textural eyidence of mafic-felsic magma interaction in

American Mineralogist, Volume 77, pages 795-809, 1992

Textural eyidence of mafic-felsic magma interaction in dacite lavas,Clear Lake, California

J,lNrBs A. Srrnrncr* THorvrls H. PrancnDepartment of Geological Sciences, Queen's University, Kingston, Ontario K7L 3N6, Canada

Ansrnlcr

At Clear Lake, California, the episode of volcanism from 0.65 to 0.30 Ma culminatedin production of mixed dacite and rhyodacite lavas with bimodal phenocryst populationsexhibiting extreme disequilibrium textures. These lavas formed by the interaction of sub-equal fractions of mafic recharge magma with the crystal-rich remnants of earlier-formedfelsic magma bodies. The blending of basaltic andesite and rhyolite magmas dominatedthe early stages of mafic-felsic magma interaction. Partial quenching and segmentation ofthe resulting hybrid magma eventually led to the formation of quenched andesitic inclu-sions. During later stages ofinteraction, quenched inclusions suffered extensive disaggre-gation, contributing crystalline debris and residual silicic liquid to mixed dacites. The largerange in mineral compositions and textures imply that given sufficient mafic input, pre-eruptive mixing is a remarkably efficient process, even in silicic magmas.

Phenocryst assemblages derived from the end-member magmas in dacite lavas preservedistinct compositions and styles of textural reequilibration. Felsic end-member pheno-crysts responded to mixing mainly by dissolution and reaction recorded by (l) a roundedand embayed crystal form, (2) compositional reversals, (3) a partial reaction progressinginward from crystal margins, and (4) mantles or coronas that formed by diffusion-limitedreactions in dissolution boundary layers. Sodic plagioclase derived from the felsic end-member shows a variety of features, including simple dissolution, abrupt shifts to morecalcic composition, and fritted texture (representing a partial to complete reaction to morecalcic compositions). Fritting occurs when sodic plagioclase experiences large changes inboth liquid composition and temperature during residence in hybrid andesitic magma,whereas simple dissolution and calcic shifts occur when sodic crystals residing in felsicmagma experience dramatic increases in temperature, followed by relatively small changesin liquid composition. Sanidine shows features analogous to plagioclase, including simpledissolution and shifts to more Ba-rich compositions. Some sanidine crystals are also man-tled by plagioclase. Mafic end-member magmas (basaltic andesite) responded to mixingby combined liquid blending and undercooled crystallization, resulting in quenched an-desitic inclusions and disseminated crystalline debris, with compositions intermediate be-tween the end-member phenocryst populations. Plagioclase microphenocrysts formed dur-ing mixing are characterized by dendritic or skeletal growth forms and strong normalzoning. Calcic plagioclase phenocrysts present prior to mixing have abrupt sodic shifts andnormal zoning at their margins, with rim compositions that match those of micropheno-crysts in inclusions.

INrnonucrtoN

Mafic-felsic magma interaction is now accepted as animportant control on the chemical and textural diversityof igneous rocks, and some workers have suggested thatmagma mixing is the dominant process leading to theeruption of intermediate composition rocks in otherwisebimodal volcanic centers (Eichelberger,1978; Grove andDonnelly-Nolan, 1986). Thus magma mixing may ob-scure the presence of compositional gaps in such systems,biasing our picture of them if left unrecognized. Deter-

*Present address: Los Alamos National Laboratory, EES-I,Geology/Geochemistry, Los Alamos, New Mexico 87545, U.S.A.

0003-o04x/92l0708-0795$02.00 795

mining the role of magma mixing in the origin of a givenlava or suite provides insight into precursor magma com-positions, their relative volumes, and the nature of theirinteraction in crustal magma chambers.

Our understanding of mafic-felsic magma interactionhas been improved by numerous studies of variably mixedand mingled magmas (e.g., Wilcox, 1944; Anderson, 1976;Heiken and Eichelberger, 1980; Sakuyama, l98l; Baconand Metz, 1984; Bacon, 1986; Frost and Mahood,1987:'Nixon and Pearce, 1987; Nixon, 1988a, 1988b; Clynne,1989), as well as by theoretical and experimental assess-ments of the fluid dynamical, thermal, and chemical con-sequences of mafic-felsic magma interaction (Huppert et

796 STIMAC AND PEARCE: MAGMA INTERACTION IN DACITE LAVAS

Or

SILICIC END-MEI\IBER IPHENOCRYSTS AND CRYSTALAGGREGATES)

o plag coexisting with sanx other ptag

GROUNDMASS

r mrcropnenocrysrs

MAFIC END.MEMBER

tr groundmassmicrophenocrystsin quenched lnclusions

O phenoc.ysts, crystal aggregates, andphenocrysts in quenched inclusions

Fig. l. Generalized geologic map of the Mount Konocti-Thurston Lake area, Clear Lake Volcanics, California (modifiedfrom Hearn etal., 1976). Only rocks from -0.65 to 0.30 Ma areshown (structural detail omitted). Inset map shows the locationof Mount Konocti (MK) relative to San Francisco Bay and theSan Andreas fault (SAB.

al., 1982, I 984; Kouchi and Sunagawa, I 98 5 ; Koyaguchi,1985; Sparks and Marshall, 1986; Turner and Campbell,1986; Ussler and Glazner, 1989; Oldenburg et al., 1989).From these studies it is clear that the style of mafic-felsicmagma interaction depends on the relative temperatures,compositions, and volumes of the magmas involved.When thermal and compositional contrasts between twomagmas are large or the relative volume of mafic magmasmall compared to silicic magma, their interaction willbe dominated by extensive crystallization of the maficcomponent, and little or no hybrid liquid will form (Ba-con, 1986; Sparks and Marshall, 1986). In extreme cases,the small mafic input into rhyolitic systems is effectivelyquenched, with little evidence for liquid blending (Baconand Metz, 1984; Stimac et al., 1990). Conversely, whenthe mass fraction of the mafic component is large (> 300/o)or where compositional differences are small (<10 wto/oSiOr), liquid blending will be more important (Kouchiand Sunagawa, 1985; Nixon and Pearce, 1987; Nixon,1988a). Bacon (1986) used the term "mingled magmas"to differentiate occurrences dominated by undercooledcrystallization of mafic magma in silicic magma fromthose where liquid blending is more important. Givensufficient mafic input into rhyolitic systems, mixing ap-pears to occur in stages. The early stages are dominatedby liquid blending and the formation of hybrid andesite,whereas in later stages, hybrid andesitic layers segment,forming quenched inclusions. Ultimately, disaggregationof inclusions may lead to fine-scale dissemination of in-clusion debris in silicic magmas (Thompson and Dungan,1985; Clynne, 1989; Stimac and Pearce, 1989), obscuringthe distinction between liquid blending and mingling (asoutlined above).

0 2 0 4 0 6 0 B 0 1 0 0

/ qsq6& \O I

- Y ' i ! x ! & x x x ' " \

0 2 0 4 0 6 0 8 0 r 0 0

Ab An

Fig.2. Sanidine (san) and plagioclase (plag) compositions indacitic lavas and andesitic inclusions from Clear I-ake. Group-ings are based on occurrence, composition, and texture, as de-scribed in text. See Table 2 for representative microprobe anal-yses of each textural type. End-member plagioclase phenocrystcompositions average -Anro and Anro, but some calcic zonesand cores in otherwise sodic crystals range to Anuo. Micropheno-crysts in dacite lavas and quenched inclusions have overlappingranges intermediate in composition between end-member phe-nocrysts.

