Role of the subducted slab, mantle wedge and continental ... · Melting of mafic lower crust may be...
Transcript of Role of the subducted slab, mantle wedge and continental ... · Melting of mafic lower crust may be...
Contrib Mineral Petrol (1996) 123: 263–281 C Springer-Verlag 1996
Charles R. Stern ? Rolf Kilian
Role of the subducted slab, mantle wedge and continental crustin the generation of adakites from the Andean Austral Volcanic Zone
Received: 1 March 1995y Accepted: 13 September 1995
Abstract All six Holocene volcanic centers of the An-dean Austral Volcanic Zone (AVZ; 49–548S) have erupt-ed exclusively adakitic andesites and dacites character-ized by low Yb and Y concentrations and high SryY ra-tios, suggesting a source with residual garnet, amphiboleand pyroxene, but little or no olivine and plagioclase.Melting of mafic lower crust may be the source for adak-ites in some arcs, but such a source is inconsistent withthe high Mg' of AVZ adakites. Also, the AVZ occurs ina region of relatively thin crust (,35 km) within whichplagioclase rather than garnet is stable. The source forAVZ adakites is more likely to be subducted oceanicbasalt, recrystallized to garnet-amphibolite or eclogite.Geothermal models indicate that partial melting of thesubducted oceanic crust is probable below the AustralAndes due to the slow subduction rate (2 cmyyear) andthe young age (,24 Ma) of the subducted oceanic litho-sphere. Geochemical models for AVZ adakites are alsoconsistent with a large material contribution from sub-ducted oceanic crust (35–90% slab-derived mass), in-cluding sediment (up to 4% sediment-derived mass, rep-resenting approximately 15% of all sediment subduct-ed). Variable isotopic and trace-element ratios observedfor AVZ adakites, which span the range reported foradakites world-wide, require multistage models involv-ing melting of different proportions of subducted basaltand sediment, as well as an important material contribu-tion from both the overlying mantle wedge (10–50%mass contribution) and continental crust (0–30% masscontribution). Andesites from Cook Island volcano, lo-cated in the southernmost AVZ (548S) where subductionis more oblique, have MORB-like Sr, Nd, Pb and O iso-
topic composition and trace-element ratios. These can bemodeled by small degrees (2–4%) of partial melting ofeclogitic MORB, yielding a tonalitic parent (intermedi-ate SiO2, CaOyNa2O.1), followed by limited interactionof this melt with the overlying mantle ($90% MORBmelt, #10% mantle), but only very little (#1%) or noparticipation of either subducted sediment or crust. Incontrast, models for the magmatic evolution of Burney(528S), Reclus (518S) and northernmost AVZ (49–508S)andesites and dacites require melting of a mixture ofMORB and subducted sediment, followed by interactionof this melt not only with the overlying mantle, but thecrust as well. Crustal assimilation and fractional crystal-lization (AFC) processes and the mass contribution fromthe crust become more significant northwards in the AVZas the angle of convergence becomes more orthogonal.
Introduction
Possible sources for convergent plate boundary magmasinclude subducted oceanic crust, the overlying mantlewedge and the arc crust. Green and Ringwood (l968)suggested direct partial melting of subducted ocean-floor basalts transformed to eclogites to produce thecalc-alkaline andesites that occur within many arcs. Lat-er experimental (Stern l974) and geochemical (Gill l981)studies concluded that most calc-alkaline andesites areunlikely to be direct partial melts of subducted mid-ocean ridge basalt (MORB). However, some recent stud-ies of subduction zone magmatism have again invoked animportant role for slab-melting in the generation of cer-tain calc-alkaline andesites and dacites, termed adakites(Defant and Drummond l990). Compared to normal calc-alkaline andesites and dacites, adakites have distinctivechemical characteristics suggesting derivation by partialmelting of basalt transformed to garnet-amphiboliteandyor eclogite.
Adakitic andesites and dacites have been identified ina number of arc segments globally, including both thetype locality on Adak Island as well as further west in the
C.R. Stern (✉)Department of Geological Science, University of Colorado,Boulder, CO 80309-0250, USA
R. KilianMineralogisches Institut, Universität Heidelberg,Im Neuenheimer Feld 236, D-69120 Heidelberg, Germany
Editorial responsibility: T.L. Grove
264
Aleutians (Kay l978; Yogodzinski et al. l995), in thesouthernmost Andes (Stern et al. l984; Futa and Sternl988; Kay et al. l993), in both the Baja, Mexico (Rogerset al. l985) and Panamanian (Defant et al. l991a,b) por-tions of the Central American arc, in Japan (Morris1995) and in the Cascade volcanic range of the north-western United States (Defant and Drummond l993). Al-though not volumetrically significant on a global scale,the implications of the distinctive chemistry of adakiteshas motivated renewed attention to the role of melting ofsubducted oceanic crust in the evolution of arc magmas,both presently (Peacock et al. l994; Sen and Dunn l994)and during the Archean (Drummond and Defant 1990;Rapp et al. l991; Martin l993).
One of the arc segments in which adakitic andesitesand dacites were first identified and attributed to meltingof subducted oceanic basalts was the Andean AustralVolcanic Zone (AVZ, Fig. 1; Stern et al. l984). All sixvolcanoes in the AVZ have erupted andesites and daciteswith adakitic chemical characteristics as identified byDefant and Drummond (l990), but these volcanoes alsoexhibit significant chemical differences among them-selves, which span the range of different types of adak-ites identified world-wide. This paper discusses thechemical differences among AVZ volcanoes and theirgenesis within the context of the regional tectonics andgeology of the southern Andes.
Regional geology and tectonics
The volcanoes of the AVZ result from the subduction of young(12–24 Ma) Antarctic plate below southernmost South America atlow convergence velocities of 2 cmyyear (Fig. 1; Cande and Lesliel986). The AVZ is separated from the Andean Southern VolcanicZone (SVZ) by a gap in active volcanism just south of the ChileRise-Trench triple junction, below which even younger (,6 Ma)oceanic crust is being subducted. West of the active volcanoes ofthe AVZ, the age of the subducted plate entering the trenchchanges from 12 Ma in the north to 24 Ma in the south. Theconvergence direction, which is nearly orthogonal at the latitudeof the northernmost volcano in the AVZ (Lautaro, 498S), becomesincreasingly oblique and finally changes to strike-slip plate mo-tion south of the southernmost volcano in the AVZ (Cook Island,548S). Compared to the other volcanoes in the AVZ, the CookIsland volcanic center is located further from the trench along thedirection of plate convergence.
The southern margin of South America is cut by a system oftransform faults, the most important being the Malvinas ChasmFault which merges westwards with the Magallanes Fault Zone(Fig. 1; Fuenzalida l974; Herron et al. l977). Forsyth (l975) de-fined a small microplate, the Scotia plate, south of this fault sys-tem. The southern tip of South America and the Cook Island vol-cano lie on this microplate.
Seismicity is negligible south of the Chile Rise-Trench triplejunction, possibly because of both the slow rate of convergenceand the young age of the oceanic plate (Forsyth l975), which maycause the subducted slab to be several hundreds of degrees hotterthan in more typical subduction environments (Peacock et al.l994). Because of the lack of seismicity, the geometry of the sub-
Fig. 1 Location map of thevolcanic centers of the AndeanAustral Volcanic Zone (AVZ).Also shown are the southernSouthern Volcanic Zone(SSVZ) north of the ChileRise-Trench triple junction,the late Miocene volcano Cer-ro Pampa (CP) in Argentinathat erupted adakitic magmas(Kay et al. l993), and lower-crustal and mantle xenolith lo-calities within the Patagonianplateau basalts (CF Cerro delFraile, Kilian l995;PA Pali-Aike, Selverstone and Sternl983, Stern et al. l989). Plateboundaries, plate convergencerates and ages of the oceaniclithosphere (in Ma) are fromForsyth (l975) and Cande andLeslie (l986)
265
ducted slab and mantle wedge below the AVZ cannot be deter-mined.
The southernmost Andes consist of Late Paleozoic to EarlyMesozoic metamorphic basement, Mesozoic and Cenozoic mafic(Rocas Verdes ophiolites), intermediate (Patagonian batholith)and silicic (Tobifera volcanics) igneous rocks, as well as sedimen-tary rocks deformed and uplifted beginning in the Middle to LateCretaceous (Bruhn and Dalziel l977). In the region of the AVZ, nosystematic north-to-south change is known to occur in either theage of the metamorphic basement or the proportion of basementand younger igneous rocks. The southernmost volcano of the AVZ,located on Cook Island, directly overlies plutons of the Patagonianbatholith, and Burney, Reclus and Aguilera volcanoes occur alongthe eastern margin of the batholith. The two northernmost volca-noes in the AVZ, Lautaro and Viedma, occur within the area of thePatagonian ice cap and the details of their bedrock geology isunknown.
Base elevations of AVZ volcanoes are less than a few hundredmeters above sea level, and summit elevations of these volcanoesand other peaks of the southernmost Andes are generally less than3,000 m. This suggests crustal thickness of,35 km, similar tobelow the southern Andes everywhere south of 368S (Ramos andKay l992; Kay et al. l993). Xenoliths of lower crustal origin, foundin the Patagonian plateau basalts both along (CF, Fig. 1; Kilianl995) and east of (PA, Fig. 1; Selverstone and Stern l983) theeastern margin of the Austral Andes, are pyroxene1 plagioclasegranulites without garnet, consistent with a relatively thin crust.
The Chile Rise was subducted below the Austral Andes duringthe late Miocene, resulting in the development of a "slab-window"and extensive basaltic plateau volcanism in southern Patagoniaprior to the reinitiation of subduction and AVZ arc magmatism(Ramos and Kay l992). Ridge subduction may have also causedsubduction erosion along the continental margin. As noted byCande and Leslie (l986), the western edge of the Patagonianbatholith lies much closer to the trench south compared to north ofthe Chile Rise-Trench triple junction. Bruhn and Dalziel (l977)also commented on the very narrow arc-trench gap west of thesouthernmost Patagonian batholith, as well as the fact that thebatholith, including plutons as young as mid-Miocene, has beendeeply exposed, yet no post-Paleozoic accretionary belt of anytype has been formed. These authors have suggested that bothsubduction of significant amounts of terrigenous sediment andyorsubduction erosion of the continental margin must be invoked toexplain the enormous amount of missing continental crust in thisregion. Some of this subduction erosion may have occurred sincethe formation of the mid-Miocene plutonic belt, because plutons inthis belt south of the Chile Rise-Trench triple junction are locatedcloser to the axis of the present trench than those north of the triplejunction. Cande and Leslie (l986) suggest that active subductionerosion is occurring in conjuction with the subduction of the ChileRise at 468S, and presumably also occurred west of the AVZ as thelocus of subduction of this spreading ridge migrated northwardsduring the Miocene.
Cook Island volcano consists of a group of nested post-glacialdomes (Puig et al. l984; Heusser et al. l990). All the other AVZvolcanic centers are glaciated composite stratovolcanoes, forwhich tephra records implyHolocene activity (Stern l990; Kilianet al. l991). Lautaro is the only center with a well documentedhistoric eruption, which occurred in l959 (Martinic l988). OnlyCook Island and Burney have been mapped in detail, largely be-cause of the logistic problems involved. However, with the excep-tion of Viedma and Lautaro, which are the least accessible, thesamples analyzed for this paper were selected from collectionsmade on multiple field excursions conducted independently by theco-authors. For this reason, two of the most significant observa-tions concerning the AVZ, the lack of rocks more mafic than ande-sites and the near uniformity of petrochemistry observed withinindividual centers compared to inter-volcano petrochemical varia-tions, are believed to be inherent aspects of these centers, notartifacts of the extent of field sampling.
