruption d<strong>at</strong>es<br />

between October 1945 and February 1946, whereas sample<br />

116289-8 (Table 1) has low Zr/Nb and was erupted on 18<br />

September 1946. It was around this time th<strong>at</strong> <strong>the</strong>re was a<br />

change in eruptive style from explosive Strombolian activity<br />

to effusive lava flows and Vulcanian explosions, accompanied<br />

by a decrease in <strong>the</strong> average mass eruption r<strong>at</strong>e<br />

(Pioli et al., 2008). Pioli et al. (2008) noted th<strong>at</strong> this shift in<br />

eruptive style preceded <strong>the</strong> rapid compositional change<br />

from basaltic andesite to andesite compositions by several<br />

months, and this means th<strong>at</strong> it was potentially linked to<br />

<strong>the</strong> appearance <strong>of</strong> <strong>the</strong> new high Zr/Nb magma b<strong>at</strong>ch.<br />

Despite <strong>the</strong> rapid change in whole-rock compositions,<br />

<strong>the</strong> melt inclusions do appear to preserve limited overlap<br />

between <strong>the</strong> end <strong>of</strong> Phase 2 and <strong>the</strong> beginning <strong>of</strong> Phase 3,<br />

implying minor mixing <strong>of</strong> <strong>the</strong>se two phases <strong>at</strong> <strong>the</strong> onset <strong>of</strong><br />

Phase 3. Overlap in melt inclusion compositions is most<br />

evident in Cl/K, Ba/Nb, and K2O/TiO2 r<strong>at</strong>ios between<br />

early Phase 3 and l<strong>at</strong>e Phase 2 (Figs 8 and 11). Ba/Nb<br />

r<strong>at</strong>ios in Phase 3 melt inclusions are as low as 45 compared<br />

with a maximum Ba/Nb <strong>of</strong> 48 in Phase 2 melt inclusions.<br />

Additionally, as discussed above, one inclusion in a l<strong>at</strong>e<br />

Phase 2 lava has low Cl/K ( 0·7) and high K 2O/TiO 2<br />

( 1·5), comparable with <strong>the</strong> Phase 3 inclusions.<br />

Importantly, this suggests <strong>the</strong> presence <strong>of</strong> Phase 3<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 12. Whole-rock (a) SiO 2,(b)Zr/Nb,(c)K 2O/TiO 2, and (d) Ba/Nb vs MgO wt %. The significant vari<strong>at</strong>ions in indices <strong>of</strong> crustal contamin<strong>at</strong>ion<br />

with little change in MgO concentr<strong>at</strong>ion should be noted. D<strong>at</strong>a sources: this study; Wilcox (1954); McBirney et al. (1987); Luhr (2001);<br />

Verma & Hasenaka (2004); Erlund et al. (2010);Cebria¤ et al. (2011),<br />

magmas within <strong>the</strong> shallow magma storage system during<br />

<strong>the</strong> waning <strong>of</strong> Phase 2 activity. This model, incorpor<strong>at</strong>ing<br />

minor mixing between Phase 2 and Phase 3, is consistent<br />

with <strong>the</strong> apparent overlap in whole-rock compositions<br />

during this time interval.<br />

Compositional vari<strong>at</strong>ions <strong>of</strong> Phase 3<br />

samples<br />

Eruptive Phase 3 represents two-thirds <strong>of</strong> <strong>the</strong> eruption<br />

dur<strong>at</strong>ion (1946^1952) but only <strong>the</strong> last 25% <strong>of</strong> erupted<br />

m<strong>at</strong>erial (McBirney et al., 1987). Phase 3 lavas are characterized<br />

by a wide range in SiO 2 ( 56·5 to60·5 wt %) and<br />

87 Sr/ 87 Sr (Fig. 1). By comparison, Phase 2 lavas record <strong>the</strong><br />

bulk <strong>of</strong> <strong>the</strong> eruption ( 60%) but only record an 2wt%<br />

increase in SiO 2. Phase 3b lava compositions are rel<strong>at</strong>ively<br />

2212<br />

homogeneous, with most <strong>of</strong> <strong>the</strong> vari<strong>at</strong>ion in Phase 3 occurring<br />

prior to August 1948 (Phase 3a). Phase 3 whole-rock<br />

compositions are characterized by higher SiO 2 concentr<strong>at</strong>ions<br />

and, based on indices <strong>of</strong> crustal contamin<strong>at</strong>ion, generally<br />

appear more contamin<strong>at</strong>ed. Whole-rock Ba/Nb is<br />

52^73 (vs 40^48 in Phase 2 lavas), and K 2O/TiO 2 is<br />

1·2^2·2 (vs 1·2^1·3 in Phase 2), with both r<strong>at</strong>ios increasing<br />

with SiO 2 and eruption d<strong>at</strong>e. The only exception is one <strong>of</strong><br />

<strong>the</strong> last erupted lavas (116289-19; 25 February 1952), which<br />

has slightly lower SiO 2 and lower Ba/Nb (69·3). Melt inclusions<br />

from this last sample are also hosted in more forsteritic<br />

olivine (average Fo 80·1) compared with more fayalitic<br />

olivine hosts (average Fo 78·6) erupted earlier in Phase 3b,<br />

suggesting th<strong>at</strong> <strong>the</strong> last erupted lavas from Paricutin were<br />

more primitive and less crustally contamin<strong>at</strong>ed than <strong>the</strong><br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

previously erupted m<strong>at</strong>erial (Fig. 4). Vari<strong>at</strong>ions in incomp<strong>at</strong>ible<br />

trace element r<strong>at</strong>ios sensitive to crustal contamin<strong>at</strong>ion<br />

(Ba/Nb, K 2O/TiO 2) in Phase 3 lavas and melt<br />

inclusions can generally be modeled by mixing an early<br />

Phase 3 lava (Par-2 <strong>of</strong> Cebria¤ et al., 2011) with a bulk xenolith<br />

composition similar to 116293-5 (Table 2). Using a<br />

bulk xenolith composition, 25^30 wt % crustal addition is<br />

required to explain <strong>the</strong> vari<strong>at</strong>ions in trace elements<br />

(Fig. 11). This assimil<strong>at</strong>ion model, coupled with minor fractional<br />

crystalliz<strong>at</strong>ion, can account for <strong>the</strong> majority <strong>of</strong> both<br />

<strong>the</strong> major and trace element variability in Phase 3 melts.<br />

The limited vari<strong>at</strong>ions in Zr/Nb in Phase 3 magmas are<br />

broadly consistent with assimil<strong>at</strong>ion <strong>of</strong> crust with similar<br />

Zr/Nb, like <strong>the</strong> xenoliths. Sc<strong>at</strong>ter in Zr/Nb values (Figs 11<br />

and 12) is most probably due to minor interlabor<strong>at</strong>ory<br />

biases, but minor mixing with Phase 2 melts could also<br />

play a role.<br />

Over <strong>the</strong> course <strong>of</strong> <strong>the</strong> eruption <strong>of</strong> Phase 3 lavas, Ba/Nb<br />

in melt inclusions increases with decreasing host olivine<br />

forsterite content, indic<strong>at</strong>ing th<strong>at</strong> crystalliz<strong>at</strong>ion <strong>of</strong> more<br />

evolved olivines is capturing more contamin<strong>at</strong>ed magma<br />

(Fig. 13). K 2O/TiO 2 shows little system<strong>at</strong>ic vari<strong>at</strong>ion with<br />

<strong>the</strong> forsterite content <strong>of</strong> olivine from Phase 3 lavas.<br />

Similarly, Cl/K is neg<strong>at</strong>ively correl<strong>at</strong>ed with K2O/TiO2,<br />

once again suggesting th<strong>at</strong> decreasing Cl/K is recording<br />

crustal assimil<strong>at</strong>ion <strong>of</strong> a high-K contaminant (Fig. 8b).<br />

Despite geochemical evidence for contamin<strong>at</strong>ion throughout<br />

<strong>the</strong> eruption <strong>of</strong> Phase 3 lavas, melt inclusions within<br />

single lavas are similar to <strong>the</strong> whole-rock compositions<br />

in terms <strong>of</strong> indices <strong>of</strong> contamin<strong>at</strong>ion and display no<br />

Fig. 13. Olivine host forsterite content vs inclusion Ba/Nb r<strong>at</strong>io for<br />

Type I melt inclusions.<br />

2213<br />

system<strong>at</strong>ic vari<strong>at</strong>ions with ei<strong>the</strong>r host composition or melt<br />