Study of the textures and compositions of minerals hasproved to be the simplest and most effective means ofestablishing the role of magma mixing because mineralscommonly preserve evidence of diverse and complex his-tories, even in magmas that are compositionally homo-geneous. Furthermore, such study provides critical con-straints on the nature of the mixing process itself. In thiswork we provide an inventory of textures and mineralzoning patterns in rhyodacite to dacite lavas from ClearLake, California. These lavas represent preeruptivemixtures of subequal volumes of crystal-rich rhyolitic andbasaltic andesite magmas. Emphasis is placed on the feld-spars, because feldspar zoning and reaction textures pro-vide a sensitive record of shifting magmatic conditions(Smith and Brown, 1988). The images used in this paperare of crystals from a number of dacitic units defined byHearn et al. (1976, and unpublished data). Only a briefoutline of rock and mineral compositions is given here.Sample locations, modal proportions, and ranges in rock

STIMAC AND PEARCE: MAGMA INTERACTION IN DACITE LAVAS 797

TABLE 1, Representative microprobe analyses of feldspars

sio, 65.39Al2o3 18.99FeO' 0.05CaO 0.18NarO 4.09K,O 10.21BaO 1.42

Total 100.33

65.7319.08

0.103.21

11.890.49

100.50

60.9823.880.096.067.49o.82

99.32

53.5628.930.11

12.134.720.23

99.68

47.2933.06o.42

17.421.89

100.08

51 .19 55.2130.27 28.340.27 0.44

14.08 11 .543.56 4.980.15 0.28

99.52 100.79

Note..Analyses correspond to feldspar types in dacite lavas and andesitic quenched inclusions plotted in Figure 2. Numbers denote the following:1 : san phe;ocryst rim ior6,An,) lcrossesi; 2: san phenocryst core (Or?,Anl) (crosses); 3: core of plag phenocryst (AnFOr5) associated with san(circles); 4 : catiic zone in plag phenocryst core (Ansorr) (x;s); 5: plag microphenocryst (An6ori) in dacite groundmass (filled triangles); 6: plag

microphenocryst (An*OrJ in quenched inclusion (squares); 7 : plag phenocryst (AnsAb,?) in quenched inclusion (diamonds).. All Fe as FeO.

and mineral compositions for individual dacitic units canbe found in Stimac (1991).

Gnor,ocrc sETTING

The Clear Lake Volcanics are located - 150 km north-northeast of San Francisco. California. within the broadSan Andreas fault system. Volcanism at Clear Lake oc-curred in at least three episodes, with the greatest volumeof silicic lava (rhyolite to dacite) erupted from 0.65 to0.30 Ma (Donnelly-Nolan et al., l98l). Silicic unitserupted from 0.65 to 0.30 Ma can be divided into severalrelated sequences informally termed (l) crystal-poor rhy-olites, (2) crystal-rich rhyolites, (3) dacites of ThurstonI-ake, and (4) dacites of Mount Konocti (Fig. l). Rhyo-lites were erupted south and west of Mount Konocti fromabout 0.65 to 0.50 Ma (Donnelly-Nolan et al., 198 l). Thefirst rhyolites erupted were an extensive series (-6 km3)of crystal-poor lavas, including the rhyolite of ThurstonCreek (see Stimac et al., 1990). They were followed bythe eruption of lesser amounts (<3 km3) of crystal-richrhyolite flows and tuffs (up to 33 volo/o phenocrysts) alongthe southern and western margin of Mount Konocti. Vol-canism culminated with the eruption of -20 km3 ofrhyodacitic to dacitic lava in the Mount Konocti-Thurs-ton Lake area from -0.5 to 0.3 Ma (Donnelly-Nolan eta1., l98l). The dacites of the Thurston Lake area (-3km3) form a series of coalescing flows and domes withK-Ar ages ranging from 0.48 to 0.44 Ma (Donnelly-No-lan et al., l98l). The dacites of Mount Konocti make upa composite volcano consisting mainly of lava flows anddomes that rises 1000 m above the shore of Clear I-ake.The dacites of Mount Konocti have an estimated volumeof over l6 km3 (Hearn et al., unpublished data) with K-Arages ranging from 0.43 to 0.26 Ma (Donnelly-Nolan etal.. 198 l). A lesser volume of mafic rocks is interbeddedwith the silicic sequences (Fig. l).

AN,ll,vrrc,q.L METHoDS

Microprobe analyses were conducted at Queen's Uni-versity and Rensselaer Polytechnic Institute (RPD. EDSanalyses of feldspars at Queen's were obtained on an ARL-SEMQ electron microprobe fitted with a Princeton Gam-ma Tech energy-dispersive spectrometer using a Tracor-Northern X-ray analysis system. Ten-element analyses

were performed using an accelerating potential of l5 kV,a beam current of 100 nA, and count times from 100 to200 s. A synthetic basaltic glass was used as a primarystandard to calibrate the EDS spectrum. WDS feldsparand glass analyses done at RPI were obtained on a JEOL733 Superprobe using an accelerating potential of l5 kVat a beam current of 15 nA and employing appropriatesecondary standards. In both cases, X-ray intensities werecorrected for matrix effects using the procedures ofBenceand Albee (1968) and a correction factors of Albee andRay (1970). Further details ofthe analysis procedure, pre-cision, and accuracy can be found in Stimac (1991).

Interference imaging techniques used in this study havebeen described previously by Nixon and Pearce (1987)and Pearce et al. ( I 98 7) and are briefly noted below. Nar-row-fringe laser interferometry and Nomarski differentialinterference contrast (DIC) studies were carried out in the

Queen's University Laser Lab, as described by Pearce( 1984a, 1984b). Nomarski DIC images were produced byviewing samples etched with fluorboric acid in reflectedlight (see Anderson, 1983; Pearce and Clark, 1989; fordetails of the technique).

Dlcrrp LAvAS AT CLEAR LAKE

Dacitic rocks of the Mount Konocti-Thurston Lakearea contain bimodal phenocryst assemblages exhibitingextreme disequilibrium textures consistent with an originby magma mixing (Anderson, 1936; Brice, 1953; Hearnet al., 198 l). Examination of the temporal trends, whole-rock compositions, and mineral assemblages and texturesof other magmas erupted from 0.65 to 0.30 Ma (Fig. 1)indicate that the end-member assemblages in mixingevents were derived from precursor rhyolitic and basalticandesite to andesite magmas (Stimac, l99l). Despite par-tial reequilibration, precursor mineral assemblages can beidentified based on their occurrence in crystal aggregateswith distinctive compositions, grain sizes, and textures.The silicic assemblage consists primarily of coarse-grainedsanidine (up to 2 cm), sodic plagioclase, quartz, Fe-richpyroxene, biotite, and ilmenite. Mineral compositions andtextures of this assemblage are nearly identical to thosefound in crystal-rich rhyolites erupted prior to the dacites(Stimac, l99l). The mafic assemblage consists primarilyof finer-grained calcic plagioclase (typically <2 mm), Mg-


rich pyroxenes, and lesser olivine. The mineral compo-sitions and textures of this assemblage are similar to thosefound in basaltic andesite lavas and andesitic quenchedinclusions in rhyolitic rocks, but compositions of maficminerals in individual units varv (Stimac et al.. 1990:

Stimac, l99l). Feldspar compositions for representativedacite lavas are summarized in Figure 2 and Table l.Plagioclase phenocrysts clearly form two populationscentered at An3o and Anro. Microphenocrysts in quenchedinclusions and microphenocrysts in the groundmass of

(--

Fig. 3. Textural features in dacite lavas from Clear Lake (planepolarized light and crossed nicols). Mineral composition basedon microprobe analyses. Abbreviations for minerals in this andfollowing figures are plagioclase, plag or P; sanidine, san or S;orthopyroxene, opx; clinopyroxene, cpx; biotite, bio; q\artz, qlz;olivine, o1; ilmenite, ilm; glass, gl; and vesicle, v. See Stimac(1991) for more information on individual samples. (A) Largeplag with a fritted exterior (-Ano) and sodic core (Anrr; center)derived from the silicic end-member and Mg-rich olivine (Foru)rimmed by amphibole (upper right) (sample SCL-18C; dacite ofKonocti Bay; DKB) derived from the mafic end-member. Re-action textures are interpreted as the result of magma mixing.(B) Augite corona on qtz (sample SCL-l4A; dacite of ThurstonLake). The corona consists of bladed cpx with abundant inter-stitial glass. We interpret such coronas as forming by diffusion-controlled growth of cpx in a dissolution boundary layer on qtz(Sato, 1975) induced by magma mixing. (C) Pseudomorph after

799

bio consisting offine-grained oxides and silicates (sample SCL-37C; dacite of Vulture Rock; DV). Similar reaction products ofbio in dacites were interpreted by Nixon (1988a) as resultingfrom dehydration above its stability limit attending magma mix-ing. (D) Fe-rich opx rimmed by more Mg-rich fine-fritted opxin a felsic crystal aggregate (sample SCL-9B; dacite of Soda Bay;DSB). Sodic plag in the agegate shows fine-fritted texture at itsmargin. (E) Quenched andesitic inclusion in rhyodacite host(sample SCL-IlBl; dacite of Benson Ridge; DBR). The inclu-sion contains Mg-rich pyroxene phenocrysts set in a groundmass

ofbladed plag, opx, opaques, and vesicular glass. (F) Fine-frittedplag in quenched inclusion thought to represent reaction prod-

ucts of a sodic phenocryst (sample SCL-10; DBR unit). Similarpartially fritted crystals with sodic cores are common in andes-itic inclusions and the rhyodacitic host. The crystal has a clearovergrowth -An' similar to plag microphenocrysts in the in-clusions groundmass (Fig. 2).