Analytical techniques
Mineral chemistry was determined with a Cameca Camebax SX-50 electron microprobe (Heidelberg University) using a 15 kVaccelerating voltage, a 10 nA beam current, a beam width of 1mmand a counting time of 20 s (15 s for Na and K).
Whole-rock major element compositions were determined bystandard wet-chemical techniques at Skyline Labs (Denver, Colo-rado). Trace-elements Rb, Sr, Y, Zr and Nb were determined usinga Kevex 0700 EDS system (University of Colorado). Based onrepeated analysis of standard BHVO-1, precision is estimated atbetter than 5% for Rb, Sr, Y and Zr and better than 10% for Nb.Samples were also analyzed by INAA at both the Oregon StateUniversity Radiation center and Karlsruhe University. Precision isbetter than 2% for Sc, Ta, La, Sm and Eu, 2–5% for Cr, Ni, Zn, Hf,Th, Ce, Nd, Tb, Yb and Lu, and 10% for Cs, Ba and U.
Sr- and Nd-isotopic compositions, and Rb, Sr, Nd and Sm con-centrations, were determined by mass-spectrometry at the UnitedStates Geological Survey (Denver, Colorado). Techniques are de-scribed in Peng et al. (l986). Precision is 1% for Rb, Sr, Nd and Smcontents.87Sry86Sr ratios were normalized to86Sry88Sr 5 0.1194.87Sry86Sr determined for the Eimer and Amand SrCO3 standardwas 0.70803+ 0.00005. Precision for87Sry86Sr is 0.01%.143Ndy144Nd ratios were normalized to144Ndy146Nd 5 0.7219.143Ndy144Nd for U.S.G.S. standard BCR-1 was 0.512643+ 0.000011.Precision for143Ndy144Nd is 0.005%. No age correction was ap-plied to the measured Sr- and Nd-isotopic ratios.
Pb-isotopic compositions were determined at Dartmouth Col-lege by B. Barreiro (Barreiro l983). O-isotopic compositions weredetermined at the University of Alberta by K. Muehlenbachs(Muehlenbachs and Clayton l972).
Petrography
The essential mineralogy of all the AVZ volcanoes involves pyrox-enes, amphibole and plagioclase (Fig. 2). Both a greater propor-tion of crustal xenoliths and larger more complexly zoned phe-nocrysts are observed from south-to-north, suggesting increasingimportance of near surface magma chamber processes such asslow cooling and crystallization, magma mixing, wall-rock assim-ilation, and volatile degassing. The petrographic and mineralchemistry variations for each individual center, from south-to-north, are detailed in the following sub-sections.
Cook Island
Andesites of Cook Island have subhedral and euhedral phenocrysts(typically from 0.2–2.5 mm, some as large as 5 mm) of amphibole(6–12 vol.%) and clinopyroxene (3–8 vol.%). Clinopyroxene phe-nocrysts have broad cores with high Mg' [Mgy(Mg 1 Fe21) 3 1005 88–94; Fig. 2], Cr2O3 (0.9 to 1.9 wt%) andCry(Cr1Al) (Fig. 3), similar to the cores of pyroxenes in Aleutianadakites (Yogodzinski et al. l995). These high Mg' clinopyroxe-nes have narrow rims exhibiting strong normal zonation (Mg' 588–72; Fig. 2), or are enclosed in amphibole. Comparision withclinopyroxenes in garnet- and spinel-peridotite, olivine-webster-ite and lower-crustal pyroxene-plagioclase-granulite xenolithsfound in Patagonian plateau basalts (Fig. 3) suggests that thesehigh Mg' clinopyroxene cores may be xenocrysts derived fromthe mantle. However, their Al2O3 (1.8 to 3.9 wt%) and Na2O (0.4to 0.7 wt%) contents are similar only to clinopyroxenes in infertileharzburgite xenoliths, but lower than those in fertile lherzolitexenoliths (Stern et al. l989). Amphibole phenocrysts also displaynormal zonation patterns, but with lower Mg' ranges (Mg' 586–75). The cores of amphiboles are often corroded and rims arepartly oxidized due to loss of water during magma uprise andextrusion. Cook Island andesites do not contain either plagioclase,orthopyroxene or olivine phenocrysts.
266
Fig. 2 Summary of chemicalzonations (Mg' for maficminerals; An content forfeldspars) of crystals in AVZadakites. Scale bars indicateprofile lengths (in mm)
dral with irregular zoning in the range An45–70 (Fig. 2). Highanorthite zones are partly corroded. Euhedral orthopyroxenes(Mg' 5 61–73; 2–5 vol.%) are less abundant, while partiallycorroded clinopyroxene (Mg' 5 73–78; 1–2 vol.%) phenocrystshave both normal and reversly zoned rims. Some clinopyroxeneshave cores with high Mg' (Mg'.90; Fig. 2) and Cry(Cr 1 Al)(Fig. 3), similar to the cores of Cook Island clinopyroxenes. Cor-roded olivine grains (Fo91; Fig. 2), with NiO 5 0.3 to 0.5 wt%,were also observed associated with high Mg' clinopyroxenes insmall aggregates. Amphibole also occurs in Burney andesites anddacites, but only rarely.
The groundmass of Burney rocks contain variable amounts ofplagioclase (An60–35) and orthopyroxene (Mg'#60) microlites,opaques, and glass of dacitic to rhyolitic composition. Small(,4 mm) xenoliths of upper crustal granitoids and greenschists arepresent (,2 vol.%).
Plagioclase and clinopyroxene microlites occur in the ground-mass of Cook Island andesites. Both are characterized by normalzonation. Plagioclase microlites evolve from An68 to An35 andclinopyroxene from Mg' 5 75 to 68 (Fig. 2). Very small grains ofgroundmass plagioclase reach An25. The groundmass also con-tains Fe-Ti oxides, but no orthopyroxene or olivine. Daciticgroundmass glass has enhanced Na2O contents (up to 6.5 wt%).Cook Island andesites have rare felsic xenoliths and corrodedxenocrysts (,1 vol.%) consisting mainly of plagioclase (An45).
Burney
Phenocrysts of Burney andesites and dacites include plagioclase(0.3–2.5 mm), orthopyroxene (0.5–1.3 mm) and clinopyroxene(0.3–1.1 mm). Plagioclase phenocrysts (15–20 vol.%) are euhe-
267
4.5 wt%. Granitic and granodioritic xenoliths are common (2–4 vol.%).
Geochemistry
All samples from the AVZ are andesites and dacites(Table 1; Figs. 4, 5), which in common with typical oro-genic andesites and dacites have high Al and low Ti.However, they have relatively high Mg' compared tomore typical arc andesites and dacites, including thosefrom the Andean SVZ north of the Chile Rise-Trenchtriple junction (Fig. 4; Stern et al. l984; Futa and Sternl988). Another significant feature of the AVZ centers isthe lack of basalts or basaltic andesites, which are thedominant rock type erupted from the volcanic centers inthe Andean SVZ.
All the AVZ andesites and dacites have chemical char-acteristics which would identify them as adakites. Asoutlined by Defant and Drummond (l990, l993) thesecharacteristics include: SiO2.56 wt%, Al2O3.15 wt%,low HREE (Yb,1.9 ppm) and Y,18 ppm, highSr.400 ppm and SryY.40, and positive Sr and Euanomalies. These characteristics have been interpreted toreflect both the presence of residual garnet and absenceof plagioclase in the source region of adakites (Defantand Drummond l990; Peacock et al. l994).
AVZ andesites and dacites also have low concentra-tions of high field strength elements (HFSE), which is a
Fig. 3 Mg' versus Cry(Cr1Al) for cores of clinopyroxenes fromCook Island and Burney andesites compared to clinopyroxenesfrom mantle-derived garnet- and spinel-peridotite and olivine-websterite xenoliths and lower-crustal pyroxene-plagioclase-gran-ulite xenoliths found in the Cerro del Fraile (CF, Fig. 1; Kilianl995) and Pali-Aike (PA, Fig. 1; Selverstone and Stern l983, Sternet al. l989) basalts east of the Austral Andes
Reclus
Phenocrysts of Reclus dacites consist mainly of euhedral plagio-clase (0.3 to 6 mm; 17–25 vol.%), with variable core compositions(An42–62), and often irregular zoning with abrupt changes fromAn35 to An65 (Fig. 2). High anorthite zones (.An60) are typicallycorroded. Less abundant amphibole phenocrysts (2–3 vol.%) arestrongly corroded, and partly oxidized or recrystallized to or-thopyroxene, suggesting extensive water loss in the uppermostcrust. Euhedral and subhedral orthopyroxene (0.2–0.6 mm;,1 vol.%; Mg' 5 65–55) may be formed during late-stage crys-tallization.
The groundmass consists of variable proportions of plagioclasegrains and microlites of orthopyroxene, sometimes hornblende,and glass. Groundmass glass has dacitic to rhyolitic composition,with variable K content produced by either local heterogeneitiesduring crystallization or magma mixing. Xenoliths and xenocrystsare more common in Reclus dacites than in those of Burney. Theyconsist mainly of corroded quartz and plagioclase grains (up to3 mm; 2–4 vol.%).
Northern AVZ volcanoes
Dacites of the three northernmost AVZ volcanoes (NAVZ; Fig. 1)display certain petrographic characteristics in common. Subhe-dral plagioclase (0.2–4 mm) is the most abundant phenocryst (8–12 vol.%). Zoning is very complex, including normal, reverse, andirregular zones suggesting a complex intra-crustal crystallizationhistory. For example, some plagioclase of Aguilera show a system-atic inverse zoning from An40 to An75 (Fig. 2), whereas some atViedma have a normal zonation from An75 to An40. Subhedralamphibole phenocrysts (0.3–3 mm) are characteristic of all NAVZdacites, but less abundant (1–3 vol.%) than plagioclase. Amphi-bole is usually corroded andyor rimmed by orthopyroxene. Biotitealso occurs in these volcanoes (,1%). Small euhedral orthopyrox-ene phenocrysts (Mg' 5 65) are common. In addition, in thedacites of Aguilera, clinopyroxene occurs as microphenocryst(0.2–0.6 mm) characterized by both high-Mg cores (Mg' 5 72)and reversly zoned rims (Mg' 5 81).
