Late Cenozoic magmatism of the Bolivian Altiplano - Springer Link

3 downloads 0 Views 3MB Size Report
Late Cenozoic magmatism of the Bolivian Altiplano. Received: 22 March 1994 / Accepted: 26 August 1994. Abstract Small basalt to dacite volcanic centers areĀ ...
Contrib Mineral Petrol (1995) 119:387-408

9 Springer-Verlag 1995

Jon R Davidson 9 Shanaka L. de Silva

Late Cenozoic magmatism of the Bolivian Altiplano

Received: 22 March 1994 / Accepted: 26 August 1994

Abstract Small basalt to dacite volcanic centers are distributed sparsely over the Bolivian Altiplano, behind the Andean volcanic front. Most are Pliocene to Recent in age, and are characterized by textural and mineralogical disequilibrium with abundant xenoliths and xenocrysts. True phenocrysts are rare in the more mafic samples. Compared with Recent volcanic rocks from Andean stratovolcanoes, the Bolivian centers overlap in major element trends. Incompatible element contents tend to be higher, particularly in the eastern Altiplano. The ranges of isotopic compositions reflect ubiquitous crustal contamination. Pb isotope compositions are dominated by Pb from isotopically heterogeneous basement, resulting in a wide scatter of data lying between inferred crustal compositions and showing little overlap with possible mantle sources in the region. Rocks sampled from the Bolivian Altiplano were probably derived from asthenospheric mantle and subjected to extensive open system differentiation during ascent through the 70 km thick crust of the region. Maj or element trends are largely controlled by the fractionating phase assemblage (olivine, clinopyroxene and amphibole). Trace element and isotope systematics, however, defy realistic attempts at modeling due to the geographic scatter of samples, the uniformity of compositions at a given center, and the heterogeneity of the contaminant. Nevertheless, there are first order systematic trace element variations that appear to relate to the geometry of the subduction zone. In particular, LILE/HFSE (exemplified by Ba/Nb), and Zr/Nb ratios decrease from the arc front eastward into the Altiplano. These variations are not easily reconciled with control by crustal contamination alone. A model is

(~) Department of Earth and Space Sciences, UCLA, Los Angeles, CA 90024, USA J. P. D a v i d s o n

S. L. de Silva Department of Geographyand Geology, Indiana State University,Terre Haute, IN 47809, USA Editorial responsibility: T. L. Grove

proposed in which the asthenospheric source is fluxed by high Ba/Nb slab-derived fluid to induce melting. Beneath the arc, high fluid flux increases the Ba/Nb ratio of the asthenosphere and leads to high degrees of partial melting (high Zr/Nb). To the east, lower or no fluid flux leads to low Ba/Nb and low degrees of partial melting (low Zr/Nb). Melting in the back arc region of the Altiplano may be facilitated by lithospheric delamination that leads to decompression melting of counter-flowing asthenosphere. There is no unequivocal evidence that requires a significant role for the lithospheric mantle.

Introduction Despite twenty years of geochemical investigation, the origin of subduction-related magmas and their role in the growth of the continental crust remain poorly constrained. Estimates of the bulk composition of the crust are similar to the average compositions of magmatic rocks emplaced along continental margins such as the Andes (Fig. 1). This begs the question of whether the magmatism represents a process of differentiation from the mantle, or whether it is largely a reflection of recycling of extant crustal material by anatexis and gross contamination of a much smaller mantle-derived magma flux. In this regard it is important to constrain differentiation processes in order to estimate the nature of primitive magmas, which in turn will enable us to evaluate whether they are derived from mantle or crustal sources, or some mixture of the two. Since the introduction of the term andesite, it has been recognized that the bulk of volcanic rocks at continental margin arcs are highly differentiated. In the Central Volcanic Zone of the Andes (CVZ) where crustal thicknesses reach 70 km, the volumetrically weighted average composition of erupted material is approximately andesitic in composition, and instances of true basalts (v

arc melts?