STIMAC AND PEARCE: MAGMA INTERACTION IN DACITE LAVAS

the host dacites have broadly overlapping ranges inter-mediate in composition between the end-member phe-nocryst populations (Fig. 2).

Mineral assemblages and compositions are consistentwith a history of episodic mafic recharge into the partlycrystallized remnants of silicic magma bodies resulting inthe production ofhybrid dacite and rhyodacite lavas (Sti-mac and Pearce. 1989: Stimac. l99l). Whole-rock anal-yses of lavas and inclusions providing plausible end-member compositions are listed in Table 2. Mass balancetests that utilize these end-members yield good fits (R'from -0. I to 0.4), whereas crystal fractionation modelsinvariably fail (Stimac, l99l). Successful mass balancemodels indicate that the fraction of mafic input rangedfrom -30 to 60 wto/o.

Onsnnvlrror.,ts

Textures of the silicic end-member phenocrysts

Phenocrysts derived from the silicic end-member inClear Lake dacite lavas display a variety of disequilibri-um textures (Fig. 3). Sodic plagioclase exhibits abrupt

shifts to a more calcic composition and a fritted or sievetexture (Figs. 3A, 3D, 3F). Sanidine is invariably anhe-dral and embayed and is commonly mantled by plagio-clase. Quartz is similarly rounded and embayed and com-monly supports coronas of clinopyroxene (Fig. 3B). Biotiteis anhedral to subhedral and rimmed or replaced byfine-grained intergrowths rich in titanomagnetite and py-roxene (Fig. 3C). Fe-rich orthopyroxene exhibits com-positional reversals or is replaced by more Mg-rich or-thopyroxene with a fine-fritted texture (Fig. 3D).Representative textures are described in more detail be-low.

Fine-scale plagioclase zoning. Zoning in plagioclasephenocrysts at Clear Lake can be divided into fine-scaleoscillatory zoning and larger-scale compositional shiftssuperimposed on fine-scale patterns. Fine-scale zoning isthought to be controlled by the interplay of growth anddiffusion rates in compositional and thermal boundarylayers (unrelated to magma mixing) surrounding individ-ual crystals (Sibley et al., 19761, Kuo and Kirkpatrick,1982; Loomis, 1982; Pearce and Kolisnik, 1990). Fine-

TABLE 2. Representative whole-rock analyses of lavas and quenched inclusions

SampleUni t

H76-174BK

SCL.14DDT/OI

SCL-11GDBR/OI

scL-14DT

scl-11DBR

scL-6RR

sio, 56.5Tio, 0.87Af2o3 17.6FeO- 5.57MnO 0.09MgO 5.8CaO 8.7Na2O 3.1K.O 0.85PrO, 0.23LOI 0.78

Total 100.09

57.19 63.390.91 0.65

16.91 18.175.19 3.880.08 0.065.56 2.617.09 5.342.88 3.841 .78 1.970.15 0.092.25 0.40

99.99 100.40

66.210.54

14.773.260.072.573.753.353.410 .101 .70

99.73

67.250.52

16.173.140.061.804.163.732.920.150.80

100.70

75.340.31

13.661.460.030.301.203.604.560.020.40

100.88

Nofei LOI : loss on ignition. See Stimac (1991) for analytical methods. BK: basaltic andesite lava interbedded with dacite. DT/QI : quenchedinclusion in dacite of Thurston Lake. DBR/QI : quenched inclusion in dacite of Benson Ridge (Mount Konocti). DT : dacite of Thurston Lake. DBR : daciteof Benson Ridge. RR : rhyolite of Red Hill Road.

. All Fe as FeO.


scale oscillations in plagioclase phenocrysts derived fromthe felsic end-member (-Anro_uoi ave. An30) at Clear Lakeare typically 10-40 pm thick (see Figs. 4,A., 4B). Theyappear to be broader, more diffuse, and of lower com-positional amplitude than similar zones described fromandesites (Nixon and Pearce, 1987; Pearce and Kolisnik,1990). Large phenocrysts (Fig. aA) commonly containbroad bands (up to I mm wide) consisting of numerousfine-scale zones, all with compositional variation of < 5molo/o An.

Larger-scale plagioclase zoning. Larger-scale compo-sitional discontinuities typically separate broad bands of

fine-scale zones in silicic-derived crystals (Figs. aA-4F).These discontinuities are commonly marked by evidenceoferosion or truncation ofunderlying zones along a rag-ged interface (Figs. 4B, 4C). Such interfaces are typicallyfollowed by shifts to more calcic compositions (-5-15molo/o An; see Figs. 4D, 4F). Some crystals contain sev-eral such rounded interfaces, each followed by a markedshift to a more calcic composition (Fig. 4A). Steep normalzoning generally follows calcic shifts, with compositionsbecoming at least as sodic as in underlying zones (Figs.4D,4D.

Plagioclase reaction textures. In Clear Lake dacites,


e-

Fig. 4. Plagioclase zoning patterns in felsic end-membercrystals (crossed nicols and laser interferograms). Sodic plag phe-

nocrysts in silicic rocks at Clear Lake are mainly from about 0.5

to 2.5 mm long, but some crystals range up to 2 cm. The com-positional range of most crystals is small (< 15 molo/o An), but

some crystals contain zones (usually cores) that are as calcic asAn* (Fig. 2, Table 1). The overall style of zoning is even tonormal oscillatory, punctuated by calcic spikes (resorption sur-face, abrupt shift to more calcic composition, and strong normalzoning) that are thought to record magma mixing events (see

discussion). (A) Plagioclase megacryst (- I cm in length) record-ing a complex history of growth and resorption (sample SCL-9.A; DSB unit). The crystal is oriented such that more calciccompositions show higher birefringence Qighter shades). The coreis calcic (Anor) surrounded by a patchy zoned region (Antn-ro).

The rim is partially resorbed and has a thin calcic overgrowth(>An,o). The remainder of the crystal consists of broad sodicbands (Anru-rr) with fine-scale oscillatory zoning, separated byresorption surfaces and compositional reversals of < 5 to l5 mol0/oAn (arrows). (B) Sodic plag with even to normal oscillatory zon-ing (compositional change across fine-scale oscillations are <5

molo/o An) cut by two ragged interfaces (arrows) that truncate

underlying zones (sample SCL-9E; DSB unit). The crystal is ori-

ented such that more calcic compositions show lower birefrin-gence (darker shades). Compositional reversals occur outside of

each interface, with the maximum compositional shift being - 10

molo/o An (arrows). (C) Sodic plag with a pronounced calcic spike

at arrow (sample SCL-I1B2; DBR unit). The crystal is oriented

such that calcic compositions show higher birefringence (lighter

shades). The crystal is fritted at its margin, with a thin over-growth ofcalcic plag. (D) The laser interferogram ofthe grain in

C records a compositional reversal of 5-10 molo/o An, followed

by steep normal zoning to the fine-fritted margin. (Laser inter-

ferograms consist of a series of parallel fringes that are shifted

across zones ofdifferent refractive index-a shift ofone {hll fringe

measures a compositional change of -32 molo/o.) (E), (F) Sodicplag with a more complicated zoning history that includes abrupt

shifts (> 5 molo/o) to both more calcic (higher birefringence) and

more sodic compositions (sample DCJ-86-228; DKB unit). The

pronounced calcic spike near the edge ofthe crystal (outer arrow)

is - 15 molo/o An.

some sodic plagioclase crystals are partially fritted or sievetextured (Figs. 3, 5). Fritted and unfritted grains may oc-cur side by side in dacite lavas, but virtually all crystalsin andesitic inclusions with sodic cores have fritted ex-teriors. Some fritted crystals in inclusions (and less com-monly in dacites) consist entirely of sieve-textured pla-gioclase (Fig. 3F). Fritted zones consist ofan intergrowthof plagioclase and glass with a highly variable morphol-ogy (Figs. 5B-5D). When fritting is sufficiently coarse toobtain reliable microprobe analyses, compositions gen-

erally range from Anro to Anro. Most fritted crystals haveclear, normally zoned overgrowths (Fig. 5C) ranging from-Ano, to An.r. The fritted texture typically truncates un-derlying zones, and in some crystals the size of the glass

inclusions decreases inward (Figs. 5A, 5B)'Sanidine zoning. Sanidine in silicic lavas at Clear Lake

has a limited range in composition, with the style of san-idine zoning being similar to that of sodic plagioclase.