The groundmass of all the NAVZ volcanoes includes variableproportions of plagiocalse (An50–28; Or1–7), orthopyroxene, Fe-Tioxides, and rhyolitic glass with K2O contents ranging from 3.2 to
Fig. 4 Mg' versus wt% SiO2 for AVZ andesites and dacitescompared to experimentally produced high pressure partial meltsof garnet-amphibolite and eclogite (Table 2; Keleman l995),mantle melts (basalts), Quaternary volcanic rocks from the An-dean southern Southern Volcanic Zone (SSVZ, Fig. 1; Lopez-Escobar et al. l993), and adakites from Adak Island, Aleutians (A;Kay l978) and Cerro Pampa, Argentina (CP; Kay et al. l993).Curve for mantle AFC was calculated following DePaolo (l981)with a mass-assimilationyfractionation ratio of r5 2, reflecting arelatively hot mantle wedge, and 80% amphibole1 20% clino-pyroxene as fractionating phases
268
Tabl
e1
Ma
jor
and
tra
cee
lem
ent
,a
ndis
oto
pic
com
po
sitio
ns
of
sam
ple
sfr
om
the
volc
an
oe
sin
the
And
ea
nA
ust
ralv
olca
nic
zon
e
Sa
mp
leL
au
taro
Vie
dm
aA
gu
ilera
La
-1L
a-2
La
-3L
a-4
La
u-5
La
-N1
Vd
-1V
d-3
Ag
-1A
g-2
Ag
-5A
g-7
SiO
26
6.7
86
7.0
26
6.9
16
6.5
66
6.3
06
6.6
06
4.6
36
5.9
46
3.9
96
4.3
66
2.3
56
2.0
4T
iO2
0.5
90
.51
0.5
90
.52
0.5
90
.54
0.6
30
.60
0.6
80
.60
0.7
90
.79
Al 2
O3
16
.61
17
.24
16
.11
17
.04
16
.09
16
.71
16
.91
16
.76
17
.16
16
.54
16
.61
16
.59
Fe
O3
.53
3.5
03
.87
3.4
53
.82
3.6
04
.21
3.9
44
.51
4.0
15
.43
5.4
6M
nO
0.0
70
.09
0.0
70
.09
0.0
70
.10
0.0
80
.07
0.0
90
.09
0.0
90
.09
Mg
O2
.05
1.9
11
.97
2.0
21
.95
2.0
22
.37
2.1
82
.79
2.0
92
.87
2.9
1C
aO
4.3
23
.77
4.1
23
.80
4.4
74
.47
4.9
74
.88
5.5
84
.93
5.7
15
.72
Na 2
O3
.76
3.2
23
.41
3.0
63
.73
3.7
03
.68
3.6
84
.22
4.2
04
.17
4.1
1K
2O
2.2
01
.98
2.1
82
.15
2.1
22
.09
1.8
41
.89
2.1
22
.04
2.0
92
.16
P 2O
50
.11
0.2
30
.13
0.1
90
.17
0.1
90
.07
0.1
20
.00
0.2
40
.23
0.2
5H
2O
0.0
00
.00
0.2
60
.00
0.0
00
.00
0.3
20
.09
0.0
00
.00
0.2
30
.23
Su
m1
00
.02
99
.47
99
.62
98
.88
99
.31
10
0.0
29
9.7
11
00
.15
10
1.1
49
9.1
01
00
.57
10
0.3
5
Cr
25
24
28
26
25
30
26
27
13
38
28
26
Ni
––
11
–2
10
14
7–
–1
91
8S
c1
01
01
01
09
11
10
11
81
31
31
3V
––
79
–7
0–
64
63
––
10
81
03
Zn
––
69
–6
5–
67
62
––
59
57
K1
82
63
16
43
71
80
97
17
84
81
76
07
17
35
01
52
74
15
68
91
75
99
16
93
51
73
50
17
93
1R
b5
96
86
46
16
35
55
25
36
05
75
76
3C
s2
.04
1.7
81
.87
1.7
12
.18
1.9
31
.85
1.9
21
.88
1.7
01
.90
1.7
7B
a5
55
57
65
43
54
95
54
56
45
04
44
85
35
44
94
47
44
4S
r5
00
49
25
27
49
64
94
50
96
00
60
84
78
47
05
11
46
3
Ta0
.67
0.6
90
.70
0.6
21
.02
0.6
90
.59
0.6
40
.71
0.6
60
.87
0.7
4N
b1
08
98
88
67
51
01
1H
f3
.94
.04
.13
.73
.74
.04
.04
.03
.64
.14
.14
.6Z
r1
34
13
71
56
13
21
50
14
02
00
19
11
31
14
11
49
15
0T
i3
53
73
05
73
53
73
11
73
53
03
23
73
77
73
59
74
07
73
59
74
73
64
73
6Y
13
12
12
13
11
11
16
17
88
15
16
Th
11
.71
0.8
10
.01
1.2
9.3
12
.09
.81
0.0
13
.61
1.6
10
.41
0.7
U1
.71
.51
.91
.31
.71
.71
.91
.62
.31
.91
.92
.5
La
30
.42
9.4
30
.62
9.7
28
.93
6.6
25
.82
8.0
31
.32
8.5
29
.62
9.3
Ce
60
.55
6.5
66
.65
7.7
56
.96
8.9
48
.55
2.3
57
.65
8.8
59
.76
0.1
Nd
19
.51
9.8
26
.62
0.0
20
.72
5.6
17
.12
0.2
24
.21
9.5
20
.02
1.1
Sm
3.3
3.3
3.9
3.6
3.2
4.4
3.0
4.0
3.9
3.4
4.0
4.0
Eu
1.2
1.0
0.9
1.0
0.8
1.3
1.0
1.2
1.2
1.4
1.1
1.1
Tb
0.5
70
.65
0.0
00
.56
0.0
00
.00
0.0
00
.00
0.7
20
.77
0.6
60
.47
Yb
0.9
90
.80
0.9
60
.83
0.8
31
.04
0.9
81
.00
1.1
01
.20
1.3
21
.54
Lu
0.1
70
.16
0.1
50
.16
0.1
50
.19
0.1
30
.14
0.2
20
.25
0.2
20
.19
d1
8O
8.3
8.2
–7
.0–
8.3
––
7.5
7.0
7.4
–8
7S
ry8
6S
r0
.70
54
1–
–0
.70
51
1–
0.7
05
39
0.7
05
21
–0
.70
50
00
.70
49
5–
–1
43 N
dy1
44 N
d0
.51
25
4–
––
––
0.5
12
59
––
0.5
12
69
––
20
6 Pby
20
4 Pb
18
.61
––
––
–1
8.6
3–
–1
8.6
6–
–2
07 P
by2
04 P
b1
5.5
9–
––
––
15
.58
––
15
.56
––
20
8 Pby
20
4 Pb
38
.52
––
––
–3
8.5
3–
–3
8.5
2–
–
269
Tabl
e1
(co
ntin
ue
d)
Sa
mp
leR
ecl
us
Mt.
Bu
rney
Co
ok
Isla
nd
Re
-1R
e-2
Re
-4R
e-5
MB
-14
MB
-7M
B-9
MB
-21
PM
B-1
2M
B-1
Ck-
3-1
98
Ck-
1-4
68
Ck-
3-1
97
SiO
26
5.8
26
7.0
06
6.4
06
5.9
46
3.0
76
4.1
76
4.5
36
9.4
96
3.8
16
4.1
05
9.5
66
1.3
66
1.5
1T
iO2
0.4
30
.40
0.4
00
.43
0.4
90
.43
0.4
20
.34
0.5
70
.46
0.7
80
.71
0.8
7A
l 2O
31
6.8
31
6.6
01
6.6
31
6.8
11
7.6
91
7.9
01
7.3
31
8.2
51
8.0
11
8.1
41
8.3
51
8.3
61
8.0
0F
eO
4.2
03
.88
3.9
24
.18
4.5
04
.03
3.8
53
.45
4.1
24
.36
3.6
53
.46
3.6
8M
nO
0.0
90
.09
0.0
90
.09
0.1
00
.09
0.0
90
.80
0.0
00
.09
0.0
70
.07
0.0
7M
gO
1.6
51
.49
1.4
81
.65
2.8
02
.26
2.1
82
.15
2.2
82
.20
4.4
23
.50
3.5
8C
aO
4.8
04
.43
4.5
04
.80
5.8
15
.79
5.5
05
.39
5.8
15
.62
8.5
47
.80
7.8
2N
a 2O
4.1
74
.19
4.2
24
.18
4.6
04
.48
4.2
14
.71
4.4
64
.33
4.5
14
.42
4.3
0K
2O
1.3
81
.53
1.4
71
.37
0.8
70
.95
1.0
80
.93
0.9
50
.79
0.6
00
.57
0.6
0P 2
O5
0.1
90
.17
0.1
70
.19
0.2
00
.21
0.2
00
.21
0.0
00
.03
0.2
90
.30
0.3
4H
2O
0.0
40
.11
0.0
00
.00
0.0
00
.00
0.0
00
.00
0.0
00
.06
0.0
00
.00
0.0
0
Su
m9
9.5
99
9.8
89
9.2
79
9.6
41
00
.13
10
0.3
19
9.3
91
00
.72
10
0.0
11
00
.18
10
0.7
71
00
.55
10
0.7
7
Cr
10
98
91
32
42
11
2–
11
76
63
53
Ni
––
––
30
24
21
20
–2
06
55
34
7S
c7
77
71
01
09
8–
99
99
V5
34
65
25
7–
––
––
64
76
63
53
Zn
55
52
50
46
––
––
–6
1–
––
K1
14
47
12
66
81
22
03
11
38
17
22
27
88
68
96
57
72
07
88
66
55
84
98
14
73
24
98
1R
b2
83
02
92
91
01
21
31
11
59
33
3C
s0
.40
0.5
10
.60
0.4
30
.16
0.2
40
.32
0.2
80
.23
0.2
5–
––
Ba
39
84
13
41
74
18
27
83
19
32
42
95
30
02
60
17
71
24
10
1S
r5
45
53
25
34
51
96
24
62
65
94
56
36
05
61
52
00
31
91
01
77
0
Ta0
.61
0.6
40
.54
0.5
9–
0.2
00
.21
0.4
0–
0.3
50
.61
0.6
60
.55
Nb
76
86
33
33
32
12
11
9H
f3
.73
.63
.33
.82
.52
.62
.42
.4–
2.5
3.3
3.7
3.8
Zr
14
91
54
15
11
64
74
86
83
83
–1
45
12
21
17
13
7T
i2
59
02
38
62
40
42
54
82
93
82
57
82
51
82
03
83
41
72
75
84
67
64
25
65
21
6Y
10
91
21
16
46
61
11
06
64
Th
3.1
3.9
3.6
3.6
1.4
1.7
1.5
1.0
–1
.34
.44
.14
.7U
1.4
1.0
1.5
1.9
0.9
0.7
0.3
––
0.5
1.3
1.3
1.4
La
21
.02
0.9
20
.52
0.9
10
.41
2.6
14
.01
0.5
10
.11
0.7
35
.32
7.9
30
.6C
e4
2.7
45
.14
1.9
45
.92
4.4
28
.12
8.3
22
.82
1.9
22
.07
5.7
63
.57
0.4
Nd
22
.12
2.2
20
.62
4.4
11
.61
2.3
12
.61
0.1
10
.49
.42
8.1
15
.02
7.7
Sm
3.0
2.9
3.0
3.0
2.4
2.6
2.5
2.1
2.2
2.1
4.2
4.6
5.3
Eu
0.7
0.8
0.9
0.8
1.0
0.9
0.9
0.8
0.8
0.8
1.6
1.6
1.7
Tb
0.4
00
.38
0.4
60
.36
––
––
––
0.6
40
.77
–Y
b1
.15
1.2
00
.99
1.2
41
.36
1.2
30
.99
0.8
91
.28
1.1
00
.99
0.8
51
.01
Lu
0.1
90
.11
0.1
60
.16
0.1
60
.18
0.2
00
.18
–0
.15
0.0
30
.04
0.0
4
d1
8O
––
––
6.5
7.2
6.9
6.9
6.3
–5
.75
.55
.18
7S
ry8
6S
r–
–0
.70
51
1–
0.7
04
18
0.7
04
15
0.7
04
26
0.7
04
06
––
0.7
02
80
0.7
02
68
–1
43 N
dy1
44 N
d–
–0
.51
25
7–
––
0.5
12
75
0.5
12
76
––
0.5
13
14
––
20
6 Pby
20
4 Pb
––
18
.67
––
–1
8.6
9–
––
18
.50
––
20
7 Pby
20
4 Pb
––
15
.58
––
–1
5.6
0–
––
15
.53
––
20
8 Pby
20
4 Pb
––
38
.50
––
–3
8.4
8–
––
38
.04
––
270
Fig. 5 K2O versus SiO2 for samples from the Andean AVZ. Alsoshown are low-K basalts, basaltic andesites and andesites from theAndean SSVZ (SSVZ low-K; Lopez-Escobar et al. l993), andesitesand dacites from the northern SVZ (NSVZ; Futa and Stern l988),and other adakites (A, CP; locations and data sources from Fig. 4)
Fig. 7 Spider diagrams of the concentration of trace-elements insamples from the AVZ normalized to N-MORB (Table 3; Sun andMcDonough l989). Locations and data sources for other fields(Adak Island, Cerro Pampa, low-K SSVZ,and NSVZ) are fromFigs. 4 and 5
Fig. 6 SryY versus Y for samples from the Andean AVZ. Fields for“adakitic” and “normal” andesites and dacites are from Defant andDrummond (l990). Locations and data sources for other fields (A,SSVZ low-K, NSVZ) are from Figs. 4 and 5
common feature of both adakites and typical convergentplate boundary magmas. However, other chemical char-acteristics, such as LREE content, LREEyHREE, largeion lithophile elements (LIL)yLREE, LREEyHFSE andisotopic composition, which may reflect source compo-sition rather than source mineralogy, vary significantlybetween, but not within, individual centers (Figs. 5–10).As above, we describe the individual centers from south-to-north.