/S / . ~. ~. 9 -

--

4

/ pr imary'

.o/~

I o CVZ stratovolcanoes

l ...............

6

8

10

12

14

J

are typically small-volume and non-porphyritic. Sparse analyses of these lavas (Thorpe et al. 1984) suggested that they may be quite primitive, and might therefore afford an opportunity to sample the mantlederived flux in the central Andes, albeit not directly along the arc front. However, all of the centers we sampled and analyzed are more differentiated than the one analysis from the Altiplano published by Thorpe et al. in 1984.

Analytical techniques 16

MgO Fig. 2 Ni vs MgO for central Andean rocks from literature sources, compared with minor Bolivian centers of this study. Data for CVZ volcanic rocks in this and subsequent figures are taken mainly from O'Callaghan and Francis (1986), Davidson etal. (1990), de Silva et al. (1993) and Feeley and Davidson (1994). The field of compositions inferred to be in equilibrium with mantle periodotite is from Hart and Davis (1978). An estimate of the composition of primary arc melts is also shown (based on primitive arc basalts described by Nye and Reid 1986, Eggins 1993, and references therein). Two model a r r o w s for olivine fractionation and bulk mixing with continental crust are shown - these are not meant to be realistic models; they merelY serve to indicate that rocks that are considered "mafic" (>6% MgO) may have been highly contaminated

1989), by extreme fractional crystallization of mantlederived magmas or by a combination of these mechanisms. Isotopic andtrace element characteristics of CVZ rocks are distinct from those found at oceanic subduction zones (Davidson et al. 1991) indicating that, if derived from a mantle source with little open system modification, the source is n o t the a s t h e n o s p h e r e , b u t m o r e likely ancient incompatible element-enriched subcontinental l i t h o s p h e r i c m a n t l e (Rogers and H a w k e s w o r t h 1989). Conversely, if a s t h e n o s p h e r e - d e r i v e d m a g m a s are involved it is i m p l i c i t that they have b e e n m a s s i v e l y c o n t a m i n a t e d d u r i n g p a s s a g e t h r o u g h the lithosphere. Clearly it is essential to resolve the relative c o n t r i b u t i o n s of crust and m a n t l e m a t e r i a l to A n d e a n m a g m a s i f we are to d e t e r m i n e the relative v o l u m e s of m a g m a t i c m a t e r i a l t h a t are m a n t l e - d e r i v e d additions and r e c y c l e d respectively. The highly d i f f e r e n t i a t e d n a t u r e of c e n t r a l A n d e a n volcanic rocks h i n d e r s our a b i l i t y to i d e n t i f y p r i m i t i v e m a g m a c o m p o s i t i o n s (Fig. 2). T h e d e v e l o p m e n t of large stratovolcanoes that f o r m the arc, and the p o r p h y r i t i c n a t u r e of the rocks, with low p r e s s u r e phase a s s e m blages, is likely due to extensive p r o c e s s i n g i n shallow m a g m a c h a m b e r s . To c i r c u m v e n t this p r o b l e m , we have studied m i n o r ( s m a l l - v o l u m e ) m a f i c volcanic centers f r o m the B o l i v i a n A l t i p l a n o , a b a c k arc r e g i o n i m m e d i ately east of the v o l c a n i c front (Fig. 1). The centers are c o m m o n l y a r r a n g e d in systematic l i n e a m e n t s , perhaps r e f l e c t i n g deep crustal structures, and their lavas