When the celsian component is excluded, sanidine coresrange from Oru, to Orrr. However, rims of strongly re-acted crystals may be as sodic as Orr, (Fig. 2). These moresodic compositions are present around crystal margins,embayments, and cracks and probably result from alkaliexchange during and following mixing events (Fig. 68)'When these areas are excluded, zoning in sanidine islargely restricted to Ba and to a lesser extent, Ca. BaOconcentrations range from about 0.4 to 1.8 wto/o in a typ-ical megacryst, with the largest increases occurring in ar-eas outside ofrounded interfaces that truncate underlyingzones (Fig. 6A). Individual zones are typically broad (from20 to 500 pm thick), diffuse, and separated by roundedinterfaces.

Sanidine reaction and plagioclase mantle textures. San-idine in Clear Lake dacitic and rhyodacitic rocks is in-variably either rounded and embayed, mantled by pla-gioclase, or both (Figs. 68, 6C). Plagioclase mantles have

a variety of forms and sizes, summarized in Figure 6.

Coarse plagioclase mantles (0.5-1.0 mm thick) typically

consist of cellular-textured grains, whereas fine mantles

are commonly bladed (compare Figs. 6C, 6D). Plagio-

clase mantles are generally optically continuous and have

a preferred optical orientation to sanidine. Locally, it can

be seen that the plagioclase of the mantle is separatedfrom the underlying sanidine by a thin zone ofclear glass

(<0.25 mm). Fine-bladed mantles are typically continu-ous on their outer projection but taper inward, terminat-ing in glass-filled channels in sanidine (Fig. 6D)- Coarse-

cellular plagioclase mantles have zoning that includes

repeated compositional reversals. Compositions of man-

tles range from about An, to Anoo Gig. 2).Sanidine is rare in quenched inclusions, but when pres-

ent, it is typically embayed and contains numerous ovoi-

dal cavities. Plagioclase mantles on sanidine in inclusions

commonly have a fine-fritted texture on their outer mar-glns.

Augite coronas on quartz. Quartz grains with augitecoronas are common in dacites at Clear Lake. Coronasconsist of radially oriented bladed mats of clinopyroxeneseparated from quartz and the surrounding dacite by in-

terstitial vesicular glass (Fig. 3B). Rimmed and un-rimmed quartz phenocrysts commonly exist side by sidein dacite lavas, whereas quartz crystals in quenched in-

clusions rarely lack coronas. Fragments of multigrainedaggregates of acicular clinopyroxene with interstitial glass

are also abundant in dacites, whether or not clinopyrox-ene occurs as phenocrysts. corona fragments are most

abundant when other felsic end-member minerals (sani-

dine, plagioclase, biotite) are also strongly resorbed. Au-gite in the coronas has a large compositional range, with

the most Mg-rich compositions being similar to augitephenocrysts derived from the mafic end-member. Inter-

stitial glass in coronas is commonly silicic (Stimac' l99l).


\

Fig. 5. Plagioclase reaction textures in the felsic end-member(Nomarski DIC images). Nomarski interference imaging yieldsan undistorted view ofplagioclase-glass intergrowths in two di-mensions and provides a qualitative estimate of changes in pla-gioclase composition too small to analyze with the microprobe(Pearce and Clark, 1989). (A), (B) Sodic plag with a fine-frittedrim that appears to coarsen outward (sample SCL-I7C; DpF,dacite of Plum Flat). Reaction has proceeded inward from thecrystal margin, producing an open framework of more calcic plagand interstitial glass. The coarser-grained intergrowth ofplag and

glass outside of the fine-fritting may have formed by rapid growth(arrows in B). (C) Plagioclase phenocryst consisting of (l) a sodiccore, (2) a region of fine-fritting that truncates earlier formedzones (arrow at F), and (3) an overgrowth of more calcic com-position (sample SCL-558; DHP, dacite of Henderson Point). Aneighboring plag consists ofa fritted core and calcic overgrowth.In both crystals the overgrowths have skeletal projections thatsuggest rapid growth. Note that the groundmass is texturally het-erogeneous. (D) Coarse-fritted core with clear calcic overgrowth(sample SCL-l4E; DT unit).

Textures of the mafic end-memberMafic end-member textures include those of pheno-

crysts in mafic crystal aggregates and quenched inclu-sions, as well as those of microphenocrysts in quenchedinclusions. In laser interference profiles, calcic plagioclasephenocrysts are seen to have abrupt shifts to more evolvedcompositions at their margins such that rim composr-tions match those of plagioclase microphenocrysts in in-clusions. Mg-rich olivine (Foro ,r) is invariably roundedand nmmed by orthopyroxene or amphibole of the samecomposition as in the groundmass of inclusions (Fig. 3A).Plagioclase, pyroxene, and amphibole microphenocrystsforming the groundmass of quenched inclusions exhibitstrong normal zoning and are characterized by dendriticor skeletal growth forms (Figs. 3E, 3F). Clino- and ortho-pyroxene microphenocrysts commonly occur as parallelintergrowths (Stimac, l99l). Zoning patterns of calcic

plagioclase phenocrysts are described in more detail be-low.

Plagioclase zoning. Calcic plagioclase phenocrysts de-rived from the mafic end-member (Fig. 7) in mixed da-cites have zoning patterns indicative oftwo very diferentstages of growth. Crystal cores have a restricted compo-sitional range (-Anro_roi Fig. 2) and fine-scale even-os-cillatory zoning (Figs. 78, 7D). Cores are jacketed by moresodic rims (-Anro_ur, not plotted in Fig. 2) that are nor-mally zoned and have skeletal or dendritic morphologies.The contact between core and rim is usually sharp andmarked by a shift to more sodic compositions (10-25mo10/o An). In addition, the normally zoned region com-monly contains one episode of reverse zoning shortly af-ter the initial normal shift. This pattern of discontinuouszoning is almost identical to that observed in large crys-tals in quenched inclusions in earlier erupted rhyolites atClear Lake (Srimac et al., 1990).


Fig. 6. Sanidine zoning and reaction textures (crossed ni-cols). Individual crystals range up to 2 cm in length, but mostare several millimeters to 1 cm long. When the celsian compo-nent is excluded, sanidine cores range from Oru, to Orrr, whereasrims may be as sodic as Or,, (Fig. 2,Table l). (A) Zoning in sanmegacryst with plag inclusions and fine-bladed plag mantles atits margin (not shown) (sample SCL-42F; DWP, dacite of WrightPeak). Optical zoning results mainly from changes in Ba concen-tration. Lower birefringence (darker shades ofgray) are regionson higher Ba concentration. Each increase in Ba (arrows) is de-

posited on a rounded interface that truncates earlier formed zones.(B) Coarse plag mantles on san (sample DCJ-86-228; DKB unit).Note the slight increase in birefringence at the sanidine margin,which corresponds to a region of higher Na. (C) Coarse-cellularplag mantle on san with abundant interstitial glass (sample SCL-l lA; DBR unit). Plagioclase inclusions in san are in optical con-

tinuity with the rim. (D) Coarse-grained (upper) and fine-bladed(lower) plag mantles on san (sample SCL-9A; DSB unit). Note

the continuous layer of plag normal to blades near the outermargin of the finer-grained mantle (arrows).