Cook Island
Volcanic rocks of Cook Island are high-CaO(.7.5 wt%), moderate to high-MgO (3.3 to 4.5 wt%;Mg'.60; Fig. 4), low-K2O (,0.6 wt%; Fig. 5) andes-
ites. Cook Island andesites differ from ordinary orogenicarc andesites by having extraordinarily low Rb (,3 ppm)and Ba (100–180 ppm) and high Sr (.1,750 ppm) con-centrations. Specific adakite characteristics of Cook Is-land andesites include low Y (4–6 ppm; Fig. 6); lowHREE (Yb,1.0 ppm; Fig. 7), high Sr (.1,750 ppm), apositive Sr anomaly (Fig. 7) and high SryY (320–440;Fig. 6). Cook Island andesites also have MORB-like Sr,Nd, Pb and O isotopic ratios (Figs. 8–10) and someMORB-like trace-element ratios such as RbySr (,0.002)and BayLa (,5). However, unlike MORB, they have bothhigh LREE (Fig. 7) and LayYb (30–36), similar to thetype adakite from Adak Island (Kay l978) and recentlydescribed late Miocene adakitic dacites from Cerro Pam-pa volcano in southern Patagonia (Kay et al. l993).
271
However, even with their low HREE, LayYb ratios arelow (8–21) and not greatly different from normal oro-genic andesite, because La contents are low (Fig. 7). Al-so BayLa ratios are significantly higher than in CookIsland andesites and more similar to typical arc magmas.
Isotopically, Burney and Reclus andesites and daciteshave higher 87Sry86Sr and lower 143Ndy144Nd ratios(Fig. 8) than in Cook Island andesites, more similar toother Andean and typical arc andesites. Pb isotopic com-positions are also more similar to other Andean andesitesand suggests an important crustal or subducted sedimentcomponent (Fig. 9). Alsod18O ratios are higher than inCook Island andesites (Fig. 10).
Northern AVZ volcanoes
Dacites of the three NAVZ volcanoes have significantlyhigher K20 (Fig. 5), Rb, Cs, Th and U concentrations thaneither Cook Island, Burney or Reclus volcanoes to thesouth. However, they also have characteristics of adak-ites such as low HREE (Yb,1.3 ppm) and Y (11–15 ppm; Fig. 7), and high SryY (36–60; Fig. 6). TheirLayYb ratios are also higher than in andesites from Bur-ney and Reclus and more similar to Cook Island andes-ites (Fig. 7). BayLa is higher than in Cook Island andes-ites, but lower than in Burney and Reclus andesites anddacites because of the low La content in the latter. RbySris significantly higher and KyRb lower than in Cook Is-land, Burney or Reclus andesites and dacites.
The Sr- and Nd-isotopic ratios of these volcanoes arehigher and lower, respectively, than those to the south(Fig. 8). Pb isotopic compositions are similar to Burneyand Reclus (Fig. 9).d18O is also higher than in the moresouthern AVZ volcanoes (Fig. 10). The volcanic centersof the NAVZ have very similar major and trace-elementand isotopic chemistry to centers from the northern SVZ(NSVZ, 33–348S; Figs. 5–10).
Discussion
All the andesites and dacites erupted from AVZ volca-noes share adakitic characteristics such as low HREEand Y and high SryY, but they also exhibit significantgeochemical differences in LIL concentration, LayYb,LIL yLREE and isotopic composition. These differences,particularly the isotopic variations, suggest contribu-tions of variable proportions of different source materi-als in the generation of different AVZ adakites. As inmore typical continental arcs (Kay l980), subductedoceanic crust (MORB1 sediment), the mantle wedgeand the continental crust may have contributed to AVZmagmas. We have attempted to model the trace-elementand isotopic composition of AVZ magmas on the basis ofcontributions from these four sources. We discuss theorigin of Cook Island andesites first because their chem-istry suggests little or no participation of either subduct-ed sediment andyor continental crust.
Fig. 8 143Ndy144Nd versus87Sry86Sr for samples from the AndeanAVZ. Field for Patagonian alkali basalts is from Stern et al. (l990)and for MORB is from Sun and McDonough (l989). Location anddata sources for other fields (Adak Island, CP, SSVZ low-K,andNSVZ) are from Figs. 4 and 5. Inset compares AVZ samples tosouthern South American metamorphic basement (S-Type) andPatagonian batholith (I-Type) crust, as well as Chile Trench sedi-ments sampled in ODP Leg 141 (Behrmann and Kilian l995)
Fig. 9 Pb-isotopic compositions of samples from the volcanoes ofthe Andean AVZ. Location and data sources for other fields(MORB, Adak Island, CP, SSVZ low-K, Patagonian alkali basalts,southern South American crust, ODP Leg 141 trench sediments)are from Figs. 4, 5 and 8
Burney and Reclus
Andesites and dacites from Burney and Reclus (Haram-bour l988) volcanoes have lower CaO (,5.8 wt%), MgO(,2.8 wt%; Mg' 5 40–53; Fig. 4) and Sr (,630 ppm)and higher K2O (0.7 to 1.6 wt%; Fig. 5), Ba (230–420 ppm) and Rb (7–30 ppm) than Cook Island andes-ites. Nevertheless, they have all the characteristics ofadakites presented by Defant and Drummond (l990,l993). HREE and Y are low (Yb,1.36 ppm; Y5 4–13 ppm). Although Sr is not as high as Cook Island ande-sites, low Y results in high SryY ratios (45–160; Fig. 6).
272
Fig. 10 d180 versus87Sry86Srand versus143Ndy144Nd forsamples from volcanoes of theAndean AVZ. Locations anddata sources for other fields(MORB, low-K SSVZ, NSVZ)are from Figs. 4, 5 and 8. Theupper panels show curves formixing of partial melt ofMORB (10% partial melting)with partial melt of subductedsediment (15% partial melt).Point P, which represents 10%of sediment partial meltmixed with 90% of MORBpartial melt (Table 4), is thestarting parental melt for mod-els of AVZ adakites as dis-cussed in the text. Lower pan-els illustrate curves for mantlefollowed by crustal AFC: P1and P2 are parental melt com-position P modified by 120%and 150% mantle AFC (r51)and curves 1 and 2 are thesecompositions modified bycrustal AFC (r50.5; Table 4)
Cook Island
Among the AVZ volcanoes, and arc volcanoes from aglobal perspective, Cook Island andesites are extraordi-nary in their MORB-like isotopic composition and cer-tain MORB-like trace-element ratios. BayLa (,5) is sig-nificantly lower than typical arc magmas, for which de-hydration of subducted oceanic crust and contaminationof the sub-arc mantle by alkali and alkaline-earth-ele-ment-rich hydrous fluids is considered an important pet-rogenetic process (Kay l980). We have previously shownthat a model of 4% partial melting of MORB convertedto eclogite reproduces many of the trace-element fea-tures of the Cook Island andesites, and on this basis sug-gested direct partial melting of subducted MORB as thesource for these andesites (Stern et al. l984; Futa andStern l988). Here we address problems with other alter-natives and refine this model.
Myers and Frost (l994) proposed that isotopicallysimilar adakitic andesites from Adak Island might haveobtained their unique composition by interaction amongmore typical calc-alkaline magmas in the magmatic con-duits of the Adak volcanic center. Yogodzinski et al.(l995) also presented a model in which silicic meltsfrom the subducted slab interact with basaltic meltsin the mantle wedge to produce adakitic andesites. How-ever, in the case of Cook Island, no evidence for acomplex system involving basaltic magmas exists, asthe volcano consists solely of adakitic andesites. Nobasalts occur here or anywhere else in the AndeanAVZ.
Derivation by partial melting of MORB is more rea-sonable. The low Yb and Y of Cook Island andesites,which are significantly lower than MORB, and their highSr, with both positive Sr and Eu anomalies, suggests thatpartial melting of MORB occurs with garnet, but withoutplagioclase, in the residue. An important question is;where in the crust-mantle section below the Cook Islandvolcano is this garnet-amphibolite or eclogitic MORBlocated? One possibility is in the lower crust, since MORbasalts were emplaced in the southernmost Andes duringthe formation of the Rocas Verdes back-arc basin in theCretaceous (Stern l980, l991). However, it is unlikelythat MORB in the crust has been transformed to garnet-amphibolite or eclogite, as there is no evidence forcrustal thickness.35 km below the southern Andes. Al-so, experimentally determined compositions of hydrousmelts of garnet-amphibolite and eclogite generally havelower MgO content and Mg' than Cook Island andesites(Fig. 4). If such melts were to form in the subducted slab,they could increase their Mg' by interaction with peri-dotite as they rose through the overlying mantle wedge(Keleman l990, l995; Keleman et al. l993). This is notpossible for melts formed in the lower crust.
In this convergent plate boundary environment, it ismore probable that the MORB source for the Cook Islandandesites is the subducted Antarctic plate oceanic crust.Peacock et al. (l994) have shown that because of theyoung age and slow rates of subduction below the south-ern Andes, the slab will reach higher temperatures atshallower depths than in typical subduction zones. In thecase of Cook Island, the fact that it is located well to the
273
east of the trench in the direction of convergence impliesa longer period of time for subducted oceanic crust toarrive below the volcano, and thus a higher temperaturefor the slab below this volcano. Peacock et al. (l994)identified the Andean AVZ as one of the few global arcsegments where temperatures might reach the meltingpoint of eclogite in the presence of H2O. The presence ofamphibole in the Cook Island andesites implies at least asmall amount of water in the source.