SiO 2 within single samples. K 2O/TiO 2 in olivine-hosted<br />

melt inclusions from <strong>the</strong> two Phase 3b lavas varies from<br />

1·29 0·16 to 2·00 0·32, with corresponding wholerock<br />

r<strong>at</strong>ios varying from 1·38 to 2·00, respectively.<br />

Orthopyroxene-hosted melt inclusions from Phase 3 lavas<br />

have K 2O/TiO 2 r<strong>at</strong>ios equivalent to those <strong>of</strong> <strong>the</strong><br />

olivine-hosted melt inclusions, indic<strong>at</strong>ing a similar timing<br />

<strong>of</strong> crystalliz<strong>at</strong>ion <strong>of</strong> olivine and orthopyroxene rel<strong>at</strong>ive to<br />

assimil<strong>at</strong>ion. Average Ba/Nb r<strong>at</strong>ios in <strong>the</strong> melt inclusions<br />

are generally <strong>of</strong>fset to slightly lower values, but are within<br />

error <strong>of</strong> <strong>the</strong> whole-rock values. These minor differences in<br />

Ba/Nb between whole-rock and average melt inclusions<br />

may be <strong>the</strong> result <strong>of</strong> minor l<strong>at</strong>e-stage contamin<strong>at</strong>ion, following<br />

melt inclusion entrapment in olivine and prior to<br />

form<strong>at</strong>ion <strong>of</strong> <strong>the</strong> Type II inclusions. However, because<br />

Type II melt inclusions (excluding sample 116289-19) all<br />

have Ba/Nb r<strong>at</strong>ios equivalent to Type I inclusions and less<br />

than <strong>the</strong> whole-rock values, <strong>the</strong> slight difference between<br />

<strong>the</strong> melt inclusions and <strong>the</strong> whole-rock r<strong>at</strong>ios is probably<br />

an analytical artifact (reflecting slight calibr<strong>at</strong>ion differences<br />

between SIMS and ICP-MS techniques). Therefore,<br />

despite <strong>the</strong> evidence for significant crustal contamin<strong>at</strong>ion<br />

in <strong>the</strong> evolution <strong>of</strong> Phase 3 magmas, <strong>the</strong> melt inclusions<br />

within single Phase 3 lavas do not preserve a record <strong>of</strong> progressive<br />

contamin<strong>at</strong>ion and fraction<strong>at</strong>ion, implying th<strong>at</strong><br />

any significant contamin<strong>at</strong>ion must have occurred prior<br />

to both Type I olivine and orthopyroxene crystalliz<strong>at</strong>ion.<br />

However, in a given lava sample, variability in melt inclusion<br />

Ba/Nb r<strong>at</strong>ios is outside analytical uncertainty<br />

(Fig. 13); <strong>the</strong>refore, instead <strong>of</strong> a consistent and progressive<br />

contamin<strong>at</strong>ion sign<strong>at</strong>ure, rel<strong>at</strong>ively enriched and variable<br />

K 2O/TiO 2 and Ba/Nb (Fig. 11) r<strong>at</strong>ios in Phase 3 melt inclusions<br />

may imply a rel<strong>at</strong>ively homogeneous ‘bulk’<br />

magma experiencing variable contamin<strong>at</strong>ion decoupled<br />

from significant fraction<strong>at</strong>ion.<br />

Magma chamber models and <strong>the</strong> timing <strong>of</strong><br />

inclusion entrapment and assimil<strong>at</strong>ion<br />

Several physical models have been suggested for magma<br />

chamber evolution and dynamics bene<strong>at</strong>h Paricutin<br />

Volcano as a means <strong>of</strong> explaining <strong>the</strong> observed compositional<br />

vari<strong>at</strong>ions. Wilcox (1954) and McBirney et al.<br />

(1987) both suggested th<strong>at</strong> <strong>the</strong> eruptive history <strong>of</strong> <strong>the</strong> volcano<br />

was too short to have developed <strong>the</strong> observed compositional<br />

vari<strong>at</strong>ions and <strong>the</strong>refore a str<strong>at</strong>ified magma<br />

chamber had to have been present for decades prior to<br />

<strong>the</strong> eruption. In a str<strong>at</strong>ified magma chamber model, density<br />

differences between compositionally distinct layers<br />

halts convection between <strong>the</strong>m, with contamin<strong>at</strong>ed<br />

magma accumul<strong>at</strong>ing <strong>at</strong> <strong>the</strong> top <strong>of</strong> <strong>the</strong> chamber (Wilcox,<br />

1954). However, convection within str<strong>at</strong>ified layers could<br />

still persist. McBirney et al. (1987) presented a similar<br />

model in which <strong>the</strong> mafic intrusion melts <strong>the</strong> wall-rock<br />

and <strong>the</strong>n back-mixes with <strong>the</strong> crustal melt as it rises to<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

collect under <strong>the</strong> ro<strong>of</strong>. This model is also capable <strong>of</strong> gener<strong>at</strong>ing<br />

<strong>the</strong> erupted volumes <strong>of</strong> zoned magma on a decadal<br />

time scale depending on <strong>the</strong> shape and r<strong>at</strong>e <strong>of</strong> cooling <strong>of</strong><br />

<strong>the</strong> intrusion. The model <strong>of</strong> a large zoned magma body as<br />

<strong>the</strong> storage site for <strong>the</strong> Paricutin lavas, however, appears<br />

to be fundamentally flawed in th<strong>at</strong> it is based on <strong>the</strong> assumption<br />

th<strong>at</strong> all <strong>of</strong> <strong>the</strong> erupted magmas are petrogenetically<br />

rel<strong>at</strong>ed by a simple progressive assimil<strong>at</strong>ion and<br />

fractional crystalliz<strong>at</strong>ion process. This is not consistent<br />

with <strong>the</strong> whole-rock trace element d<strong>at</strong>a, in particular incomp<strong>at</strong>ible<br />

trace element r<strong>at</strong>ios, from both this study and<br />

<strong>the</strong> liter<strong>at</strong>ure th<strong>at</strong> indic<strong>at</strong>e th<strong>at</strong> <strong>the</strong> eruption is composed<br />

<strong>of</strong> <strong>at</strong> least two different magma b<strong>at</strong>ches. The distinction is<br />

most easily observed in Zr/Nb r<strong>at</strong>ios (Figs 11 and 12),<br />

which show differences between Phase 1 and Phase 2<br />

lavas and between Phase 2 and Phase 3 lavas th<strong>at</strong> cannot<br />

be explained by crustal assimil<strong>at</strong>ion. Additionally, a<br />

pre-existing zoned magma chamber model does not explain<br />

<strong>the</strong> compositionally distinct Phase 1 lavas, which<br />

record olivine crystalliz<strong>at</strong>ion <strong>at</strong> a significantly deeper<br />

crustal level. The presence <strong>of</strong> short-lived, compositionally<br />

distinct magma b<strong>at</strong>ches <strong>at</strong> <strong>the</strong> onset <strong>of</strong> <strong>the</strong> eruption is<br />

hard to reconcile with a model dependent on a<br />

pre-existing, str<strong>at</strong>ified magma chamber, especially given<br />

<strong>the</strong> lack <strong>of</strong> interaction between Phase 1 and Phase 2, as<br />

indic<strong>at</strong>ed by both whole-rock d<strong>at</strong>a and melt inclusions.<br />

Erlund et al. (2010) also argued against a pre-existing<br />

zoned magma chamber based on <strong>the</strong> concept th<strong>at</strong> some<br />

<strong>of</strong> <strong>the</strong> melt inclusions from <strong>the</strong> early stages <strong>of</strong> <strong>the</strong> eruption<br />

should have shown evidence for contamin<strong>at</strong>ion, given<br />

th<strong>at</strong> most crystalliz<strong>at</strong>ion would occur in contamin<strong>at</strong>ed<br />

<strong>the</strong>rmal boundary layers.<br />

<strong>An</strong> altern<strong>at</strong>ive model argues th<strong>at</strong> <strong>the</strong>re is no<br />

pre-existing magma chamber and th<strong>at</strong> basaltic andesite<br />

magma is injected <strong>into</strong> <strong>the</strong> crust with subsequent rapid<br />

crystalliz<strong>at</strong>ion and assimil<strong>at</strong>ion (Dungan, 2005;<br />