DrscussroN

Silicic end-member

Large-scale plagioclase zoning. Rounded interfaces, likethose found in sodic plagioclases in silicic rocks at ClearLake, are a common feature of plagioclase and have beeninterpreted as dissolution surfaces that formed during ep-isodes of plagioclase resorption (Homma' 1932, 1936;Phemister, 1934; Nixon and Pearce, 1987; St. Seymouret al., 1990; Kolisnik, 1990; Pearce and Kolisnik, 1990).This interpretation is also in accord with the experimen-tal evidence of Tsuchiyama (1985), who found that whena melt is undersaturated in plagioclase (above the liqui-dus for that particular melt composition), plagioclase un-dergoes simple dissolution, leading to rounded crystals.The combination of a rounded or ragged discontinuity(resorption surface) truncating underlying zones followedby a shift to a more calcic composition and subsequentnormal zoning has been observed in numerous studies of

plagioclase zoning in andesitic and dacitic rocks (Nixon

and Pearce, 1987: Pearce et al., 1987; St. Seymour et al.,1990; Kolisnik, 1990; Blundy and Shimizu, l99l). Thiszoning pattem has been referred to as a "calcic spike"(e.g., Fig. 4E). In rocks with other petrographic and chem-ical evidence for mafic-felsic magma interaction, this gen-

eral pattern has been attributed to shifts in temperatureand melt composition attending magma mixing (Nixon

and Pearce, 1987; Kolisnik, 1990). Ifcalcic spikes record

mixing episodes, then multiple reversals imply repeated

mixing events.Plagioclase reaction textures. Numerous petrographic

studies have suggested that a fritted or sieve texture in

plagioclase results from the reaction of sodic crystals to

a more calcic composition (Kuno, 1950; Anderson, 1976',

Nixon and Pearce, 1987). Furthermore, several experi-

mental studies have confirmed that a fritted texture forms

when sodic plagioclase is immersed in a melt that is in

804

Fig.7. Plagioclase zoning patterns of the mafic end-memberunder crossed nicols and as laser interferograms. (A), (B) phe-nocryst in a quenched inclusion (sample SCL-9C; DSB unit).Zoning in the crystal core is even oscillatory (Anro_ro). There isa sharp interface between core and rim, marked by a sodic shiftof -20 molo/o An (arrow in A). The rim shows strong normal

STIMAC AND PEARCE: MAGMA INTERACTION IN DACITE LAVAS

zoning with a small compositional reversal (arrow in B). (C), (D)Calcic plag in dacitic host with an even oscillatory-zoned core,sharp interface, and sodic shift followed by a small composi-tional reversal (arrow), within a normally zoned rim (plane po-larized light and laser interferogram) (sample SCL9A; DSB unit).

equilibrium with more calcic plagioclase (Lofgren andNorris, l98l; Tsuchiyama, 1985; Glazner et a1., 1988,1990). Alternatively, some authors have suggested that afritted texture may form in plagioclase by rapid growth,

during episodes of undercooled crystallization (Hibbard,1981; Anderson, 1984). Kuo and Kirkpatrick (1982) de-scribed fritted crystals compatible with both origins.

Several criteria may be used to differentiate between


Fig. 8. Comparison ofresorption and rapid growth texturesin plagioclase from Clear Lake dacite lavas and andesitic inclu-sions. (A) Sodic plag crystal in a hybrid andesitic inclusion (sam-ple SCL-llG; DBR unit). The crystal consists of a sodic core(An.r), fine-fritted region (more calcic), and a clear rim withskeletal outgowths (An,o). @) A plag crystal with a fine-frittedinterior and clear calcic overgrowth similar to crystal in A butwith remnants of quenched inclusion groundmass (arrows) pre-served in an embayment (sample SCL-55E; DHP unit). (C) Por-tion of a mafic crystal aggregate in a dacite host consisting ofplag and pyroxene (sample SCL-9B; DSB unit). Plag exhibits

fine-scale even oscillatory-zoned cores (Anr-ro) and an abruptshift to more sodic rims along the boundaries of the ag$egate.Rims show normal zoning and skeletal crystal form suggestiveof rapid grolvth (arrow). (D) Plagioclase on the edge of a maficcrystal aggregate (sample SCL-558; DHP uniti. The crystal hasa calcic core (Anro-ro). There is a sharp interface between coreand rim marked by a sodic shift of -20 molo/o An (at a), skeletalregion (from a to b), and a normally zoned outer rim (from b toc). The skeletal region was probably formed by rapid gowth.Also note textural heterogeneity in the gtoundmass surroundingthe crystal.

these hypotheses in a given occurrence. Rapid growthtextures should not disturb underlying zones (Figs. 7A,8D) and may lead to skeletal crystal terminations (Figs.5C, 5D). Conversely, resorption should truncate under-lying zones (Figs. 3A, 58, 5C) and progress inward fromcrystal margins along cracks and grain boundaries (Figs.3D, 8A). Evaluation of compositional information is alsocritical in distinguishing between resorption and rapidgrowth. Ifa fritted texture represents the reaction ofsodicplagioclase to more calcic compositions by preferentialdissolution of the sodic component (Tsuchiyama andTakahashi, I 983), analyses ofthe fritted region should bemore calcic than those of the underlying zones (Figs. 3,A,,3D, 58, 5C, 8A). Conversely, plagioclase formed rapidlyby undercooling should be more sodic than underlyingzones (Figs. 7A,78,8D) (Smith and Lofgren, 1983).

Textural and compositional relations in Clear Lake

dacites indicate that fritted textures have formed by bothreaction and rapid growth (compare Figs. 58, 8D), but afritted texture formed by the reaction of sodic plagioclaseto a more calcic composition appears to be much morecommon. The occurrence of partially fritted grains withsodic cores in quenched andesitic inclusions demon-strates that this texture forms when sodic plagioclase (de-rived from the silicic end-member) is assimilated by maf-ic liquids (see Figs. 3F, 8A). In these grains, fritted regionstruncate early-formed zones, and crystals have clear over-growths with compositions that match those of plagio-clase microphenocrysts in quenched inclusions (Figs. 5C,5D). Similar partially or completely fritted grains in da-cites have remnants of quenched inclusion groundmasspreserved in embayments (Fig. 8B), suggesting that suchcrystals have been cycled through quenched inclusions,which have since been disaggregated.


Sanidine zoning. The abrupt increase in Ba, Ca, and Alin sanidine in areas outside ofthe rounded or ragged in-terfaces (Fig. 7A) appears to result from dissolution ofsanidine, followed by the growth of more Ba-rich com-positions. Ba (and Ca) zoning in sanidine may be moreresistent to solid-state reequilibration than K-Na zoningbecause Ba2* and Ca2* are coupled to Al3* in a mannersimilar to Ca2* in plagioclase (Grove et al., 1984). Theslow rate of coupled Ba-Al and Ca-Al diffusion in sani-dine is consistent with the observation that Ba zoning ismore commonly preserved in potassium feldspar in plu-tonic rocks than is Na-K zoning (Vernon, 1986). We pro-pose the following explanation for Ba zoning. The firstgrowth of sanidine following an episode of dissolution islikely to be Ba rich because the concentration of Ba wouldbe higher in the dissolution boundary layer than in thebulk melt (provided it was not stripped by convection orremoved by diffusion). Subsequent growth would depleteBa in the boundary layer, eventually lowering its concen-tration. Thus we consider the pattern of Ba zoning insanidine to be analogous to certain types ofCa zoning inplagioclase.

Sanidine reaction textures. The plagioclase mantles onsanidine in dacite lavas at Clear Lake bear a striking re-semblance to a rapakivi texture in granitic rocks (Stimacand Wark, 1992). Hibbard (1981) examined numerousexamples of rapakivi texture in plutonic rocks and con-cluded that mixing of mafic and felsic magmas offeredthe best mechanism for producing it. He noted that manyplutonic rocks that contain mantled potassium feldsparalso contain mafic inclusions. More recent studies haveconcluded that the rapakivi texture in fine-grained maficto intermediate composition inclusions results from mag-ma mixing (Vernon, 1986, 1990; Bussy, 1990).