Subducted oceanic crust normally would be expectedto include altered MORB and sediment, as well as freshMORB, but based on the isotopic and trace-element data,only the latter appears to have participated in the genera-tion of the Cook Island andesites. As discussed below,subducted sediment may make a significant contributionto the adakitic magmas erupted from other AVZ volca-noes. However, Cook Island is located both further fromthe trench in the direction of plate convergence, as wellas in a region of more oblique subduction angle, thanother volcanoes in the AVZ. This may influence the ex-tent to which different materials are either subducted orunderplated. Among other adakites world-wide, thosefrom both Cerro Pampa, Argentina (Kay et al. l993), andthe western Aleutians (Yogodzinski et al. 1995), haveisotopic compositions that indicate little involvement ofsubducted sediment and altered oceanic crust, and bothalso occur either relatively far from the trench or in anarea of oblique subduction.
The REEyHFSE ratios of the Cook Island andesitesare higher than MORB and similar to typical arc mag-mas. The explanation normally invoked to explain simi-lar high REEyHFSE in more typical arc magmas, enrich-ment of the mantle source region by hydrous fluids, canbe ruled out for Cook Island andesites because of theirMORB-like BayLa ratios. High REEyHFSE ratios mayresult from refractory residual phases retaining HFSE,such as rutile, which might be expected for low degreesof partial melting of eclogitic MORB (Rapp et al. l991;Ayers and Watson l991; Keleman et al., l993). Yogodzin-ski et al. (l995) have shown that the presence of 1% rutilein an eclogitic MORB source can produce the amount ofTa depletion, relative to light-rare-earth elements, ob-served in both western Aleutian and Cook Island adak-ites. Also Rb, Ba and K are lower in Cook Island andes-ites than predicted for 4% melting of N-MORB (Stern etal. l984). This suggests a small proportion of other resid-ual phase containing these elements, such as mica, am-phibole or K-feldpar.
Although the simple eclogite-melt trace-element dis-tribution pattern and MORB-like isotopic compositionof Cook Island andesites are consistent with the deriva-tion of these magmas by partial melting of subductedMORB, their high Mg' (Fig. 4) and Ni and Cr contentsmay result from interaction of these slab-melts withmantle peridotite as they rise through the overlyingmantle wedge, as has been suggested for bajaites (Rogerset al. l985) and other adakites including the type adakitefrom Adak Island (Kay l978) and those from Cerro Pam-pa, Argentina (Kay et al. l993). Since the mantle wedge
is expected to be hotter than either the slab or any slab-melts, it is hard to imagine how such interaction wouldnot occur (Kelemen et al. l993). High Mg' and Cry(Cr 1 Al) clinopyroxene cores in Cook Island andesites(Figs. 2, 3) may be relic xenocrysts from such interac-tion. However, Cook Island adakites do not have eitherthe very high MgO, Ni and Cr observed in bajaites(Rogers et al. l985), nor olivine, suggesting that suchinteraction must have been limited. Also, Cook Islandandesites are isotopically distinct from all the Patagoni-an plateau alkali basalts and included mantle xenolithswhich are considered representative of the southernSouth American lithosphere (Figs. 8, 9). Although thelow Sr, Nd and Pb concentration of mantle lithospherewould make limited amounts of mantle contaminationdifficult to detect, large amounts of such contaminationcan be ruled out by these isotopic considerations.
Different possible processes have been proposed forthe interaction of hydrous slab-derived melts with theoverlying maltle wedge, including fluxing of diapiric up-rise and decompression melting of the mantle resultingin mixing of slab and mantle derived melts (Yogodzinskiet al. l995), and slab-meltyperidotite reaction and hy-bridization (Keleman 1990, l995; Keleman et al. l993).We prefer slab-meltyperidotite reaction to explain thehigh Mg' of Cook Island andesites, since there is nodirect evidence for mantle-derived basaltic magmas inthe AVZ. Physically slab-meltyperidotite interaction in-volves the reaction of mantle olivine and pyroxenes withrelatively silica-rich hydrous melts to produce new min-erals (amphibole, pyroxenes, olivine andyor garnet) andhybridized melts (Sekine and Wyllie 1983; Caroll andWyllie l989; Keleman l990, l995). Keleman (l990, 1995)has suggested that heating of hydrous slab-derived meltsascending through the mantle wedge, due either to theinverted temperature gradient in the wedge (Davis andStevenson l992) or to heating resulting from isenthalpicascent (Baker et al. 1994), provides substantial thermalenergy for peridotite dissolution. He concluded that aslab-melt formed at 9008C and heated to 1,2008C has thethermal energy to dissolve an equal weight of peridotite.We have calculated a simple AFC model for a hypotheti-cal slab-melt composition equal to the average of all ex-perimentally determined high-pressure (1–3 GPa), hy-drous partial melts of basalts (Fig. 4). This model in-volves bulk mantle assimilation combined with crystal-lization of amphibole and clinopyroxenes, the phe-nocrysts ocurring in Cook Island andesites, with mantlemass-assimilation greater (r5 2; DePaolo l981) than theamount of fractionation, reflecting a relatively hot mantle(Keleman l990, l995). This model suggests that dissolu-tion of only between 10–20% peridotite is required toraise the Mg' of a slab-melt to that of the Cook Islandandesites (Fig. 4).
Cook Island adakites, as well as many other adakitesworld-wide, have high CaO contents and CaOyNa2O ra-tios compared to typical calc-alkaline andesites. Theseresemble experimentally determined slab-melts oftonalitic (CaOyNa2O.1) rather than trondhjemitic
274Ta
ble
2C
alc
ula
ted
pa
rent
alm
elt
sw
hic
hca
nyi
eld
Co
ok
Isla
nda
nde
site
sb
ya
ssim
ilatio
no
f1
0%
ma
ntle
-ma
ssco
mb
ine
dw
ith
fra
ctio
na
tion
of
am
ph
ibol
ea
ndcl
ino
pyr
oxe
ne
,a
ndp
ubl
ish
ed
hig
hp
ress
ure
exp
eri
me
nta
lpa
rtia
lme
lto
fb
asa
lts
Co
ok
Pri
mit
ive
Fra
ctio
na
ted
min
era
ls:
CO
OK
CO
OK
CO
OK
Tro
ndh
jem
itic
Ton
alit
icM
ant
leP
are
nt1
Pa
rent
2P
are
nt3
Cp
xH
blS
um
Ma
ntle
ass
imila
tion51
0%
9'1
0'1
1'1
2'1
3'1
4'1
5'2
0%
80
%1
00
%r5
1r5
2r5
8
SiO
26
0.2
34
5.9
65
1.0
24
5.0
34
6.2
36
0.2
66
1.0
06
1.6
06
7.6
67
0.1
06
5.9
36
2.4
76
2.3
06
3.1
16
1.7
0T
iO2
0.7
80
.18
0.6
32
.08
1.7
90
.94
0.9
00
.86
0.6
10
.94
1.3
10
.42
–0
.73
0.8
0A
l 2O
31
8.0
74
.06
4.7
01
1.6
11
0.2
31
8.6
91
9.1
31
9.5
01
7.4
51
5.5
41
7.5
52
0.9
32
2.7
01
6.5
01
9.5
0F
eO
3.5
67
.54
5.8
99
.06
8.4
33
.65
3.4
03
.19
1.8
91
.78
3.1
54
.93
2.3
05
.26
5.6
0M
nO
0.0
70
.20
0.1
20
.07
0.0
80
.06
0.0
60
.06
–0
.07
0.1
20
.10
–0
.10
0.0
6M
gO
4.0
53
7.7
81
4.4
21
5.7
41
5.4
81
.82
1.1
00
.51
0.6
20
.43
1.0
21
.33
0.2
02
.48
1.5
0C
aO
7.9
73
.21
22
.48
11
.29
13
.53
9.0
08
.76
8.5
71
.61
1.8
91
.47
8.5
37
.90
5.7
26
.40
Na 2
O4
.37
0.3
30
.73
2.9
12
.47
4.5
84
.70
4.7
94
.89
5.4
46
.72
1.1
33
.60
3.4
23
.40
K2O
0.5
80
.00
0.0
00
.13
0.1
00
.59
0.6
20
.64
5.2
93
.82
2.5
50
.28
0.7
0–
1.3
0P 2
O5
0.3
10
.00
0.0
00
.00
0.0
00
.31
0.3
30
.34
––
––
––
–
Tota
l1
00
.00
99
.26
10
0.0
09
7.9
19
8.3
31
00
.00
10
0.0
01
00
.00
10
0.0
21
00
.01
99
.82
10
0.1
29
9.7
09
7.3
21
00
.26
Mg'
(Fe
tot)
67
.08
9.9
81
.47
5.6
Bu
lkD
47
.13
6.6
21
.83
6.9
30
.13
6.6
32
.51
3.4
45
.73
2.3
Rb
3.0
00
.54
0.0
80
0.1
00
0.0
96
2.9
83
.11
3.2
19
5S
en
&D
un
n(1
99
4);
2G
Pa
,ru
nB
16
Ba
17
76
0.0
60
0.0
70
0.0
68
17
71
86
19
21
05
Ra
pp
eta
l.(1
99
1);
3.2
GP
a,
Ro
ck'2
,1
07
58C
Th
4.4
00
.08
0.0
50
0.0
50
0.0
50
4.4
14
.60
4.7
81
15
Ra
pp
eta
l.(1
99
1);
3.2
GP
a,
Ro
ck'1
,1
10
08C
U1
.30
0.0
20
.01
50
.04
00
.03
51
.30
1.3
71
.41
12
5W
olf
and
Wyl
lie(1
99
4);
1G
Pa
,1
38
SD
:3K
49
81
25
80
.02
00
.10
00
.08
44
99
75
22
65
39
61
35
Ste
rna
ndW
yllie
(19
78
);3
GP
a,
5%
H2O
,9
008
CTa
0.6
10
.04
0.2
40
0.2
10
0.2
16
0.6
20
.64
0.6
51
45
Yog
od
zin
skie
tal.
(19
95
);sm
lt2
La
35
.30
0.6
14
0.6
00
0.0
50
0.1
60
35
.80
37
.30
38
.41
55
Joh
an
ne
sa
ndW
olke
(19
94
);1
GP
a,
90
08
CC
e7
5.7
01
.60
10
.90
00
.06
00
.22
87
7.3
08
0.2
08
2.4
Pri
mit
ive
ma
ntle
5H
ofm
an
n(1
98
8)
Nd
28
.10
1.1
89
0.4
00
1.2
00
1.0
40
31
.00
30
.90
30
.8D
Cp
xa
ndD H
blfr
om
Bre
nn
an
eta
l.(1
99
4),
Sr
20
03
18
0.1
00
0.2
00
0.1
80
20
38
21
20
21
81
Joh
nso
n(1
99
4),
Ma
rtin
(19
93
),S
en
and
Du
nn
Zr
12
21
00
.30
00
.25
00
.26
01
24
12
91
32
(19
94
)a
ndS
chn
etg
er
(19
93
)Y
6.0
03
.90
0.8
00
2.5
00
2.1
60
7.0
16
.58
6.3
Yb
0.9
90
0.4
14
0.9
00
2.2
00
1.9
40
1.1
60
1.1
00
1.0
6L
u0
.04
00
.06
40
.90
03
.00
02
.58
00
.04
00
.04
20
.04
2C
r6
51
20
03
0.0
00
30
.00
03
0.0
00
55
02
00
87S
ry8
6S
r0
.70
28
00
.70
36
0.7
02
80
0.7
02
80
0.7
02
80
14
3 Ndy
14
4 Nd
0.5
13
15
0.5
12
80
.51
31
50
.51
31
50
.51
31
5d
18O
5.5
6.5
5.5
5.5
5.4
51
00y
Nd
3.5
68
4.0
93
.23
3.2
43
.25
10
00y
Sr
0.5
05
4.9
50
.49
0.4
70
.46
Sry
Y3
34
4.7
29
13
22
34
6
275
(CaOyNa2O,1) composition (Table 2). Yogodzinski etal. (l995) concluded that transforming low CaOyNa2Otrondhjemitic slab-melts into high CaOyNa2O adakiticandesites would require extensive assimilation of mantleperidotite (roughly twice the mass of the original melt toget enough CaO from garnet and clinopyroxene into themelt), followed by equally extensive fractionation ofolivine and orthopyroxene (roughly 50% of the mass ofthis mixture to get the MgO and FeO out). Both the ab-sence of either olivine or orthopyroxene in the Cook Is-land andesites, and the isotopic differences betweenthese andesites and samples of the mantle lithospherebelow Southern South America, rule out such extensiveinteraction. Also, such extensive assimilation and frac-tionation would not preserve the simple eclogite-melttrace-element distribution pattern observed in the CookIsland andesites.