Erlund et al., 2010; Cebria¤ et al., 2011). This model is based<br />

on observ<strong>at</strong>ions th<strong>at</strong> crustal xenoliths can potentially be<br />

assimil<strong>at</strong>ed rapidly (less than decadal time scales) and<br />

th<strong>at</strong> <strong>the</strong>y would not survive in a hot basaltic andesite for<br />

more than 10 years, as might be implied by <strong>the</strong> presence<br />

<strong>of</strong> accumul<strong>at</strong>ed contamin<strong>at</strong>ed magma in a str<strong>at</strong>ified<br />

magma chamber (Wilcox, 1954; McBirney et al., 1987).<br />

This is supported by <strong>the</strong> observ<strong>at</strong>ion <strong>at</strong> Paricutin th<strong>at</strong><br />

crustal xenoliths are found only in Phase 1 and 2 lavas,<br />

and not in <strong>the</strong> high-SiO 2,high- 87 Sr/ 86 Sr Phase 3 lavas. In<br />

<strong>the</strong> Dungan (2005) model, <strong>the</strong> eruption initially pulls m<strong>at</strong>erial<br />

from <strong>the</strong> middle <strong>of</strong> <strong>the</strong> intrusion where crustal fragments<br />

have not yet been fully assimil<strong>at</strong>ed. Magma <strong>at</strong> <strong>the</strong><br />

edges <strong>of</strong> <strong>the</strong> intrusion continues assimil<strong>at</strong>ing crust and is<br />

l<strong>at</strong>er mixed and erupted with <strong>the</strong> more primitive m<strong>at</strong>erial.<br />

In this scenario, <strong>the</strong> presence <strong>of</strong> an early, distinct magma<br />

composition is easier to r<strong>at</strong>ionalize. However, using<br />

mass-balance calcul<strong>at</strong>ions McBirney et al. (1987)<br />

2214<br />

demonstr<strong>at</strong>ed th<strong>at</strong> although an average xenolith composition<br />

is an appropri<strong>at</strong>e contaminant for most <strong>of</strong> <strong>the</strong> trace<br />

elements, none <strong>of</strong> <strong>the</strong> crustal xenoliths or basement samples<br />

are appropri<strong>at</strong>e for explaining <strong>the</strong> Sr isotopic variability<br />

between Phases 2 and 3, which requires a<br />

contaminant with a more radiogenic Sr isotope composition<br />

and gre<strong>at</strong>er Sr abundance. This requires a crustal<br />

component not found in ei<strong>the</strong>r <strong>the</strong> exposed basement or<br />

<strong>the</strong> entrained crustal xenoliths: this is difficult to explain<br />

if <strong>the</strong> magma is rapidly evolving by assimil<strong>at</strong>ion <strong>of</strong> its<br />

local surroundings as suggested by Dungan (2005).<br />

Additionally, assuming th<strong>at</strong> <strong>the</strong> melt inclusions provide a<br />

represent<strong>at</strong>ive sampling <strong>of</strong> <strong>the</strong> evolving magma, <strong>the</strong>re is<br />

no melt inclusion evidence for this rapid, progressive<br />

magma evolution. In <strong>the</strong> Dungan (2005) model, one <strong>of</strong><br />

<strong>the</strong> key requirements is th<strong>at</strong> <strong>the</strong> same basaltic andesite is<br />

essentially present throughout <strong>the</strong> eruption, with erupted<br />

magma compositions varying as a result <strong>of</strong> rapid, l<strong>at</strong>e assimil<strong>at</strong>ion<br />

<strong>of</strong> crustal xenoliths entrained in <strong>the</strong> magma. If<br />

this model is correct, we might expect to find melt inclusion<br />

compositions in <strong>the</strong> Phase 3 lavas th<strong>at</strong> record a mixture<br />

<strong>of</strong> both contamin<strong>at</strong>ed and uncontamin<strong>at</strong>ed (Phase 1<br />

or 2) magma b<strong>at</strong>ches. Although Type I melt inclusions in<br />

<strong>the</strong> first sample in Phase 3 do record a wide range <strong>of</strong> Ba/<br />

Nb r<strong>at</strong>ios <strong>the</strong> inclusions predominantly tend to record discrete<br />

compositions more characteristic <strong>of</strong> crystalliz<strong>at</strong>ion<br />

within an already contamin<strong>at</strong>ed magma body. Erlund<br />

et al. (2010) presented a model similar to th<strong>at</strong> <strong>of</strong> Dungan<br />

(2005), in which during high mass flux eruption r<strong>at</strong>es<br />

deeper magmas (as indic<strong>at</strong>ed by measurable CO2 concentr<strong>at</strong>ions)<br />

fed <strong>the</strong> eruption and caused sill form<strong>at</strong>ion.<br />

Magma stored <strong>at</strong> shallow depths in dikes and sills would<br />

<strong>the</strong>n erupt subsequently as mass flux r<strong>at</strong>es from depth<br />

decreased. This model is <strong>at</strong>tractive in th<strong>at</strong> it provides an<br />

explan<strong>at</strong>ion for <strong>the</strong> initial compositionally distinct Phase<br />

1 magmas.<br />

Rel<strong>at</strong>ive timing <strong>of</strong> AFC processes and<br />

melt inclusion entrapment<br />

Johnson et al. (2008) suggested for Volcan Jorullo (Mexico)<br />

th<strong>at</strong> whereas melt inclusions record crystalliz<strong>at</strong>ion <strong>of</strong> olivine<br />

during ascent and degassing <strong>of</strong> <strong>the</strong> magma, <strong>the</strong> compositional<br />

trends defined by lavas are driven by a deeper<br />

fraction<strong>at</strong>ion th<strong>at</strong> is not recorded in <strong>the</strong> erupted phenocrysts<br />

and inclusions. Essentially <strong>the</strong> crystal^melt inclusion<br />

record is missing <strong>the</strong> deeper processes. A similar model<br />

may be applicable to Paricutin and would be consistent<br />

with <strong>the</strong> generally low crystalliz<strong>at</strong>ion pressures <strong>of</strong> Phase 2<br />

and 3 lavas estim<strong>at</strong>ed from w<strong>at</strong>er concentr<strong>at</strong>ions (and no<br />

detectable CO 2) in melt inclusions (Luhr, 2001). This<br />

would provide a mechanism to explain whyType I melt inclusions<br />

have a restricted compositional range similar to<br />

<strong>the</strong> whole-rock composition for single samples ra<strong>the</strong>r than<br />

a range <strong>of</strong> compositions from primitive to more evolved as<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

would be expected as a consequence <strong>of</strong> a progressive AFC<br />

process.<br />

<strong>An</strong> altern<strong>at</strong>ive interpret<strong>at</strong>ion <strong>of</strong> <strong>the</strong> inclusion d<strong>at</strong>a is th<strong>at</strong><br />

assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion were decoupled.<br />

One <strong>of</strong> <strong>the</strong> fundamental underpinnings <strong>of</strong> <strong>the</strong> coupled<br />

assimil<strong>at</strong>ion^fractional crystalliz<strong>at</strong>ion (AFC) model (e.g.<br />

DePaolo, 1981) is th<strong>at</strong> <strong>the</strong> l<strong>at</strong>ent he<strong>at</strong> <strong>of</strong> crystalliz<strong>at</strong>ion<br />

during fractional crystalliz<strong>at</strong>ion provides <strong>the</strong> he<strong>at</strong> source<br />

for crustal assimil<strong>at</strong>ion, such th<strong>at</strong> magmas evolve through<br />

progressive and concurrent assimil<strong>at</strong>ion and fractional<br />

crystalliz<strong>at</strong>ion as observed in olivine-hosted melt inclusions<br />

by Kent et al. (2002) in Yemen flood basalts. If this model<br />

is correct, melt inclusions and phenocrysts in crustally contamin<strong>at</strong>ed<br />

magmas should record a progression from a<br />

primitive uncontamin<strong>at</strong>ed magma toward a contamin<strong>at</strong>ed<br />

more evolved magma. However, this progressive contamin<strong>at</strong>ion<br />

is not recorded by melt inclusions <strong>at</strong> Paricutin.<br />

Instead, ParicutinType I melt inclusions record a rel<strong>at</strong>ively<br />

restricted compositional range in each <strong>of</strong> <strong>the</strong> sampled<br />

lavas (Fig. 4; Table 5). This is most evident when comparing<br />

inclusions between eruptive phases. In a progressive AFC<br />

model, inclusions in Phase 3 lavas might be expected to<br />

record a range <strong>of</strong> compositions, from more primitive<br />

(Phase 2) to more evolved and contamin<strong>at</strong>ed. However,<br />

Type I inclusions (<strong>the</strong> most primitive) in Phase 3 lavas<br />

appear to be more contamin<strong>at</strong>ed rel<strong>at</strong>ive to Phase 2 inclusions<br />

and record a restricted compositional range with<br />

only minor overlap with Phase 2 compositions <strong>at</strong> <strong>the</strong><br />

onset <strong>of</strong> Phase 3 (Figs 8 and 11). This may suggest th<strong>at</strong> contamin<strong>at</strong>ion<br />

in <strong>the</strong> Paricutin magm<strong>at</strong>ic system is decoupled<br />

from crystalliz<strong>at</strong>ion, or <strong>at</strong> least from <strong>the</strong> crystalliz<strong>at</strong>ion recorded<br />

by melt inclusions in <strong>the</strong> lavas. A similar interpret<strong>at</strong>ion<br />

has previously been suggested by Grove et al. (1988)<br />

to explain <strong>the</strong> gener<strong>at</strong>ion <strong>of</strong> an andesite lava <strong>at</strong> Medicine<br />