Hibbard (1981) proposed a scenario for the formationof rapakivi texture where mafic magma was injected intofelsic magma and partially quenched by it, producing maf-ic inclusions and hybrid magma. Potassium feldspar de-rived from the felsic magma was resorbed and eventuallyovergrown by dendritic plagioclase as it came into con-tact with hybrid magma. Hibbard interpreted plagioclasemantles on potassium feldspar and a fritted texture inplagioclase phenocrysts as resulting from rapid gowth byundercooling (Lofgren, 1974a). At Clear Lake, mantlecompositions are too sodic to have formed in hybrid an-desite (Fig. 2). In fact, sanidine in andesitic inclusionstypically exhibits evidence of dissolution, and plagioclasemantles have a fritted texture at their margins, implyingsodic mantles were also undergoing reaction. Using thedistribution of mantles, their textures, and compositionsas a basis, we suggest that the growth of mantles may berelated to the formation of dissolution boundary layerson sanidine. Bussy (1990) independently suggested a sim-ilar mechanism for rapakivi formation using observa-tions from plutonic rocks.

Augite coronas on quartz. At Clear Lake, it seems likelythat augite coronas formed in hybrid andesitic magmaderived by the mixing of quartz-bearing silicic magmawith basaltic andesite. Similar coronas have been de-

scribed from basaltic to dacitic rocks worldwide and re-produced experimentally in mafic liquids by a number ofworkers (Sato, 1975; Watson, 1982; Ussler and Glazner,1987 ; Glazner et a1., I 98 8). Sato ( I 975) proposed a modelfor corona growth by diffusion in a boundary layer sur-rounding qrarlz. He documented uphill diffusion of al-kalis into glass surrounding augite and concluded that thehigh alkali content of the boundary layer increased thechemical potential of CaO relative to the groundmass,resulting in the growth of augite, even in magmas notsaturated with clinopyroxene. Quenched inclusions insome dacitic units at Clear I-ake representing hybrid an-desitic magma clearly record this process (Stimac, l99l).They contain quartz with augite coronas, as well as sodicplagioclase with the fritted texture described above. It isalso possible that some augite coronas grew in more si-licic hybrid magmas that were undersaturated with re-spect to quartz. The presence of abundant fragments ofaugite coronas in dacites indicates that quarlz dissolutionsometimes went to completion or that overgrowth tex-tures were disrupted after formation.

Mafic end-member

Plagioclase zoning. The zoning pattern and composi-tional range of calcic cores of phenocrysts in Clear Lakedacite lavas is consistent with crystallization from basal-tic andesite (Fig. 2). The marked shift to a more evolvedcomposition, followed by strong normal zoning, is com-patible with rapid undercooled crystallization of plagio-clase from a restricted supply ofliquid. Lofgren (1974b)experimentally produced similar patterns of discontinu-ous normal zoning, followed by continuous reverse zon-ing in plagioclase by rapid drops in temperature (dis-cussed by Smith and Lofgren, 1983, p. 157, Fig. 5).Lofgren's experiments indicate that the magnitude of thenormal shift was directly proportional to the magnitudeof the temperature drop. In their experiments, each tem-perature drop was followed by a period of isothermalcrystallization that resulted in even to reverse zoning.Lofgren (1974b) argued that rapid temperature drops re-sult in the growth of plagioclase more sodic than thatpredicted at equilibrium, and that continuous reversezoning forms as the system reapproaches equilibrium. Astepwise cooling history is exactly what is expected for acalcic plagioclase phenocryst growing from a mafic mag-ma that was suddenly blended with, and partiallyquenched by, more felsic magma, and it may explain smallcompositional reversals present in some crystals just out-side the core-rim interface (see Figs. 6, 7, 8 in Stimac etal., 1990; Figs. 7B, 7D in this work). Such sodic shifts areobserved in large crystals in both quenched inclusionsand mafic crystal aggregates in dacite lavas, suggestingthat mafic crystal aggregates were once part of quenchedinclusions or that they underwent a similar stage of un-dercooled crystallization (compare Figs. 7B, 7D, 8). Thewidespread occurrence of similar sodic shifts and strongnormal zoning at crystal rims in calc-alkaline rocks (Smithand Lofgren, 1983; Barbarin, 1990; Blundy and Shimizu,199 l) suggests that recycling ofmafic-derived plagioclase


into more silicic magmas is both common and wide-spread.

Disequilibrium textural features of dacite lavas

The large degree of thermal and chemical disequilib-rium induced by the interaction of subequal volumes ofmafic and felsic magma is recorded by textures and themineral assemblages in dacite lavas at Clear Lake (Figs.3-8). Not only do end-member phenocrysts preserve dis-tinct compositions and styles of textural reequilibration,but comparison of textures in quenched inclusions withthose in the dacite host provide constraints on the natureof the mixing process. The phenocryst assemblage de-rived from the felsic end-member responded to mixingmainly by dissolution or reaction. Disequilibrium result-ed in (l) simple dissolution, (2) partial reaction progress-ing inward from crystal margins, (3) local internal meltingof crystals, and (4) formation of mantles or coronas bydiffusion-limited reactions in dissolution boundary layersat the margins of crystals. Silicic end-member crystals inandesitic inclusions consistently display extreme disequi-librium, whereas those in dacite exhibit the complete rangeoftextures described.

The mafic end-member magma responded to mixingmainly by undercooled crystall ization, resulting inquenched inclusions and disseminated crystalline debriswith compositions intermediate between the end-mem-ber phenocryst compositions (Figs. 3E, 3F). Inclusion mi-crophenocrysts are characterized by dendritic or skeletalgrowth forms. Plagioclase microphenocrysts in quenchedinclusions show strong normal zoning and are character-ized by dendritic or skeletal growth forms. Clino- andorthopyroxene microphenocrysts in quenched inclusionscommonly occur as intergrowths, inferred to have formedby epitaxial nucleation in an undercooled environment(Kuno, 1950). Calcic plagioclase phenocrysts in quenchedinclusions and mafic aggegates show abrupt shifts to moreevolved compositions. Compositions of rims on the cal-cic plagioclase phenocrysts match those of plagioclase mi-crophenocrysts in the groundmass of inclusions. Mg-richolivine is rimmed by orthopyroxene or amphibole of thesame composition as that in the groundmass ofinclusions(Fig. 3A). Identical zoning styles and overgrowth texturesare present in both mafic end-member minerals inquenched inclusions and the dacite host. These texturesand zoning patterns reflect rapid growth from liquid inan undercooled environment.

Crystal textures and their environrnent of formation

The variety of mineral textures and compositions inClear Lake dacite lavas and their common distributionin lavas along with hybrid andesite inclusions implies tous distinctly different environments of formation for sometextures. This observation of different textures occurringwithin the same hand specimen (or even thin section) canbe reconciled by a multistage mixing model for daciteformation such as the one illustrated in Figure 9. In thissimplified model, mixing is triggered by the influx of ba-saltic andesite into a crystal-rich silicic magma body.

Fig. 9. Relationships between disequilibrium textures andtheir environments of formation. Stages I and 2 may occur onthe scale of magma body, or locally at a contact between maficand felsic magma. Individual crystals may experience one ormore mixing events, including either or both stages. Disequilib-rium textures ascribed to stage I processes include a fritted tex-ture in sodic plagioclase, augite coronas on quartz, and dissolu-tion of sanidine. Disequilibrium textures ascribed to stage 2include simple dissolution and compositional reversals in pla-gioclase, sanidine, and pyroxene, and plagioclase mantles onsanidine. Rapid growth and reaction textures in the mafic end-member are attributed primarily to stage 1.

Blending of basaltic andesite and rhyolite dominates theearly stages of mafic-felsic magma interaction (stage l).As the mafic magma assimilates silicic liquid and crys-tals, it undergoes rapid crystallization, eventually seg-menting into discrete hybrid inclusions in a more silicichost (compare with Kouchi and Sunagawa, 1985; Nixon,1988a). As crystallization proceeds and the viscosity ofinclusions and host become similar, inclusions suffer ex-tensive disaggregation, contributing crystalline debris andhybrid liquid to mixed dacite (stage 2) (compare withThompson and Dungan, 1985; Clynne, 1989; Poli andTommasini. 1991).