If the original slab-melt has higher CaOyNa2O, thensignificantly less mantle assimilation is required to pro-duce an adakitic andesite. For instance, Yogodzinski etal. (1995) proposed 26% assimilation of mantle peri-dotite, followed by 34% orthopyroxene fractionation, toproduce the major-element composition of westernAleutians adakitic andesites from a tonalitic slab-melt.However, only clinopyroxene and amphibole phe-nocrysts are observed in Cook Island andesites. We haveback-calculated possible parental slab-melts for CookIsland andesites with AFC models (DePaolo l981) in-volving fractionation of these phases in their observedproportions (80% amphibole and 20% clinopyroxene),and variable proportions of mantle assimilation rangingfrom equal (r5 1) to greater (r5 2 and r5 8) than theamount of fractionation, reflecting a relatively hot mantlewedge (Keleman l990, l995). Our calculated parentalslab-melts (Table 2), from which Cook Island andesitescan be derived by assimilation of 10% mantle peridotite,are high CaOyNa2O magmas with between 2.0 and0.5 wt% MgO; reasonably similar to suggested tonaliticslab-melts. With up to 10% assimilation of mantle, thetrace-element concentrations calculated for theseparental magmas of Cook Island andesites (Table 2) arestill consistent with models of derivation by 2–3% partialmelting of eclogitic MORB. Greater than 10% mantlemass assimilation would produce parental magmas re-quiring #2% MORB melting, which is below the lowerlimit for the segregation of a magma suggested by Wyllieand Wolf (l993).
In summary, we propose a simple slab-melt origin forCook Island andesite, followed by limited (#10% massassimilation) interaction in the overlying mantle. Rapidrise through both the mantle and the upper crust, perhapsalong pathways related to the strike-slip plate boundarybetween the South American and Scotia plates (Fig. 1), issuggested by this limited extent of interaction with themantle, and by the lack of plagioclase phenocrysts inthese andesites. Also their MORB-like O- and Pb-iso-topic compositions, combined with their low Pb concen-tration, preclude crustal contamination.
Other AVZ volcanoes
The other volcanoes of the AVZ do not have the distinc-tive MORB-like isotopic signature of Cook Island. How-ever, they all have low HREE and Y and high SryY whichsuggests garnet in their source, and like Cook Island theyoccur in a region of relatively thin crust (,35 km) withinwhich plagioclase rather than garnet is likely to be stable.If the thermal conditions for melting of subductedMORB occur below Cook Island, it seems likely that theymay also occur below the other AVZ volcanoes, underwhich even younger oceanic crust is being subducted(Fig. 1). However, it is clear that other material such ascontinental crust, subducted sediment, andyor the mantlewedge must have participated in the generation of mag-mas erupted from these volcanoes to create the observedtrace-element ratios (Fig. 7) and isotopic (Figs. 8–10)variations. We discuss below models for the generationof the other AVZ volcanoes involving these other materi-als, first individually, and then together.
From south to north, the volcanoes of the AVZ definea trend of decreasing MgO (Fig. 4) and Sr content and143Ndy144Nd ratios (Fig. 8), and increasing SiO2, K2O(Fig. 5), 87Sry86Sr (Fig. 8) andd18O (Fig. 10). Theserelations, as well as the petrologic features of increasedmodal phenocryst and xenocryst proportions, suggest anorthward increase in the importance of combined frac-tional crystallization and crustal assimilation (AFC) inshallow level magma chambers (Stern et al. l984). How-ever, the relatively high Mg' (Fig. 4) and Sr, as well as lowY b and Y (Fig. 6) of all the AVZ volcanoes rule out typicalAndean basalts as possible parental magmas (Stern et al.l984). A crustal AFC model involving a parental magmasimilar to the Cook Island andesite, with r5 0.5, canproduce the observed Sr versus87Sry86Sr relations by30% fractionation for Burney and 50% for the NAVZ.However, this model fails to produce many other signifi-cant geochemical characteristics of these AVZ centerscompared to Cook Island, in particular the low LIL andLREE of Burney adakites. Crustal AFC may be impor-tant in the generation of all the AVZ centers north ofCook Island, but can not be the only processes involvedin creating the variations observed among these centers.
Subduction of both pelagic and terrigenous sediment,andyor subduction erosion of continental crust, are alsolikely processes by which greater proportions of thesehigh87Sry86Sr andd18O materials might have been incor-porated in the AVZ magmas erupted from the volcanoesnorth of Cook Island. Also, AVZ adakites have Pb-iso-topic compositions which occur along a mixing line be-tween MORB and Chile Trench sediment (Fig. 9). How-ever a model involving mixing of MORB melts withmelts derived from subducted sediments (Figs. 10, 11),cannot by itself produce the isotopic variations observedamong the AVZ magmas. Specifically, the amount ofsubducted sediment required to produce the87Sry86Sr ofBurney and NAVZ andesites and dacites (Fig. 11a) ismuch less than that required to produce theird18O(Fig. 10) and143Ndy144Nd (Fig. 11b) values.
276
Fig. 11a, b 1000ySr versus87Sry86Sr and 100yNd versus143Ndy144Nd for slab-melts followed by mantle and crustal AFC process-es as described in the text. Trends for mantle AFC with r5 0.75(arrow to high Sr) and r5 8 (arrow to low Sr) are illustrated for1000ySr versus87Sry86Sr
models are listed in Table 3. The subducted sediment isfrom ODP Leg 141 in the Chile Trench (Behrmann andKilian l995), and the continental crust is a mixture ofaverage southern South American metamorphic base-ment (S-type) and plutonic rocks (I-type; Behrmann andKilian l995). In order to model the melting of subductedMORB, we calculated a bulk distribution coefficientDMORB (Table 3) by assuming that the Cook Islandparental magma'2 (Table 2) was generated by 2.5%melting of N-MORB. In order to model melting of sub-ducted sediment, we caluclated the bulk distribution co-efficient DSedimentby adding 20% biotite to DMORB. Bulkdistribution coefficients for the mineral assemblages in-volved in mantle and crustal AFC processes were calcu-lated based on published distribution coefficients forthese minerals.
We presume that if MORB melts, so do subductedsediments (Nichols et al. l994). We varied the percentageof partial melting of MORB and subducted sediment in-dependently, and mixed the resulting melts together indifferent proportions, to produce one best parental com-position (P; Figs. 10–13; Table 4), which can yield bothBurney and NAVZ magmas by subsequent mantle andcrustal AFC. Parental compositionP corresponds to amixture of 10% sediment partial melt (15% partial melt-ing) with 90% MORB partial melt (10% partial melt-ing). Lower degrees of partial melting of either sedimentor MORB resulted in melts with unrealistically high in-compatible element concentrations, while higher degreesof partial melting of sediment decreased Sr concentra-tions so that the resulting mixtures with MORB meltshave unacceptably highd18O at relatively low87Sry86Sr.Mixing higher percentages of sediment partial melts
The genesis of most typical arc magmas is generallyattributed to melting of the mantle wedge fluxed by addi-tion of small (,2%) amounts of slab-derived fluids(Pearce and Parkinson l993). Slab-melts rising into thehot mantle wedge are also likely to interact with thismantle peridotite. Interaction of slab-melts with mantleperidotite may lower incompatible LIL and LRE elementconcentrations relative to more compatible trace-ele-ments (Keleman et al. l993). Although variations in theextent of interaction with the mantle wedge alone cannotexplain the isotopic variations observed among the AVZadakites, interaction of slab-derived melts within themantle wedge could potentially produce magmas withthe low LIL and LREE of Burney adakites. MgO-richolivine and clinopyroxene xenocrysts in Burney andes-ites (Figs. 2, 3) might be relics of such interaction, al-though the general lack of olivine in all AVZ adakitesindicates that such interactions were limited, and as forCook Island andesites, other AVZ adakites have not equi-librated with mantle peridotite.
We have attempted to model Burney and NAVZ adak-ites by multi-stage processes of slab (MORB1 sedi-ment) melting, followed by interaction of these meltswith the mantle wedge, and subsequently crustal AFC(Figs. 10–13; Table 4). The composition of N-MORB,primitive mantle, subducted sediment and continentalcrust, and the distribution coefficients used in these
277
Fig. 12 SryY versus Y forslab-melts followed by mantleand crustal AFC processes asdescribed in the text
with MORB melts results in compositions with87Sry86Srequal to or greater than Burney (Figs. 10, 11), whichprecludes subsequent crustal AFC, but at the same timedoes not produce the required low143Ndy144Nd of Bur-ney magmas. CompositionP is essentially the maximumamount of sediment that may be incorporated in theparental slab-melt and still allow a specific amount ofsubsequent crustal AFC to produce both the87Sry86Srand143Ndy144Nd of the Burney magmas. With this sameparental slab-melt composition, NAVZ magmas can bemodeled with greater amounts of crustal AFC. Alterna-tively, more subducted sediment could be incorportatedin the slab-melt parental magmas for NAVZ volcanoes,but the model presented here attempts to explain bothBurney and NAVZ magmas from similar parental slab-melt.
The resultant parental slab-meltP has both higher Srand higher Nd concentrations than either Burney orNAVZ magmas (Fig. 11; Table 4). Although Sr decreases,Nd concentration increases during crustal AFC. For thisreason, interaction of slab-melts with mantle peridotitemust be an important process below these AVZ volcanoesin order to lower Nd concentrations of these slab-meltsprior to crustal AFC. We have modeled this interactionassuming mantle assimilation combined with fractionalcrystallization of the evolving magmas. We calculatedmantle AFC models with variable mass-assimilationyfractionation ratios (r5 0.75, r5 1 and r5 8), using themafic phenocryst phases observed in AVZ magmas (am-phibole, clinopyroxene, orthopyroxene and Ti-mag-netite) as fractionating phases. All these different r-val-ues result in decreasing Nd with increasing mantle mass-assimilation. However, high mass-assimilationyfraction-ation ratios (r5 8 as well as bulk mantle assimilation),while physically possible in a hot mantle wedge (Kele-man et al. l993), result as well in an unsuitable decreaseof Sr (Fig. 11a) and other LIL element concentrations. Incontrast, low mass-assimilationyfractionation ratios(r 5 0.75) result in an unsuitable increases in Sr(Fig. 11a) and other LIL element concentrations. Forr 5 1, Sr and LIL concentrations remain nearly constantas Nd, Y and HREE decrease (Figs. 11–13).