Lake, California. In this example, <strong>the</strong> arguments for<br />

decoupling <strong>of</strong> crystalliz<strong>at</strong>ion and assimil<strong>at</strong>ion are based<br />

on petrological evidence ra<strong>the</strong>r than a direct record <strong>of</strong><br />

magma compositions, and <strong>the</strong> observ<strong>at</strong>ion th<strong>at</strong> in silic<strong>at</strong>e<br />

melts, diffusion <strong>of</strong> he<strong>at</strong> is substantially faster than diffusion<br />

<strong>of</strong> chemical constituents.<br />

One mechanism to explain <strong>the</strong> apparent decoupling <strong>of</strong><br />

assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion in <strong>the</strong> melt inclusion<br />

record <strong>at</strong> Paricutin is th<strong>at</strong> <strong>the</strong> <strong>the</strong>rmal conditions<br />

required for inclusion entrapment were not favorable to<br />

coupled crustal assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion.<br />

In this case <strong>the</strong> melt inclusions represent a biased record<br />

<strong>of</strong> magma evolution with much <strong>of</strong> <strong>the</strong> crystalliz<strong>at</strong>ion and<br />

assimil<strong>at</strong>ion taking place within a main magma chamber<br />

or body essentially missed by <strong>the</strong> inclusion record.<br />

Petrographic investig<strong>at</strong>ions and experimental studies <strong>of</strong><br />

melt inclusion form<strong>at</strong>ion have identified several ways <strong>of</strong><br />

forming melt inclusions, with iso<strong>the</strong>rmal crystalliz<strong>at</strong>ion<br />

following rapid cooling (e.g. Roedder, 1979, 1984; Kohut &<br />

Nielsen, 2004) and entrapment in crystal defects or<br />

2215<br />

disloc<strong>at</strong>ions during slower cooling (e.g. Faure & Schiano,<br />

2005), which is probably <strong>the</strong> most realistic for basaltic compositions<br />

(Kent, 2008). Similarly, Goldstein & Luth (2006)<br />

found th<strong>at</strong> inclusions formed <strong>at</strong> a range <strong>of</strong> cooling r<strong>at</strong>es<br />

(7^2508C h 1 ) but not <strong>at</strong> very low cooling r<strong>at</strong>es (18C h 1 ;<br />

Kent, 2008). The potential need for undercooling to form<br />

melt inclusions may preclude significant assimil<strong>at</strong>ion and<br />

may be rel<strong>at</strong>ed to magma chamber processes, particularly<br />

where in <strong>the</strong> magma chamber <strong>the</strong> <strong>the</strong>rmal conditions are<br />

favorable for inclusion form<strong>at</strong>ion and entrapment rel<strong>at</strong>ive<br />

to wallrock assimil<strong>at</strong>ion. If this is <strong>the</strong> case, melt inclusions<br />

in short-lived, small-volume magm<strong>at</strong>ic systems might provide<br />

a biased sampling <strong>of</strong> magma chamber processes and<br />

melt evolution.<br />

Multiple magma b<strong>at</strong>ches and magma<br />

mixing<br />

The preferred explan<strong>at</strong>ion for <strong>the</strong> overall evolution <strong>of</strong> <strong>the</strong><br />

Paricutin magma system is <strong>the</strong> presence <strong>of</strong> several compositionally<br />

distinct magma b<strong>at</strong>ches <strong>at</strong> shallow levels bene<strong>at</strong>h<br />

<strong>the</strong> erupting volcano. Abrupt temporal changes in r<strong>at</strong>ios<br />

<strong>of</strong> highly incomp<strong>at</strong>ible trace element r<strong>at</strong>ios such as Zr/Nb<br />

th<strong>at</strong> are difficult to explain using a progressive assimil<strong>at</strong>ion<br />

and fractional crystalliz<strong>at</strong>ion model mark <strong>the</strong> arrival <strong>of</strong><br />

distict magma b<strong>at</strong>ches. The Phase 1 and Phase 2 magmas<br />

have similar 87 Sr/ 86 Sr but are compositionally different in<br />

terms <strong>of</strong> Zr/Y r<strong>at</strong>ios ( 6 vs 7^8) and LREE^HREE fraction<strong>at</strong>ion<br />

(La/Yb N:5·3^5·7 vs6·6^7·2), and <strong>the</strong>se fe<strong>at</strong>ures<br />

can be explained by small differences in <strong>the</strong> degree <strong>of</strong> melting<br />

to produce distinct magma b<strong>at</strong>ches. The transition<br />

from Phase 2 to Phase 3 records <strong>the</strong> shift and potential<br />

minor mixing between two independently evolving but<br />

interconnected magma bodies. This model was also<br />

favored by Luhr and Housh based on mineral phase equilibria<br />

<strong>of</strong> <strong>the</strong> Phase 2 and Phase 3 lavas (T. Housh, personal<br />

communic<strong>at</strong>ion, 2007). The Phase 3 magmas represent<br />

only a small proportion <strong>of</strong> <strong>the</strong> total erupted volume, and<br />

<strong>the</strong>y record a wide compositional range th<strong>at</strong> is consistent<br />

with crustal assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion.<br />

The rel<strong>at</strong>ionship, if any, <strong>of</strong> <strong>the</strong> Phase 3 magmas to <strong>the</strong> two<br />

preceding magma b<strong>at</strong>ches is uncertain. Even <strong>the</strong> least contamin<strong>at</strong>ed<br />

Phase 3 sample has higher 87 Sr/ 86 Sr and K 2O/<br />

TiO 2 than any <strong>of</strong> <strong>the</strong> earlier Phase 1 and 2 magmas, indic<strong>at</strong>ive<br />

<strong>of</strong> some crustal influence. The difference in Zr/Nb<br />

<strong>at</strong> similar MgO makes it difficult to rel<strong>at</strong>e <strong>the</strong> Phase 3<br />

and Phase 2 magmas. On some trace element plots (e.g.<br />

Fig 11), <strong>the</strong> composition <strong>of</strong> Phase 3 lavas could potentially<br />

be explained by assimil<strong>at</strong>ion <strong>of</strong> an average xenolith composition<br />

by Phase 1 lavas, although this is not <strong>the</strong> case<br />

with o<strong>the</strong>r trace element r<strong>at</strong>ios (e.g. Fig. 14). Work is currently<br />

under way to assess <strong>the</strong> rel<strong>at</strong>ionships between <strong>the</strong>se<br />

different magma b<strong>at</strong>ches using high-precision Pb isotope<br />

analyses.<br />

The evolution <strong>of</strong> Paricutin volcano appears to have<br />

been controlled by <strong>the</strong> eruption <strong>of</strong> three<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. 14. (a) Nb/Yb vs Th/Yb diagram, after Pearce & Pe<strong>at</strong>e (1995), demonstr<strong>at</strong>ing <strong>the</strong> variability <strong>of</strong> magm<strong>at</strong>ic compositions <strong>at</strong> Paricutin compared<br />

with those <strong>of</strong> <strong>the</strong> MGVF. <strong>An</strong>alyses <strong>of</strong> crustal m<strong>at</strong>erial have been placed <strong>into</strong> one <strong>of</strong> four c<strong>at</strong>egories with <strong>the</strong> average composition for<br />

each c<strong>at</strong>egory plotted: (1) xenolith glass; (2) basement; (3) high 87 Sr/ 86 Sr xenoliths; (4) low 87 Sr/ 86 Sr xenoliths. Compiled MGVF d<strong>at</strong>a sources:<br />