Zr,ningand reaction textures of plagioclase provide clearevidence for the nature of these two environments (Fig.9). Fritting occurs when sodic plagioclase was assimilatedby hybrid magmas formed during the initial stages ofmafic-felsic magma interaction (stage l). These interme-diate magmas rapidly lose heat as they hybridize withadjacent silicic magma and crystallize abundant plagio-clase and pyroxene, eventually forming partially quenchedinclusions. Tsuchiyama (1985) showed experimentallythat sodic plagioclase immersed in a magma saturatedwith more calcic plagioclase develops fritted texture. In

STAGE 2 (RHYOLITE TOLOW Tj but undergoing rapid heatingundersaturated in plagioclasedominated by crystal dissolution and

CRYSTAL TEXTURES AND THEIRENVIRONMENT OF FORMATION

STAGE 1HYBRID ZONE (ANDESITIC}

HIGH Ti but undorgolng rapid coolingoversaturated in prag'oclase (-an4o60)dominated by liquid blend,ng and

808 STIMAC AND PEARCE: MAGMA

most natural examples, fritted regions are overgrown byintermediate composition plagioclase as their host mag-ma crystallizes (Fig. 9). In contrast, sodic crystals notdirectly incorporated into andesitic hybrids experience arapid increase in temperature, possibly followed by a rel-atively small change in liquid composition as their silicichost magma undergoes mechanical mixing with inclusiondebris (stage 2). Sodic crystals with calcic spikes (simpledissolution followed by compositional reversal) record thisprocess. Factors leading to renewed growth of a morecalcic composition probably include (l) changes in liquidcomposition resulting from limited liquid blendiry, (2)the effects of fractionation and cooling following mixing(Nixon, 1988a), and (3) growth from the dissolutionboundary layer previously enriched in plagioclase con-stituents (Kolisnik, 1990).

Although some homogenization undoubtedly occurs inconduits, careful study ofcalc-alkaline plutonic rocks hasrevealed the same range oftextural features as present inextrusive equivalents (Frost and Mahood, 1987; Vernon,1990; Blundy and Shimizu, l99l;Zorpi et al., l99l). Ourobservations at Clear Lake suggest that, provided maficinput is sufficient, mechanical mixing of mafic and silicicmagma is remarkably efficient, despite initially large con-trasts in composition and viscosity.

AcxNowlnocMENTs

This work is a contribution from the Laser Iaboratory at Queen's Uni-versity, supported by National Sciences and Engineering Research Coun-cil of Canada (NSERC) grant no. 0656. Additional financial support forthis study was provided by NSERC grant no. 8709 (T.H P.). The cost offield work was defrayed in part by a grant from Sigma Xi (J.A.S.). Wethank J.M. Donnelly-Nolan, B.C. Heam, Jr., and D. Jacobs for assistancein our study ofthe Clear Lake Volcanics. Useful reviews were providedby M. Clynne, S. Linnernan, and C. Bacon. Thanks also to A. Kolisnikand D. Wark for stimulating discussions concerning mineral zoning andreaction patterns. D. Crawford provided valuable assistance with manu-script preparation.

RnrnnnNcrs crrEDAlbee, A.L., and Ray, L. (1970) Correction factors for electron micro-

analysis of silicates, oxides, carbonates, phosphates and sulphates. An-alytical Chemistry, 42, | 4O8-l 4 | 4.

Anderson, A.T., Jr. (1976) Magma mixing: Petrologic process and vol-canological tool. Journal ofVolcanology and Geothermal Research, l,J _ J J .

-(1983) Oscillatory zoning of plagioclase: Nomarski interferencecontrast microscopy of etched sections. American Mineralogist, 68,r25 -129 .

-(1984) Probable relations between plagioclase zoning and magmadynamics, Fuego Yolcano, Guatemala. American Mineralogist, 69,660-o / o .

Anderson, C.A. (1 936) volcanic history ofthe Clear lake area, California.Geological Society of America Bulletin, 47 , 629-664.

Bacon, C.R. (1986) Magmatic inclusions in silicic and intermediate vol-canic rocks. Journal of Geophysical Research, 9 l, 609 l-6 | 12.

Bacon, C.R., and Metz, J. (1984) Magmatic inclusions in rhyolites, con-taminated basalts, and compositional zonation beneath the Coso vol-canic field, California. Contributions to Mineralogy and Petrology, 85,f46-365.

Barbarin, B. (1990) Plagioclase xenocrysts and mafic magmatic enclavesin some granitoids of the Sierra Nevada Batholith, California. Journalof Geophysical Research, 95, 177 47 -177 56.

Bence, A.E., and Albee, A.L. (1968) Empirical correction factors for elec-

INTERACTION IN DACITE LAVAS

tron microanalysis ofsilicates and oxides. Journal ofGeology, 76,382-386.

Blundy, J.D., and Shimizu, N. (1991) Trace element evidence for plagio-

clase recycling in calc-alkaline magmas. Earth and Planetary Sciencelrtters. 102. 178-197.

Brice, J.C. (1953) Geology of the Lower lake Quadrangle, California.California Division of Mines Bulletin, 166, 34-49.

Bussy, F. (1990) The rapakivi texture offeldspars in a plutonic mixingenvironment: A dissolution-recrystallization process? Geological Jour-nal.25.319-324.

Clynne, M.A. (1989) Disaggregation of quenched magmatic inclusionscontributes to chemical diversity in silicic lavas of Lassen Peak, Cali-fornia. New Mexico Bureau of Mines and Mineral Resources Bulletin,1 3 1 , 5 4 .

Donnelly-Nolan, J.M., Hearn, B.C. Jr., Curtis, G.H., and Drake, R.E.( 198 1) Geochronology and evolution of the Clear Lake volcanics. U.S.Geological Survey Professional Paper | 1 41, 47 -66.

Eichelberger, J.C. (1978) Andesitic volcanism and crustal evolution. Na-l.lre, 275, 2l-27.

Frost, T.P., and Mahood, G.A. (1987) Style of mafic-felsic magma inter-action: The lamarck Granodiorite, Sierra Nevada, Califomia. Geolog-icaf Society ofAmerica Bulletin, 99,272-291.

Glazner, A.F., Ussler, W., III, and Mies, J.W. (19E8) Fate of graniticminerals in mafic magmas. Eos, 69, 1504.

Glazner, A.F., Ussler, W., III, and Mathis, A.C. (1990) Interpretation ofplagioclase texture in volcanic rocks. Eos, 71,1678.

Grove, T.L., and Donnelly-Nolan, J.M. (1986) The evolution of youngsilicic lavas at Medicine [,ake Volcano, California: Implications for theorigin ofcompositional gaps in calc-alkaline series lavas. Contributionsto Mineralogy and Petrology, 92,281-302.

Grove, T.L., Baker, M.B., and Kinzler, R.J. (1984) Coupled CaAlNaSidiffirsion in plagioclase feldspar: Experiments and application to cool-ing rate speedometry. Geochimica et Cosmochimica Actz, 48,2ll3-2 t 2 l

Heam, B.C., Jr., Donnelly-Nolan, J.M., and Goff, F.E. (1976) Preliminarygeologic map and cross-section of the Clear Lake volcanic field, LakeCounty, California. U.S. Geological Survey Open-File Report 76-751.

-(1981) The Clear Lake Volcanics: Tectonic setting and magmasources. U.S. Geological Survey Professional Paper ll4I,25-45.

Heiken, G., and Eichelberger, J.C. (1980) Eruptions at Chaos Crags, Las-sen Volcanic National Park, California. Journal of Volcanology andGeothermal Research. 7. 443-481

Hibbard, M.J. (1981) The magma mixing origin of mantled feldspars.Contributions to Mineralogy and Petrology, 76,158-170.

Homma, F. (1932\ Uber das Ergebnis von Messungen an zonaren Pla-gioklasen aus Andesiten mit Hilfe des Universaldrehtisches. Schweizer-ische mineralogische und petrographische Mitteilungen, 12, 345-352.

-(1936) Classification of zonal structure in plagioclase. MemorialCollege of Science, Kyoto Imperial University Serial B I I, I 35- I 5 5.

Huppert, H.E., Sparks, R.S.J., and Turner, J.S. (1982) Effects ofvolatileson mixing in calc-alkaline magma systems. Nature, 297, 554-557.

-(1984) Some effects of viscosity on the dynamics of replenishedmagma chambers. Journal of Geophysical Research, 89, 6857 -687 7.

Kolisnik, A. (l 990) Phenocryst zoning and heterogeneity in andesites anddacites of Volciin Popocat6petl, Mexico, 247 p. M Sc. thesis, Queen'sUniversity, Kingston, Ontario, Canada.