For mantle AFC with r51, 120% and 150% mantlemass-assimilation by parental slab-meltP produce com-
Fig. 13 Spider diagrams showing multi-element fit for the bestmodel of slab melting followed by mantle and crustal AFC asdescibed in the text
positionsP1 andP2, respectively (Figs. 10–13; Table 4).From compositionP1, NAVZ magmas may be modeledby 20–30% crustal AFC, using the commonly employedvalue r 5 0.5 (DePaolo l981). From compositionP2,Burney magmas may be modeled by 10–15% crustalAFC. The greater amount of mantle mass-assimilationrequired to produce Burney compared to NAVZ magmasreflects the significantly lower LIL and LREE of Burneyadakites. The smaller proportion of crustal AFC required
278
Table 3 Compositions of materials and distribution coefficients used in slab melting followed by mantle and crustal AFC for adakitesa
N-MORB Primitive ODP 141 S-type I-type Crustal Dbiotite DMORB DSediment Ds for Ds forMantle sediments crust crust mixture mantle crustal
(50: 50) AFC AFC
Rb 0.56 0.54 74 101 74 87 3.00 0.17 0.74 0.083 0.032Ba 6.3 6 445 482 489 485 5.60 0.01 1.13 0.060 0.111Th 0.12 0.08 6 8.4 7.2 7.8 2.00 0.002 0.40 0.053 0.020U 0.047 0.02 1.8 1.6 2 1.75 2.00 0.01 0.41 0.042 0.020K 600 258 17000 13695 21266 17481 5.00 0.10 1.08 0.072 0.061Ta 0.132 0.035 0.6 1 1.1 1.05 1.60 0.20 0.48 0.243 0.010La 2.5 0.6 19 24.6 24 24.3 0.30 0.05 0.10 0.164 0.143Ce 7.5 1.6 49 52.9 45.4 49.2 0.34 0.07 0.12 0.253 0.141Nd 7.3 1.2 21 28.8 18.6 23.6 0.32 0.22 0.24 0.910 0.140Sr 90 18 305 173 267 220 0.30 0.02 0.08 0.156 1.025Zr 74 10 149 180 160 170 0.08 0.60 0.50 0.236 0.073Y 28 3.9 27 27 21 24 0.10 4.10 3.30 1.875 0.010Yb 3.05 0.41 2.4 1.96 1.94 1.95 0.10 2.70 2.18 1.682 0.295Lu 0.455 0.064 0.4 0.27 0.3 0.26 0.10 11.60 9.30 2.208 0.29587Sry86Sr 0.7025 0.7036 0.708 0.723 0.707 0.715 (66% Hbl, (50% Pl,143Ndy144Nd 0.51315 0.5128 0.5124 0.512 0.5125 0.5123 20% Cpx, 25% Cpx,d18O 5.5 6.5 10 14 10 12 10% Cpx, 25% Cpx,
4% Ti-Mt)
a N-MORB from Sun and McDonough (1989); primitive mantlefrom Hofmann (1988); DMORB (calculated as described in thetext), DSediment5(80% DMORB120% Dbiotite); Dbiotite, DHbl, DCpx,
DTi-Mt from references in Table 2; ODP Leg 141 sediments, S-type (metamorphic basement) and I-type (Patagonian Batholith)crust from Behrmann and Kilian (1995) and unpublished data
Fig. 14 Schematic cross-sections below the different volcaniccenters, from north-to-south, of the Andean AVZ. The angle ofsubduction is surmised to increase southwards from near the ChileRise-Trench triple junction to below Burney, and than decreaseagain as convergence direction becomes more oblique. The extentof crustal compression is surmised to increase northwards as theangle of subduction becomes more orthogonal. These factors af-fect the extent of mantle and crustal AFC processes as descibed inthe text
to produce the Burney magmas reflects the more primi-tive isotopic composition of these magmas.
Conclusions
Although considerable uncertainties relate to both thecomposition of the different components and the equi-librium bulk assimilation AFC processes, particularlyfor slab-meltymantle peridotite interaction, employed inour multi-stage, four-component (MORB, subductedsediment, mantle wedge, crust) model, this model gives areasonable fit to the observed trace-element and isotopiccompositions of AVZ andesites, specifically Sr, Nd andY contents and O, Sr and Nd isotopic compositions (Figs.10–13; Table 4). The mass contribution from the sub-ducted slab in this model is high compared to typical arcmagmas. This is consistent with geothermal modelswhich suggest melting of the young and slowly subduct-ing slab in this region of the Andean arc (Peacock et al.l994). Our model also involves a significant amount ofsubducted sediment in the generation of all the AVZadakites except those from Cook Island.
The proportions of the different source materials in-volved in the genesis of AVZ magmas vary significantlyfrom north-to-south. A basis for rationalizing these vari-
279
Tabl
e4
Ca
lcu
late
dtr
ace
-ele
me
nta
ndis
oto
pic
com
po
sitio
ns
of
the
sla
b-m
elt
pa
rent
alm
ag
ma
sa
ndth
ism
ag
ma
mo
dif
ied
firs
tb
ym
ant
lea
ndsu
bse
que
ntly
cru
sta
lAF
C(
Ma
ma
ssa
ssim
ilate
d,M
cm
ass
crys
talli
zed
)
Pa
rtia
lme
ltin
ga9
0%
MO
RB
AF
CM
od
elli
ng
Ave
rag
eA
FC
Mo
de
llin
gAv
era
ge
&1
0%
sed
i-N
AVZ
Bu
rney
MO
RB
Se
dim
ent
me
ntm
elt
Ma
ntle
Cru
stC
rust
Ma
ntle
Cru
stC
rust
10
%1
5%
PM
a51
20
%M
a52
0%
Ma5
30
%M
a51
50
%M
a51
0%
Ma5
15
%P
art
ialm
elt
Pa
rtia
lme
ltM
c51
20
%M
c54
0%
Mc5
60
%M
c51
50
%M
c52
0%
Mc5
30
%P
-1P
-2
Rb
2.2
95
.41
1.5
11
.13
4.8
51
.35
9.0
10
.92
1.6
24
.81
0.0
Ba
58
40
19
29
32
17
29
85
14
93
15
01
93
27
8T
h1
.21
2.2
2.3
2.2
4.9
6.4
10
.92
.23
.34
.01
.4U
0.4
3.6
0.8
0.7
1.3
1.8
1.8
0.7
1.0
1.2
0.9
K3
15
81
59
18
44
34
43
65
96
27
13
50
91
71
98
43
48
67
02
80
74
72
22
Ta0
.47
1.0
80
.53
0.4
30
.78
1.0
20
.70
0.4
10
.57
0.6
60
.20
La
17
.28
0.9
23
.62
0.1
27
.83
2.8
29
.81
9.3
22
.82
4.8
10
.4C
e4
6.0
19
1.9
60
.64
6.4
60
.16
8.6
56
.44
3.5
49
.85
3.2
24
.4N
d2
4.5
59
.32
8.0
10
.31
7.0
21
.52
1.0
8.1
11
.11
2.8
11
.6S
r7
65
14
23
83
17
09
54
14
66
50
06
82
60
05
60
62
4Z
r1
16
26
11
30
10
81
69
21
11
35
10
41
31
14
67
4Y
7.4
9.1
7.6
2.7
8.4
12
.11
0.9
2.4
5.1
6.6
6.0
Yb
1.2
11
.20
1.2
00
.37
0.8
51
.15
1.0
00
.32
0.5
50
.66
0.9
9L
u0
.04
0.0
50
.04
0.0
30
.10
0.1
40
.16
0.0
30
.06
0.0
80
.18
Cr
20
20
42
21
34
26
33
87S
ry8
6S
r0
.70
28
00
.70
80
00
.70
36
90
.70
36
90
.70
45
50
.70
51
50
.70
52
00
.70
36
90
.70
40
90
.70
43
30
.70
42
01
43 N
dy1
44 N
d0
.51
31
50
.51
24
00
.51
29
90
.51
29
70
.75
12
80
.51
26
80
.51
26
20
.51
29
70
.51
28
10
.51
27
60
.51
27
5d
18O
5.5
01
0.0
05
.98
5.9
97
.20
7.8
07
.70
6.0
06
.60
6.9
06
.80
Sry
Y1
04
15
61
10
26
56
43
94
62
81
11
88
51
04
10
00y
Sr
1.3
10
.70
1.2
01
.41
1.8
52
.15
2.0
01
.47
1.6
71
.78
1.6
01
00y
Nd
4.1
1.7
3.6
9.7
5.9
4.7
4.8
12
.39
.07
.88
.6
aM
elt
ing
of
MO
RB
and
sed
ime
ntw
ith
bu
lkD
sfr
om
Tabl
e3
280
ations in the tectonic context of the AVZ is provided bythe schematic cross-sections in Fig. 14. Just south of thetriple junction (46–498S), a gap in active volcanism oc-curs in the region where very young oceanic lithosphereis being subducted, presumably at very low angle. Fur-ther to the south below the NAVZ (49–518S), melting ofthe slab (MORB1 sediment) and interaction of thesemelts with a relatively small mantle wedge, followed bysignificant crustal AFC, produces the NAVZ adakites.Even further south below Reclus and Burney (51–528S),the lower LIL and LREE of magmas erupted from thesevolcanoes suggest greater interaction within the mantleof slab-derived melts. In this region the subducted litho-sphere is older and the subduction angle may be greater,so that the mantle wedge may be larger or more activelyconvecting and thus hotter. On the other hand, crustalAFC process are less significant than below the NAVZ.This may be due to the increasingly oblique nature ofsubduction as well as the influence of the strike-slipMalvinasyMagallanes fault system (Fig. 1) in providingpaths for the relatively rapid rise of sub-crustal magmasat this latitude. Even further south below Cook Island(548S), oblique subduction, fragmentation of the sub-ducted slab and deep lithospheric faults allow small pro-portions of MORB melt to rise through both the mantleand crust with only very limited interaction.
Acknowledgements We thank A. Puig and L. Dickenson forproviding some samples, and A. Skewes, E. Leonard, R. Fuenzali-da, S. Harambour and J. Lobato for logistic support in obtainingothers. K. Futa, K. Muehlenbachs, L. Farmer and B. Barreiro pro-vided some of the isotopic data, and H.Meyer washelpful inproducing the microprobe data. The final mauscript was improvedby comments from T. Dunn, T. Grove and an anonymous reviewer.This research was supported by NSF grants EAR76-03816,EAR80-20791 and EAR83-13884, and German Research Founda-tion grants Ki-345y1 and Ki-456y1.