Luhr & Carmichael (1985); Hasenaka & Carmichael (1987); Luhr (2001); Chesley et al. (2002); Righter et al. (2002); Verma & Hasenaka (2004);<br />

Johnson et al. (2009); Cebria¤ et al. (2011).<br />

compositionally distinctive and independently evolving<br />

magma b<strong>at</strong>ches. The existence <strong>of</strong> compositionally distinct<br />

magma b<strong>at</strong>ches th<strong>at</strong> are erupted <strong>at</strong> different times during<br />

a monogenetic eruption is not unique to Paricutin.<br />

Numerous studies <strong>of</strong> small-volume basaltic systems have<br />

shown <strong>the</strong> eruption <strong>of</strong> compositionally distinct magma<br />

b<strong>at</strong>ches over rel<strong>at</strong>ively short time periods (e.g. Camp<br />

et al., 1987; Reagan & Gill, 1989; Cervantes & Wallace,<br />

2003; Strong & Wolff, 2003). Reagan & Gill (1989)<br />

observed Nb-enriched and Nb-depleted magmas from <strong>the</strong><br />

same eruptive episode <strong>at</strong> Turrialba Volcano, Costa Rica,<br />

and argued for <strong>the</strong> tapping and flux melting <strong>of</strong> compositionally<br />

distinct mantle source domains. Similarly, Strong<br />

& Wolff (2003) noted <strong>the</strong> contemporaneous eruption <strong>of</strong><br />

calc-alkaline and ocean island basalt (OIB)-like magmas<br />

in sou<strong>the</strong>rn Cascades monogenetic centers. They argued<br />

th<strong>at</strong> <strong>the</strong> different melts required distinct magma sources,<br />

and th<strong>at</strong> ei<strong>the</strong>r <strong>the</strong> melts travel through <strong>the</strong> crust without<br />

storage within a magma chamber or th<strong>at</strong> <strong>the</strong> magma<br />

chambers are inefficient <strong>at</strong> mixing <strong>the</strong>se compositionally<br />

distinct magma b<strong>at</strong>ches.<br />

There is good evidence for <strong>the</strong> eruption <strong>of</strong> compositionally<br />

diverse magmas in close proximity in <strong>the</strong><br />

2216<br />

Michoacan^Guanaju<strong>at</strong>o volcanic field (MGVF) where<br />

Paricutin is loc<strong>at</strong>ed (Luhr & Carmichael, 1985; Verma &<br />

Hasenaka, 2004; Johnson et al., 2009). Luhr & Carmichael<br />

(1985) showed th<strong>at</strong> magmas <strong>of</strong> considerably different composition<br />

were erupted only 3 km apart, albeit <strong>at</strong> different<br />

times, <strong>at</strong> Volcan Jorullo (K2O/TiO2 1·0; La/Sm 3^4) and<br />

Cerro la Pilita (K 2O/TiO 2<br />

2·1; La/Sm 10^12). Johnson<br />

et al. (2009) noted th<strong>at</strong> <strong>the</strong>re is no system<strong>at</strong>ic across-arc<br />

vari<strong>at</strong>ion in magma composition within <strong>the</strong> MGVF.<br />

Figure 14 shows how <strong>the</strong> compositions <strong>of</strong> <strong>the</strong> Phase 1, 2<br />

and 3 magmas <strong>at</strong> Paricutin compare with compositional<br />

diversity found within <strong>the</strong> MGVF, using a plot <strong>of</strong> Nb/Yb<br />

vs Th/Yb (after Pearce & Pe<strong>at</strong>e, 1995).<br />

CONCLUSIONS<br />

The two distinct melt inclusion popul<strong>at</strong>ions preserved in<br />

single lava samples provide new insights <strong>into</strong> <strong>the</strong> history <strong>of</strong><br />

magma evolution and <strong>the</strong> rel<strong>at</strong>ive timing <strong>of</strong> crystalliz<strong>at</strong>ion<br />

and assimil<strong>at</strong>ion <strong>at</strong> Paricutin. The Type II (high-SiO 2)<br />

melt inclusions record a very l<strong>at</strong>e-stage crystalliz<strong>at</strong>ion<br />

event with little assimil<strong>at</strong>ion and are compositionally similar<br />

to tephra m<strong>at</strong>rix glasses reported by Luhr (2001) and<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

Erlund et al. (2010). Because this popul<strong>at</strong>ion has low vol<strong>at</strong>ile<br />

contents and is not observed in contemporaneous<br />

tephra samples, we suggest th<strong>at</strong> <strong>the</strong>se high-SiO 2, low-S inclusions<br />

record very shallow crystalliz<strong>at</strong>ion after segreg<strong>at</strong>ion<br />

<strong>of</strong> gases from denser magma near <strong>the</strong> base <strong>of</strong> <strong>the</strong><br />

edifice.<br />

The low-SiO 2, high-S popul<strong>at</strong>ion (Type I) <strong>of</strong> inclusions<br />

records a rel<strong>at</strong>ively restricted compositional range, more<br />

similar to <strong>the</strong> whole-rock compositions. If <strong>the</strong> melt inclusion<br />

compositions record a represent<strong>at</strong>ive sampling <strong>of</strong><br />

magma chamber processes, none <strong>of</strong> <strong>the</strong> previous models<br />

for magma chamber development and melt evolution <strong>at</strong><br />

Paricutin are appropri<strong>at</strong>e. The similarity <strong>of</strong> Type I melt inclusions<br />

to single whole-rock compositions may instead<br />

imply th<strong>at</strong> crystalliz<strong>at</strong>ion, or <strong>at</strong> least <strong>the</strong> crystalliz<strong>at</strong>ion recorded<br />

by <strong>the</strong> erupted crystal cargo, must have occurred<br />

after significant crustal contamin<strong>at</strong>ion. The potential<br />

decoupling <strong>of</strong> assimil<strong>at</strong>ion and fractional crystalliz<strong>at</strong>ion<br />

may also be a function <strong>of</strong> <strong>the</strong> mechanism for inclusion entrapment<br />

in th<strong>at</strong> inclusions may be forming only <strong>at</strong> specific<br />

intervals in <strong>the</strong> magm<strong>at</strong>ic evolution when <strong>the</strong>rmal conditions<br />

(i.e. degree <strong>of</strong> undercooling) are optimal for inclusion<br />

entrapment.<br />

In conjunction with <strong>the</strong> melt inclusion d<strong>at</strong>a, new<br />

whole-rock trace element analyses indic<strong>at</strong>e th<strong>at</strong> <strong>the</strong> abrupt<br />

compositional vari<strong>at</strong>ions in incomp<strong>at</strong>ible trace element<br />

r<strong>at</strong>ios observed in erupted lavas <strong>at</strong> Paricutin are probably<br />

<strong>the</strong> result <strong>of</strong> multiple, independently evolving, small<br />

magma b<strong>at</strong>ches ra<strong>the</strong>r than <strong>the</strong> progressive assimil<strong>at</strong>ion<br />

and fractional crystalliz<strong>at</strong>ion <strong>of</strong> a single b<strong>at</strong>ch <strong>of</strong> basaltic<br />

andesite magma. Compositionally diverse magmas erupting<br />

in close proximity to one ano<strong>the</strong>r are observed elsewhere<br />

locally in <strong>the</strong> Michoaca¤n^Guanaju<strong>at</strong>o volcanic<br />

field and have been <strong>at</strong>tributed to mantle heterogeneity.<br />

Therefore, although fractional crystalliz<strong>at</strong>ion and crustal<br />

assimil<strong>at</strong>ion may be important processes within single<br />

eruptive phases, <strong>the</strong> complexity and timing <strong>of</strong> compositional<br />

shifts in <strong>the</strong> erupted magma <strong>at</strong> Paricutin are instead<br />

a function <strong>of</strong> <strong>the</strong> co-evolution <strong>of</strong> multiple, compositionally<br />

distinct magma b<strong>at</strong>ches derived by variable melt<br />

gener<strong>at</strong>ion processes within a heterogeneous mantle<br />

source region, with additional crustal assimil<strong>at</strong>ion<br />

effects superimposed prior to <strong>the</strong>ir input to shallow crustal<br />

levels.<br />

ACKNOWLEDGEMENTS<br />

The authors would like to thank Frank Tepley (Oregon<br />

St<strong>at</strong>e University EMPA labor<strong>at</strong>ory) and Rick Hervig and<br />