Kouchi, A., and Sunagawa, I. (1985) A model for mixing basaltic anddacitic magmas as deduced from experimental data. Contributions toMineralogy and Petrology, 89, 17-23.

Koyaguchi, T. (1985) Magma mixing in a conduit. Journal of Yolcanologyand Geothermal Research. 25. 365-369.

Kuno, H. (1950) Petrology ofHakone volcano and adjacent areas, Japan.Geological Society of American Bulletin, 61,957-1020.

Kuo, L., and Kirkpatrick, R.J. (1982) Pre-eruption history ofphyric ba-salts from DSDP kgs 45 and 46: Evidence from morphology andzoning pattems in plagioclase. Contributions to Mineralogy and Pe-trology,79, 13-27.

I-ofgen, G.E. (197 4a) An experimental study of plagioclase crystal mor-phology: Isothermal crystallization. American Journal of Science, 274,243-273.

- (197 4b\ Temperature induced zoning in synthetic plagioclase feld-


spar. In W.W. Mackenzie and J. Zussman, Eds., The feldspars, p. 362-375. Manchester University Press, Manchester, England.

l,ofgren, G.E., and Norris, N. (1981) Experimental duplication of plagio-clase sieve and overgowth textures. Geological Society of AmericaAbstracts with Programs, 13,498.

Loomis, T.P. (1982) Numerical simulations of crystallization processes ofplagioclase in complex melts: The origin of major and oscillatory zon-ing in plagioclase. Contributions to Mineralogy and Petrology, 81, 219-229.

Nixon, G.T. (1988a) Petrology of the younger andesites and dacites ofIztaccihuatl volcano, Mexico: I. Disequilibrium phenocryst assem-blages as indicators ofmagma chamber processes. Journal ofPetrology,29 ,2 t t - 264 .

-(1988b) Petrology of the younger andesites and dacites of lztzcci-huatl volcano, Mexico: II. Chemical stratigraphy, magma mixing, andthe composition of basaltic magma influx. Journal of Petology, 29,265-303.

Nixon, G.T., and Pearce, T.H. (1987) Iaser-interferornetry study of os-cillatory zoning in plagioclase: The record of magma mixing and phe-nocryst recycling in calc-alkaline magma chambers, Iztaccihuatl vol-cano, Mexico. American Mineralogist, 72, 1144-1162.

Oldenburg, C.M., Spera, F.J., Yuen, D.A., and Sewell, G. (1989) Dynamicmixing in magma bodies: Theory, simulations, and implications. Jour-nal of Geophysical Research, 94, 9215-9236.

Pearce, T.H. (1984a) Multiple frequency laser interference microscopy: Anew technique. The Microscope, 32, 69-81.

-(1984b) Optical dispersion and zoning in magmatic plagioclase:Laser-interference observations. Canadian Mineralogist, 22, 383-390.

Pearce, T.H., and Clark, A.H. (1989) Nomarski interference contrast ob-servations of textural details in volcanic rocks. Geology, 17 , 7 57 -7 59 .

Pearce, T.H., and Kolisnik, A.M. (1990) Observations of plagioclase zon-ing using interference imaging. Earth Science Reviews, 29,9-26.

Pearce, T.H., Russell, J.K., and Wolfson, I. (1987) l,aser-interference andNomarski interference imaging of zoning profiles in plagioclase liomthe May 18, 1980, eruption of Mount St. Helens. American Mineral-og|st ,72, l 131-1 143.

Phemister, J. (1934) Zoning in plagioclase feldspar. Mineralogical Mag-azine. 23. 54 l-5 55.

Poli, G.E., and Tommasini, S. (1991) Model for the origin and signifi-cance ofmicrogranular enclaves in calc-alkaline granitoids. Journal ofPetrology, 32, 657-666.

Sakuyama, M. (1981) Petrological study of the Myoko and Kurochimevolcanoes, Japan: Crystallization sequences and evidence for magmamixing. Journal of Petrology, 22, 553-583.

Sato, H. (1975) Diffirsion coronas around quartz xenocrysts in andesitesand basalts from Tertiary volcanic region in northeastern Shikoku, Ja-pan. Contributions to Mineralogy and Petrology, 50,40-64.

Sibley, D.F., Vogel, T.A., Walker, B.M., and Byerly, G. ( I 976) The originofoscillatory zoning in plagioclase: A difftrsion and growth controlledmodel. American Journal of Science. 276, 275-284.

Smith, J.V., and Brown, W.L. (1988) Feldspar minerals, vol. I (2nd edi-tion), 828 p. Springer-Verlag, Berlin.

Smith, R.K., and I-ofgren, G.E. (1983) An analytical and experimentalstudy ofzoning in plagioclase. Lithos, I, 153-168.

Sparks, R.S.J., and Marshall, L.A. (1986) Thermal and mechanical con-

straints on mixing between mafic and silicic magmas. Journal of Vol-

canology and Geothermal Research, 29,99-124.

St. Seymour, K., Vlassopoulos, D., Pearce, T.H., and Rice, C. (1990) The

record ofmagma chamber processes in plagioclase phenocrysts at The-

ra Volcano, Aegean Volcanic Arc, Greece. Contributions to Mineralogy

and Petrology, 104, 7 3-84.Stimac, J.A. (1991) Evolution of the silicic magmatic system at Clear

Iake, California, from 0.6 to 0.3 Ma, 399 p. Ph.D. dissertation, Queen'sUniversity, Kingston, Ontario, Canada.

Stimac, J.A., and Pearce, T.H. (1989) Origin of dacite lavas at Clear l-ake,

California: Disseminated of mafic crystalline debris in silicic magma.

Eos.70. 1409.Stimac, J.A., and Wark, D.A. (1992) Plagioclase mantles on sanidine in

silicic lavas, Clear Lake, California: Implications for the origin of ra-

pakivi texture. Geological Society of America Bulletin, 104, 7 28-7 44'

Stimac, J.A., Pearce, T.H., Donnelly-Nolan, J.M., and Hearn, B.C' Jr'(1990) Origin and implications ofundercooled andesitic inclusions in

rhyolites, Clear Lake Volcanics, Califomia' Joumal of Geophysical Re-

search. 95. 17729-177 46.Thompson, R., and Dungan, M.A. (1985) The petrology and geochem-

istry of the Handkerchief Mesa mixed magma complex, San Juan

Mountains, Colorado. Journal of Volcanology and Geothermal Re-

search. 26. 251-274.Tsuchiyama, A. (1985) Dissolution kinetics of plagioclase in melt of the

system diopside-albite-anorthite and the origin ofdusty plagioclase in

andesites. Contributions to Mineralogy and Petrology, 89' 1-16.

Tsuchiyama, A., and Takahashi, E. (1983) Melting kinetics of a plagio-

clase feldspar. Contributions to Mineralogy and Petrology, 84,345-

f54.Turner, J.S., and Campbell, I.H. (1986) Convection and mixing in magma

chambers. Earth Science Reviews, 23,255-352.

Ussler, W., III, and Glazner, A.F. (1987) Origin of augite coronas aroundquartz xenocrysts in basalts and andesites Eos, 19, 874.

- (1989) Phase equilibria along a basalt-rhyolite mixing line: Impli-

cations for the origin ofcalc-alkaline intermediate lavas. Contributions

to Mineralogy and Petrology, 101,232-244.Yernon, R.H. (1986) K-feldspar megacrysts in granites-phenocrysts, not

porph),roblasts. Earth Science Reviews, 23, 1 -63.-(1990) Crystallization and hybridism in microgranitoid enclave

magmas: Microstructural evidence. Journal of Geophysical Research,

95, 17849-17859.Watson, E.B. (1932) Basalt contamination by continental crust: Some

experiments and models. Contributions to Mineralogy and Petrology,

80, 73-87.Wilcox, R.E. (1944) Rhyolite-basalt complex on Gardiner River, Yellow-

stone Park, Wyoming. Geological Society of America Bulletin' 55' 1047-

1080.Zotpt,M.J., Coulon, C., and Orsini, J'B. (1991) Hybridization between

felsic and mafic magmas in calc-alkaline granitoids-a case study in

northern Sardinia, Italy. Chemical Geology, 92, 45-86.

Mnr.nrscnrsr REcETvED MeY 20, l99lMewuscnrrr ACCEPTED MencH 13, 1992

textural eyidence of mafic-felsic magma interaction in

Documents