References
Ayers JC, Watson EB (1991) Rutile solubility and mobility insupercritical aqueous fluids. Contrib Mineral Petrol 114: 321–330
Baker MB, Grove TL, Price R (l994) Primitive basalts and andes-ites from the Mt. Shasta region, N. California: products ofvarying melt fraction and water content. Contrib MineralPetrol 118: 111–129
Barreiro B (l983) Lead isotopic compositions of South SandwichIsland volcanic rocks and their bearing on magmagenesis inintra-oceanic island arcs. Geochim Cosmochim Act 47: 817–822
Behrmann J, Kilian R (l995) Chemical and isotopic constraints onthe origin of Pleistocene to Pliocene trench sediments at theChile triple junction (abstract). IUGG XXI General Assembly:A450
Brennan JM, Shaw HF, Phinney DL (l994) Experimental determi-nation of trace element partitioning between pargasitic amphi-bole and hydrous melt. Mineral Mag 58: 121–122
Bruhn RL, Dalziel IWD (1977) Destruction of the Early Creta-ceous marginal basin in the Andes of Tierra del Fuego. In:Talwani M, Pitman WC (eds) Island arcs, deep sea trenches,and back-arc basins (Maurice Ewing Series 1) AmericanGeophysical Union, Washington, pp 395–406
Cande SC, Leslie RB (1986) Late Cenozoic tectonics of the South-ern Chile Trench. J Geophys Res 91: 471–496
Carroll MR, Wyllie PJ (l989) Experimental phase relations in thesystem tonalite-peridotite-H2O at 15 kb: implications for as-similation and differentiation processes near the crust-mantleboundary. J Petrol 30: 1351–1382
Davis JH, Stevenson DJ (l992) Physical model of source region ofsubduction zone volcanics. J Geophys Res 97: 2037–2070
Defant MJ, Drummond MS (1990) Derivation of some modern arcmagmas by melting of young subducted lithosphere. Nature347: 662–665
Defant MJ, Drummond MS (1993) Mount St. Helens: potentialexample of partial melting of subducted lithosphere in a vol-canic arc. Geology 21:547–550
Defant MJ, Clark LF, Stewart RH, Drummond MS, de Boer JZ,Maury RC, Bellon H, Jackson TE and Restrepo JF (1991a)Andesite and dacite genesis via contrasting processes: the ge-ology andgeochemistry of El Valle Volcano, Panama. ContribMineral Petrol 106: 309–324
Defant MJ, Richerson PM, De Boer JZ, Stewart RH, Maury RC,Bellon H, Drummond MS, Feigenson MD, Jackson TE (1991b)Dacite genesis via both slab melting and differentiation: petro-genesis of La Yeguada Volcanic Complex, Panama. J Petrol 32:1101–1142
DePaolo DJ (1981) Trace element and isotopic effects of combinedwall-rock assimilation and fractional crystallization. EarthPlanet Sci Lett 53: 189–202
Drummond MS, Defant MJ (1990) A model for trondhjemite-tonalite-dacite genesis and crustal growth via slab melting:Archean to modern comparisons. J Geophys Res 95: 21503–21521
Forsyth DW (1975) Fault plane solution and tectonics of the SouthAtlantic and Scotia Sea. J Geophys Res 80: 1429–1443
Fuenzalida RP (1974) The Magallanes Fault zone. In: Gonzales-Ferran O (ed) Andean and Antarctic volcanologic problems.Proceedings IAVCEI Intern Volcan Sympos, Santiago, Chile,pp 373–391
Futa K, Stern CR (1988) Sr and Nd isotopic and trace elementcompositions of Quaternary volcanic centers of the southernAndes. Earth Planet Sci Lett 88: 253–262
Gill JB (1981) Orogenic andesites and plate tectonics. Springer,Berlin Heidelberg New York
Green TH, Ringwood AE (1968) Genesis of calc-alkaline igneousrock suite. Contrib Mineral Petrol 18: 105–162
Harambour SM (l988) Sobre el hallazgo del mitico volcan Reclus,ex Mano del Diablo, Hielo Patagonico Sur, Magallanes, Chile.Rev Geol Chile 15: 173–180
Herron EM, Bruhn R, Winslow M, Chuaqui L (1977) Post Miocenetectonics of the margin of southern Chile. In: Talwani M, Pit-man WC (eds) Island arcs, deep sea trenches and back-arcbasins (Maurice Ewing Series 1). American GeophysicalUnion, Washington, pp 273–283
Heusser CJ, Heusser LE, Hauser A (l990) A 12,000 B.P. tephralayer at Bahia Inutil (Tierra del Fuego, Chile). Ans Ins Pat l9:39–49
Hofmann AW (1988) Chemical differentiation of the Earth: rela-tionship between mantle, continental crust and oceanic crust.Earth Planet Sci Lett 90: 297–314
Johannes W, Wolke C (l994) Bildung von tonalitischenTeilschmelzen - ein Schritt in der Entwicklung der kontinental-en Kruste. Eur J Mineral 6: 126
Johnson KTM (l994) Experimental cpxyand garnetymelt parti-tioning of REE and other trace elements at high pressures:petrologic implications. Mineral Mag 58: 454–455
Kay RW (1978) Aleutian magnesium andesites: melts from subduc-ted Pacific oceanic crust. J Volcanol Geotherm Res 4: 117–132
Kay RW (1980) Volcanic arc magmas: implications of a melting-mixing model for element recycling in the crust upper mantlesystem. J Geol 88: 497–522
Kay SM, Ramos VA, Marquez M (1993) Evidence in Cerro Pampavolcanic rocks for slab-melting prior to ridge-trench collisionin Southern South America. J Geol 101: 703–714
281
Keleman PB (l990) Reaction between ultramafic rock and frac-tionated basaltic magma. I. Phase relations, the origin of calc-alkaline magma series, and the formation of discordant dunite.J Petrol 31: 51–98
Keleman PB (l995) Genesis of high Mg' andesites and the conti-nental crust. Contrib Mineral Petrol 120: 1–19
Keleman PB, Shimizu N, Dunn T (1993) Relative depletion ofniobium in some arc magmas and the continental crust: parti-tioning of K, Nb, La and Ce during meltyrock reaction in theupper mantle. Earth Planet Sci Lett 120: 111–134
Kilian R (l995) Interaction of tonalitic melts with mantle peri-dotite: evidence from a xenolith of the southern Andes (ab-stract). IUGG XXI General Assembly: A468-469
Kilian R, Lopez-Escobar L, Lobato J (1991) Quaternary volcan-ism of the Austral Volcanic Zone of the Andes (49–528S):geochemistry and petrology. Zentralbl Geol Paläontol Teil 1:1709–1721
Lopez-Escobar L, Kilian R, Kempton PD, Tagiri M (l993) Petrol-ogy and geochemistry of Quaternary rocks from the SouthernVolcanic Zone of the Andes between 418309 and 468009S,Chile. Rev Geol Chile 20: 33–56
Martin H (1993) The mechanisms of petrogenesis of the Archaeancontinental crust: comparison with modern processes. Lithos30: 373–388
Martinic M (1988) Actividad volcanica historica en la region deMagallanes. Rev Geol Chile 15: 181–186
Morris P (1995) Slab melting as an explanation of Quaternaryvolcanism and aseismicity in southwest Japan. Geology 23:395–398
Muehlenbachs K, Clayton RN (l972) Oxygen isotope studies offresh and weathered submarine basalts. Can J Earth Sci 9:172–184
Myers JD, Frost CD (1994) A petrologic re-investigation of theAdak volcanic center, central Aleutian arc, Alaska. J VolcanolGeotherm Res 60: 109–146
Nichols GT, Wyllie PJ, Stern CR (l994) Subduction zone meltingof pelagic sediments constrained by melting experiments. Na-ture 371: 785–788
Peacock SM, Rushmer T, Thompson AL (1994) Partial melting ofsubducted oceanic crust. Earth Planet Sci Lett 121: 227–244
Pearce JA, Parkinson IJ (1993) Trace element models for mantlemelting: application to volcanic arc petrogenesis. In: PrichardHM, Alabaster T, Harris NBW, Neary CR (eds) Magmatic pro-cesses and plate tectonics. Geol Soc London Spec Pub 76:373–403
Peng ZC, Zartman RE, Futa K, Chen DG (l986) Pb-, Sr- andNd-isotopic systematics and chemical characteristics of Ceno-zoic basalts, eastern China. Chem Geol 59: 3–33
Puig AG, Herve M, Suarez M, Saunders AD (1984) Calc-alkalineand alkaline Miocene and calc-alkaline recent volcanism in theSouthermost Patagonian Cordillera, Chile. J VolcanolGeotherm Res 21: 149–163
Ramos VA, Kay SM (l992) Southern Patagonian plateau basaltsand deformation: backarc testimony of ridge collision. Tec-tonophysics 205: 261–282
Rapp RP, Watson EB, Miller CF (1991) Partial melting of amphi-boliteyeclogite and the origin of Archean trondhjemites andtonalites. Precambrian Res 51: 1–25
Rogers G, Saunders AD, Terrell DJ, Verma SP, Marriner GF(1985) Geochemistry ofHolocene volcanic rocks associatedwith ridge subduction in Baja, California, Mexico. Nature 315:389–392
Schnetger B (l993) Partial melting during the evolution of theamphibolite- to granulite gneisses of the Ivrea Zone, northernItaly. Chem Geol 113: 71–101
Sekine T, Wyllie PJ (l983) Experimentasl similation of mantlehybridization in subduction zones. J Geol 91: 511–528
Selverstone J, Stern CR (l983) Petrochemistry and recrystalliza-tion history of granulite xenoliths from the Pali-Aike volcanicfield, Chile. Am Mineral 68: 1102–1112
Sen C, Dunn T (1994) Dehydration melting of a basaltic composi-tion amphibolite at 1.5 and 2.0 GPa: implications for the originof adakites. Contrib Mineral Petrol 117: 394–409
Stern CR (1974) Melting products of olivine tholeiite basalt insubduction zones. Geology 2: 227–230.
Stern CR (1980) Chemistry of Chilean ophiolites: evidence for thecompositional evolution of the mantle source of back-arc basinbasalts. J Geophys Res 85: 955–966
Stern CR (1990) Tephrochronology of southermost Patagonia. Na-tional Geograph Res 6: 110–126
Stern CR (l991) Isotopic composition of late Jurassic and earlyCretaceous mafic igneous rocks from the southernmost Andes:implications for sub-Andean mantle. Rev Geol Chile 18: 15–23
Stern CR, Wyllie PJ (1978) Phase compositions through crystal-lization intervals in basalt-andesite-H2O at 30 kbar with impli-cations for subduction zone magmas. Am Mineral 63: 641–663
Stern CR, Futa K, Muehlenbachs K (1984) Isotope and trace ele-ment data for orogenic andesites from the Austral Andes. In:Harmon RS, Barreiro BA (eds) Andean magmatism: chemicaland isotopic constraints. Shiva Publishing, Cheshire, pp 31–46
Stern CR, Saul S, Skewes MA,Futa K (l989) Garnet peridotitexenoliths from the Pali-Aike basalts of southernmost SouthAmerica. Spec Publ Geol Soc Aust 14: (2) 735–744
Stern CR, Frey FA, Futa K, Zartman RE, Peng Z, Kyser TK (1990)Trace element and Sr, Nd, Pb, and O isotopic composition ofPliocene and Quaternary alkali basalts of the Patagonianplateau lavas of southernmost South America. Contrib MineralPetrol 104: 294–308
Sun SS, McDonough WF (1989) Chemical and isotopic systemat-ics of oceanic basalts: implications for mantle composition andprocesses. In: Saunders AD, Norry MJ (eds) Magmatism inOcean Basins. Geol Soc London Spec Pub 42: 313–345
Wolf MB, Wyllie PJ (1994) Dehydration melting of amphibolite at10 kbar: the effect of temperature and time. Contrib MineralPetrol 115: 369–383
Wyllie PJ, Wolf MB (1993) Dehydration melting of solid amphi-bolite at 10 kbar: textural development, liquid interconnectivi-ty and applications to segregation of magmas. Mineral Petrol44: 151–179
Yogodzinski GM, Kay RW, Volynets ON, Loloskov AV, Kay SM(1995) Magnesian andesite in the western Aleutian Koman-dorsky region: implications for slab melting and processes inthe mantle wedge. Geol Soc Am Bull 107: 505–519