Linda Williams (Arizona St<strong>at</strong>e University SIMS facility)<br />

for <strong>the</strong>ir assistance. We thank Paul Wallace, K<strong>at</strong>e<br />

Saunders, and an anonymous reviewer for feedback and<br />

comments. Whole-rock samples and thin sections for this<br />

study were provided on loan from <strong>the</strong> Smithsonian<br />

N<strong>at</strong>ional Museum <strong>of</strong> N<strong>at</strong>ural History.<br />

2217<br />

FUNDING<br />

Funding for this project came from N<strong>at</strong>ional Science<br />

Found<strong>at</strong>ion, Division <strong>of</strong> Earth Sciences, grant 0609652.<br />

The SIMS facility <strong>at</strong> ASU is partly supported by a grant<br />

from <strong>the</strong> Instrument<strong>at</strong>ion and Facilities Program,<br />

Division <strong>of</strong> Earth Sciences, N<strong>at</strong>ional Science Found<strong>at</strong>ion.<br />

SUPPLEMENTARY DATA<br />

Supplementary d<strong>at</strong>a for this paper are available <strong>at</strong> Journal<br />

<strong>of</strong> Petrology online.<br />

REFERENCES<br />

Ariskin, A. A., Frenkel, M. Y., Barmina, G. S. & Nielsen, R. L. (1993).<br />

COMAGMAT; a Fortran program to model magm a differenti<strong>at</strong>ion<br />

processes. Computers and Geosciences 19, 1155^1170.<br />

Asimow, P. D. & Ghiorso, M. S. (1998). Algorithmic modific<strong>at</strong>ions extending<br />

MELTS to calcul<strong>at</strong>e subsolidus phase rel<strong>at</strong>ions. American<br />

Mineralogist 83, 1127^1131.<br />

Bannister, V., Roeder, P. & Poustovetov, A. (1998). Chromite in <strong>the</strong><br />

Paricutin lava flows (1943^1952). Journal <strong>of</strong> Volcanology and<br />

Geo<strong>the</strong>rmal Research 87, 151^171.<br />

Bowen, N. L. (1928). The Evolution <strong>of</strong> <strong>the</strong> Igneous Rocks. Princeton,NJ:<br />

Princeton University Press.<br />

Camp, V. E., Hooper, P. R., Roobol, M. J. & White, D. L. (1987). The<br />

Madinah eruption, Saudi Arabia: Magma mixing and simultaneous<br />

extrusion <strong>of</strong> three basaltic chemical types. Bulletin <strong>of</strong><br />

Volcanology 49, 489^505.<br />

Cebria¤, J. M., Martiny, B. M., Lo¤pez-Ruiz,J.&Mora¤n-Zenteno, D. J.<br />

(2011). The Paricutin calc-alkaline lavas: new geochemical and<br />

petrogenetic modelling constraints on <strong>the</strong> crustal assimil<strong>at</strong>ion process.<br />

Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal Research 201, 113^125.<br />

Cervantes, P. & Wallace, P. (2003). Magma degassing and basaltic<br />

eruption styles; a case study <strong>of</strong> approxim<strong>at</strong>ely 2000 year BP Xitle<br />

Volcano in central Mexico. Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal<br />

Research 120, 249^270.<br />

Chesley, J., Ruiz, J., Righter, K., Ferrari, L. & Gomez-Tuena, A.<br />

(2002). Source contamin<strong>at</strong>ion versus assimil<strong>at</strong>ion: and example<br />

from <strong>the</strong> Trans-Mexican volcanic arc. Earth and Planetary Science<br />

Letters 195, 211^221.<br />

Danyushevsky, L. V., Della-Pasqua, F. N. & Sokolov, S. (2000).<br />

Re-equilibr<strong>at</strong>ion <strong>of</strong> melt inclusions trapped by magnesian olivine<br />

phenocrysts from subduction-rel<strong>at</strong>ed magmas: petrological implic<strong>at</strong>ions.<br />

Contributions to Mineralogy and Petrology 138,68^83.<br />

Danyushevsky, L.V., Sokolov, S. & Fallon, T. J. (2002). Melt inclusions<br />

in olivine phenocrysts: Using diffusive re-equilibr<strong>at</strong>ion to determine<br />

<strong>the</strong> cooling history <strong>of</strong> a crystal, with implic<strong>at</strong>ions for <strong>the</strong><br />

origin <strong>of</strong> olivine-phyric volcanic rocks. Journal <strong>of</strong> Petrology 43,<br />

1651^1671.<br />

Davidson, J. P. & Wilson, I. R. (1989). Evolution <strong>of</strong> an alkali basalt^<br />

trachyte suite from Jebel Marra volcano, Sudan, through assimil<strong>at</strong>ion<br />

and fractional crystalliz<strong>at</strong>ion. Earth and Planetary Science Letters<br />

95, 141^160.<br />

DePaolo, D. J. (1981). Trace element and isotopic effects <strong>of</strong> combined<br />

wallrock assimil<strong>at</strong>ion and fractional crystallis<strong>at</strong>ion. Earth and<br />

Planetary Science Letters 53, 189^202.<br />

Devine, J., Gardner, J., Brack, H., Layne, G. & Ru<strong>the</strong>rford, M. (1995).<br />

Comparison <strong>of</strong> microanalytical techniques for estim<strong>at</strong>ing H2O<br />

contents <strong>of</strong> silicic volcanic glasses. American Mineralogist 80, 319^328.<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Dungan, M. A. (2005). Partial melting <strong>at</strong> <strong>the</strong> Earth’s surface: implic<strong>at</strong>ions<br />

for assimil<strong>at</strong>ion r<strong>at</strong>es and mechanisms in subvolcanic intrusions.<br />

Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal Research 140, 193^203.<br />

Eggler, D. H. (1972). W<strong>at</strong>er-s<strong>at</strong>ur<strong>at</strong>ed and unders<strong>at</strong>ur<strong>at</strong>ed melting rel<strong>at</strong>ions<br />

in a Paricutin andesite and an estim<strong>at</strong>e <strong>of</strong> w<strong>at</strong>er content in<br />

<strong>the</strong> n<strong>at</strong>ural magma. Contributions to Mineralogy and Petrology 34,<br />

261^271.<br />

Erlund, E. J., Cashman, K. V., Wallace, P. J., Pioli, L., Rosi, M.,<br />

Johnson, E. & Delgado Granados, H. (2010). Compositional evolution<br />

<strong>of</strong> magma from Paricutin Volcano, Mexico: <strong>the</strong> tephra record.<br />

Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal Research 197,167^187.<br />

Faure, F. & Schiano, P. (2005). Experimental investig<strong>at</strong>ion <strong>of</strong> equilibr<strong>at</strong>ion<br />

conditions during forsterite growth and melt inclusion form<strong>at</strong>ion.<br />

Earth and Planetary Science Letters 236, 882^898.<br />

Foshag, W. F. & Gonzalez-Reyna, J. (1956). Birth and development <strong>of</strong><br />

Paricutin volcano, Mexico. US Geological Survey Bulletin 135,<br />

355^489.<br />

Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer in magm<strong>at</strong>ic<br />

processes IV. A revised and internally consistent <strong>the</strong>rmodynamic<br />

model for <strong>the</strong> interpol<strong>at</strong>ion and extrapol<strong>at</strong>ion <strong>of</strong> liquid^<br />

solid equilibria in magm<strong>at</strong>ic systems <strong>at</strong> elev<strong>at</strong>ed temper<strong>at</strong>ures and<br />

pressures. Contributions to Mineralogy and Petrology 119, 197^212.<br />

Goldstein, S. B. & Luth, R. W. (2006). The importance <strong>of</strong> cooling<br />

regime in <strong>the</strong> form<strong>at</strong>ion <strong>of</strong> melt inclusions in olivine crystals in haplobasaltic<br />

melts. Canadian Mineralogist 44, 1543^1555.<br />

Grove, T. L., Kinzler, R. J., Baker, M. B., Donnelly-Nolan, J. M. &<br />

Lesher, C. E. (1988). Assimil<strong>at</strong>ion <strong>of</strong> granite by basaltic magma <strong>at</strong><br />

Burnt Lava flow, Medicine Lake volcano, nor<strong>the</strong>rn California:<br />

Decoupling <strong>of</strong> he<strong>at</strong> and mass transfer. Contributions to Mineralogy<br />

and Petrology 99, 320^343.<br />

Hasenaka, T. (1994). Size, distribution, and magma output r<strong>at</strong>e for<br />

shield volcanoes <strong>of</strong> <strong>the</strong> Michoaca¤n^Guanajuanto volcanic field,<br />

Central Mexico. Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal Research 63,<br />

13^31.<br />

Hasenaka, T. & Carmichael, I. S. E. (1987). The cinder cones <strong>of</strong><br />

Michoacan^Guanaju<strong>at</strong>o, central Mexico: petrology and chemistry.<br />

Journal <strong>of</strong> Petrology 28, 241^269.<br />

Hauri, E. H. (2002). SIMS analysis <strong>of</strong> vol<strong>at</strong>iles in silic<strong>at</strong>e glasses, 2: isotopes<br />

and abundances in Hawaiian melt inclusions. Chemical<br />

Geology 183,115^141.<br />

Johnson, E. R., Wallace, P. J., Cashman, K.V., Delgado Granados, H.<br />

& Kent, A. J. R. (2008). <strong>Magm<strong>at</strong>ic</strong> vol<strong>at</strong>ile contents and<br />

degassing-induced crystalliz<strong>at</strong>ion <strong>at</strong> Volcan Jorullo, Mexico:<br />

Implic<strong>at</strong>ions for melt evolution and <strong>the</strong> plumbing systems <strong>of</strong> monogenetic<br />

volcanoes. Earth and Planetary Science Letters 269, 477^486.<br />

Johnson, E. R., Wallace, P. J., Delgado Granados, H., Manea, V. C.,<br />

Kent, A. J. R., Bindeman, I. N. & Donegan, C. S. (2009).<br />

Subduction-rel<strong>at</strong>ed vol<strong>at</strong>ile recycling and magma gener<strong>at</strong>ion bene<strong>at</strong>h<br />

Central Mexico: insights from melt inclusions, oxygen isotopes<br />

and geodynamic models. Journal <strong>of</strong> Petrology 50, 1729^1764.<br />

Kaneko, K. & Koyaguchi,T. (2004). Experimental study on <strong>the</strong> effects<br />

<strong>of</strong> crustal temper<strong>at</strong>ure and composition on assimil<strong>at</strong>ion with fractional<br />

crystalliz<strong>at</strong>ion <strong>at</strong> <strong>the</strong> floor <strong>of</strong> magma chambers. Journal <strong>of</strong><br />

Volcanology and Geo<strong>the</strong>rmal Research 129,155^172.<br />

Kent, A. J. R. (2008). Melt inclusions in basaltic and rel<strong>at</strong>ed volcanic<br />

rocks. In: Putirka, K. D. & Tepley, F. J., III (eds) Minerals,<br />

Inclusions and Volcanic Processes. Mineralogical Society <strong>of</strong> America and<br />

Geochemical Society, Reviews in Mineralogy and Petrology 69, 273^331.<br />

Kent, A. J. R., Norman, M. D., Hutcheon, I. D. & Stolper, E. M.<br />

(1999). Assimil<strong>at</strong>ion <strong>of</strong> seaw<strong>at</strong>er-derived components in an oceanic<br />

volcano: Evidence from m<strong>at</strong>rix glasses and glass inclusions from<br />

Loihi Seamount, Hawaii. Chemical Geology 156, 299^319.<br />

2218<br />

Kent, A. J. R., Baker, J. A. & Wiedenbeck, M. (2002). Contamin<strong>at</strong>ion<br />

and melt aggreg<strong>at</strong>ion processes in continental flood basalts: constraints<br />

from melt inclusions in Oligocene basalts from Yemen.<br />

Earth and Planetary Science Letters 202, 577^594.<br />

Knaack, C., Cornelius, S. B. & Hooper, P. R. (1994). Trace element<br />

analyses <strong>of</strong> rocks and minerals by ICP-MS. Technical Notes,<br />

Geo<strong>An</strong>alytical Lab, Washington St<strong>at</strong>e University.<br />

Kohut, E. & Nielsen, R. L. (2004). Melt inclusion form<strong>at</strong>ion mechanisms<br />

and compositional effects in high-<strong>An</strong> feldspar and high-Fo<br />

olivine in anhydrous mafic silic<strong>at</strong>e liquids. Contributions to<br />

Mineralogy and Petrology 147, 684^704.<br />

Krauskopf, K. B. (1948). Mechanism <strong>of</strong> eruption <strong>at</strong> Paricutin Volcano,<br />

Mexico. Geological Society <strong>of</strong> America Bulletin 59,711^731.<br />

Kuritani, T., Kitagawa, H. & Nakamura, E. (2005). Assimil<strong>at</strong>ion and<br />

fractional crystalliz<strong>at</strong>ion controlled by transport process <strong>of</strong> crustal<br />

melt: implic<strong>at</strong>ions from an alkali basalt^dacite suite from Rishiri<br />

Volcano, Japan. Journal <strong>of</strong> Petrology 46, 1421^1442.<br />

Kuritani, T., Yokoyama, T. & Nakamura, E. (2007). R<strong>at</strong>es <strong>of</strong> <strong>the</strong>rmal<br />

and chemical evolution <strong>of</strong> magmas in a cooling magma chamber:<br />

a chronological and <strong>the</strong>oretical study on basaltic and andesitic<br />

lavas from Rishiri Volcano, Japan. Journal <strong>of</strong> Petrology 48, 1295^1319.<br />

Leitch, A. M. (2004). <strong>An</strong>alog experiments on melting and contamin<strong>at</strong>ion<br />

<strong>at</strong> <strong>the</strong> ro<strong>of</strong> and walls <strong>of</strong> magma chambers. Journal <strong>of</strong><br />

Volcanology and Geo<strong>the</strong>rmal Research 129, 173^197.<br />

Luhr, J. (2001). Glass inclusions and melt vol<strong>at</strong>ile contents <strong>at</strong> Paricutin<br />

Volcano, Mexico. Contributions to Mineralogy and Petrology 142,<br />

261^283.<br />

Luhr, J. F. & Carmichael, I. S. E. (1985). Jorullo Volcano, Michoacan,<br />

Mexico (1759^1774): The earliest stages <strong>of</strong> fraction<strong>at</strong>ion in<br />

calc-alkaline magmas. Contributions to Mineralogy and Petrology 90,<br />

142^161.<br />

McBirney, A. R., Baker, B. H. & Nilson, R. H. (1985). Liquid fraction<strong>at</strong>ion.<br />

Part 1: basic principles and experimental simul<strong>at</strong>ions.<br />

Journal <strong>of</strong> Volcanology and Geo<strong>the</strong>rmal Research 24,1^24.<br />

McBirney, A. R., Taylor, H. P. & Armstrong, R. L. (1987). Paricutin<br />

re-examined: a classic example <strong>of</strong> crustal assimil<strong>at</strong>ion in<br />

calc-alkaline magma. Contributions to Mineralogy and Petrology 95,<br />

4^20.<br />

McDonough, W. F. & Sun, S.-S. (1995). The composition <strong>of</strong> <strong>the</strong> Earth.<br />

Chemical Geology 120, 223^253.<br />

Nielsen, R. L., Michael, P. & Sours-Page, R. (1998). Chemical and<br />

physical indic<strong>at</strong>ors <strong>of</strong> compromised melt inclusions. Geochimica et<br />

Cosmochimica Acta 62, 831^839.<br />

Pearce, J. A. & Pe<strong>at</strong>e, D. W. (1995). Tectonic implic<strong>at</strong>ions <strong>of</strong> <strong>the</strong> composition<br />

<strong>of</strong> volcanic arc lavas. <strong>An</strong>nual Review <strong>of</strong> Earth and Planetary<br />

Sciences 23, 251^285.<br />

Pearce, J. A., Perkins, W. T., Westg<strong>at</strong>e, J. A., Gorton, M. P., Jackson, S.<br />

E., Neal, C. R. & Chenery, S. P. (1997). A compil<strong>at</strong>ion <strong>of</strong> new and<br />

published major and trace element d<strong>at</strong>a for NIST SRM 610 and<br />

NIST SRM 612 glass reference m<strong>at</strong>erials. Geostandards Newsletter 21,<br />

115^144.<br />

Pichavant, M., HerreraJac<strong>into</strong>,V., Boulmier, S., Briqueu, L., Joron, J.,<br />

Juteau, M., Marin, L., Michard, A., Sheppard, S., Treuil, M. &<br />

Vernet, M. (1987). The Macusani glasses, SE Peru; evidence <strong>of</strong><br />

chemical fraction<strong>at</strong>ion in peraluminous magmas. In: Mysen, B. O.<br />

(ed.) <strong>Magm<strong>at</strong>ic</strong> Processes, Physicochemical Principles; a Volume in Honor <strong>of</strong><br />

H<strong>at</strong>ten S. Yoder, Jr. Geochemical Society Special Public<strong>at</strong>ion 1,359^373.<br />

Pioli, L., Erlund, E., Johnson, E., Cashman, K., Wallace, P., Rosi, M.<br />

& Delgado Granados, H. (2008). Explosive dynamics <strong>of</strong> violent<br />

Strombolian eruptions: The eruption <strong>of</strong> Paricutin Volcano 1943^<br />

1952 (Mexico). Earth and Planetary Science Letters 271, 359^368.<br />

Reagan, M. K. & Gill, J. B. (1989). Coexisting calcalkaline and<br />

high-niobium basalts from Turrialba Volcano, Costa Rica;<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


ROWE et al. MAGMATIC PLUMBING SYSTEM, PARICUTIN<br />

implic<strong>at</strong>ions for residual titanites in arc magma sources. Journal <strong>of</strong><br />

Geophysical Research 94,4619^4633.<br />

Righter, K., Chesley, C. T. & Ruiz, J. (2002). Genesis <strong>of</strong> primitive,<br />

arc-type basalt: constraints from Re, Os, and Cl on <strong>the</strong> depth <strong>of</strong><br />

melting and role <strong>of</strong> fluids. Geology 30, 619^622.<br />

Roedder, E. (1979). Origin and significance <strong>of</strong> magm<strong>at</strong>ic inclusions.<br />

Bulletin <strong>of</strong> Mineralogy 102, 487^510.<br />

Roedder, E. (1984). Fluid Inclusions. Mineralogical Society <strong>of</strong> America,<br />

Reviews in Mineralogy 12, , 644 p.<br />

Roedder, E. & Emslie, R. (1970). Olivine^liquid equilibrium.<br />

Contributions to Mineralogy and Petrology 29, 275^289.<br />

Rowe, M. C., Nielsen, R. L. & Kent, A. J. R. (2006). <strong>An</strong>omalously<br />

high Fe contents in rehomogenized olivine hosted melt inclusions<br />

from oxidized magmas. American Mineralogist 91, 82^91.<br />

Rowe, M. C., Kent, A. J. R. & Nielsen, R. L. (2007). Determin<strong>at</strong>ion <strong>of</strong><br />

sulfur speci<strong>at</strong>ion and oxid<strong>at</strong>ion st<strong>at</strong>e <strong>of</strong> olivine hosted melt inclusions.<br />

Chemical Geology 236,303^322.<br />

Rowe, M. C., Kent, A. J. R. & Nielsen, R. L. (2009). Subduction influence<br />

on oxygen fugacity and trace and vol<strong>at</strong>ile elements in basalts<br />

across <strong>the</strong> Cascade volcanic arc. Journal <strong>of</strong> Petrology 50, 61^91.<br />

Rowe, M. C., Pe<strong>at</strong>e, D. W. & Newbrough, A. (2011). Compositional<br />

and <strong>the</strong>rmal evolution <strong>of</strong> olivine-hosted melt inclusions in<br />

small-volume basaltic eruptions: a ‘simple’ example from Dotsero<br />

Volcano, NW Colorado. Contributions to Mineralogy and Petrology 161,<br />

197^211.<br />

Rudnick, R. L. & Gao, S. (2004). Composition <strong>of</strong> <strong>the</strong> continental<br />

crust. In: Holland, H. D. & Turekian, K. K. (eds) Tre<strong>at</strong>ise on<br />

Geochemistry,Vol. 3. Amsterdam: Elsevier, pp. 1^64.<br />

Shimizu, N., Semet, M. P. & Alle' gre, C. J. (1978). Geochemical applic<strong>at</strong>ions<br />

<strong>of</strong> quantit<strong>at</strong>ive ion-microprobe analysis. Geochimica et<br />

Cosmochimica Acta 42, 1321^1334.<br />

Spandler, C., O’Neill, H. St. C. & Kamenetsky, V. S. (2007). Survival<br />

times <strong>of</strong> anomalous melt inclusions from element diffusion in olivine<br />

and chromite. <strong>N<strong>at</strong>ure</strong> 447, 303^306.<br />

2219<br />

Spera, F. J. & Bohrson, W. A. (2001). Energy constrained open-system<br />

magm<strong>at</strong>ic processes I: general model and energy-constrained assimil<strong>at</strong>ion<br />

and fractional crystalliz<strong>at</strong>ion (EC-AFC) formul<strong>at</strong>ion.<br />

Journal <strong>of</strong> Petrology 42,99^1018.<br />

Spera, F. J. & Bohrson, W. A. (2004). Open-system magma chamber<br />

evolution: an energy-constrained geochemical model incorpor<strong>at</strong>ing<br />

<strong>the</strong> effects <strong>of</strong> concurrent eruption, recharge, variable assimil<strong>at</strong>ion<br />

and fraction crystalliz<strong>at</strong>ion (EC-E’RAwFC). Journal <strong>of</strong> Petrology 45,<br />

2459^2480.<br />

Strong, M. & Wolff, J. (2003). Compositional vari<strong>at</strong>ions within scoria<br />

cones. Geology 31, 143^146.<br />

Toplis, M. J. (2005). The <strong>the</strong>rmodynamics <strong>of</strong> iron and magnesium partitioning<br />

between olivine and liquid: criteria for assessing and predicting<br />

equilibrium in n<strong>at</strong>ural and experimental systems.<br />

Contributions to Mineralogy and Petrology 149,22^39.<br />

Urrutia-Fucugauchi, J. & Uribe-Cifuentes, R. A. (1999). Lower-crustal<br />

xenoliths from <strong>the</strong> Valle de Santiago Maar Field, Michoacan^<br />

Guanaju<strong>at</strong>o Volcanic Field, Central Mexico. Intern<strong>at</strong>ional Geology<br />

Review 41, 1067^1081.<br />

Verma, S. P. & Hasenaka, T. (2004). Sr, Nd, and Pb isotopic and trace<br />

element geochemical constraints for a veined-mantle source <strong>of</strong><br />

magmas in <strong>the</strong> Michoacan^Guanaju<strong>at</strong>o Volcanic Field, west^central<br />

MexicanVolcanic Belt. Geochemical Journal 38, 43^65.<br />

von Seckendorff,V. & O’Neill, H. St. C. (1993). <strong>An</strong> experimental study<br />

<strong>of</strong> Fe^Mg partitioning between olivine and orthopyroxene <strong>at</strong> 1173,<br />

1273 and 1423 K and 1·6 GPa. Contributions to Mineralogy and<br />

Petrology 113, 196^207.<br />

Wilcox, R. E. (1954). Petrology <strong>of</strong> Paricutin volcano, Mexico. US<br />

Geological Survey Bulletin 965C, 281^353.<br />

Zinner, E. & Crozaz, G. (1986). Ion probe determin<strong>at</strong>ion <strong>of</strong> <strong>the</strong> abundances<br />

<strong>of</strong> all <strong>the</strong> rare earth elements in single mineral grains. In:<br />

Benninghoven, A., Colton, R. J., Simons, D. S. & Werner, H. W.<br />

(eds) Secondary Ion Mass Spectrometry (SIMS-V). Berlin: Springer,<br />

pp. 444^446.<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013


APPENDIX<br />

JOURNAL OF PETROLOGY VOLUME 52 NUMBER 11 NOVEMBER 2011<br />

Fig. A1. (a) Loc<strong>at</strong>ion map for Paricutin volcano within <strong>the</strong><br />

Michoacan^Guanaju<strong>at</strong>o volcanic field (MGVF) <strong>of</strong> central Mexico;<br />

(b) map <strong>of</strong> <strong>the</strong> Paricutin region showing <strong>the</strong> local distribution <strong>of</strong><br />

cinder cones; modified from McBirney et al. (1987) and Erlund et al.<br />

(2010). The pl<strong>at</strong>e configur<strong>at</strong>ion is also indic<strong>at</strong>ed in (a). P, Paricutin volcano;<br />

RP, Rivera Pl<strong>at</strong>e.<br />

2220<br />

Downloaded from<br />

http://petrology.oxfordjournals.org/ by guest on April 3, 2013

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!