Rock samples were prepared by removing all weathered surfaces and crushing to chip size. At this stage any obvious xenoliths were separated - but note that fine xenocrystic material which was abundant in most samples (see Petrology and Mineral Chemistry, below) could not be removed. Rock chips were powdered in an alumina shatterbox. Major and trace element abundances were measured by XRF using a Rigaku spectrometer at University of Southern California (USC). Estimated precision is bettei- than 1% for all elements except Y, Nb and Cr (better than 5%). Results for a duplicate analysis and two international standards are in excellent agreement (Appendix 1). Most of the samples were analyzed by INAA for a further set of elements (see Table 3). Samples were irradiated and analyzed at the Oregon State University Triga Reactor facility. Precision is better than 2% for Sc, Ta, La, Sm and Eu, 2-5% for Hf, Th, Ce, Nd, Tb, Yb and Lu, and ca. 6% for U. For isotopic analyses, ca. 30 mg of rock powder was dissolved in an HF-HNO3 mixture in sealed teflon beakers for >6 h on a hot plate. Sr and REE were separated by standard cation exchange techniques, and Nd subsequently separated from the other REE using DEHP-coated teflon resin. Pb was separated from a separate dissolution using 600 pA anion exchange columns. Total process blanks were ca. 350, 100 and 1000 pg for Sr, Nd and Pb respectively. Sr was run with TaO2 and phosphoric acid on single Re filaments using a dynamic routine on the multicollector VG Sector mass spectrometer at UCLA. S7Sr/86Sr ratios were normalized to S6Sr/SaSr=0.1194. NBS 987 over the period of analysis was measured as 0.71022_ + 13 (n = 15). Pb was run on single Re filaments with silica gel and phosphoric acid, on the same instrument. Mass fractionation was monitored by including at least one NBS 981 standard in each turret of 10 samples, and corrected for by normalizing to accepted values of NBS 981. Nd was loaded in dilute HN% on one side of a separable triple Re filament assembly, and run using a static routine on the VG Sector 54-30 mass spectrometer at UCLA. 143Nd/te4Ndratios were normalized to 146Nd/ H4Nd=0.7219. Analyses of the La Jolla standard gave a value of 0.511840_+11 (n=4). 4~ age determinations were performed following irradiation at the University of Michigan facility. Two laser fusion analyses were performed on the VG 3600 mass spectrometer at UCLA, and one furnace step heating analysis was undertaken on a MAP instrument at the New Mexico Bureau of Mines. Further details are given in Table 1. Mineral compositions were determined on the Cameca SX50 microprobe at Indiana University using an accelerating voltage of 15 kV, a beam current of 20 nA and a beam diameter of 1 Ixm. Na was determined first in order to minimize loss through volatilization.

Geology and stratigraphy The locations of minor centers sampled in this study are shown in Fig. lb. Volcanic centers were targeted on the basis of Landsat and Space Shuttle imagery. There are three main areas in which the centers are located: (1) immediately behind the arc in the region of

390 Sajama, (2) immediately behind the arc in the region of Oliagtie, and (3) well to the east in the Altiptano. The centers near Sajama are trachybasalts from small cinder cones. Some major element analyses of volcanic rocks from this region have been published previously by Avila Salinas et al. (I 985). Our samples BC9029 and BC9034 probably correspond to their samples L-9 and L-8 respectively. The centers consist of one or two small lava flows. The largest of these is Sunu Kkollu from which samples BC9032 and BC9033 were taken. Locality BC9029 (Cerro Pucari) is a columnar jointed flow which overlies bedded strombolian and phreatomagmatic deposits, which in turn overlie a 2.2 Ma ignimbrite of the Mauri Formation. The phreatomagmatic deposits indicate interaction With a shallow water reservoir, and suggest that eruption of the trachybasalt took place during a much wetter period in the Altiplano's history. The field of centers to the east of Ollagtie covers a wide area and includes both silicic andesite and dacite flow complexes (BC9006-11) and trachybasalt cinder cones (BC9005, 12). The most evolved centers are located in the Pampa Luxsar region just to the north and northwest of Cerro Luxsar, a Pliocene to Pleistocene composite cone (de Silva and Francis 1991). Many of the centers are encrusted by lake terrace deposits related to the highest stand of the ancestral lake Minchin that occupied the Altiplano during late Pleistocene times. The high stand of this lake has been dated at - 1 7 ka (Bills et al. 1994). All these centers overlie regional ignimbrites which are late Miocene to Pliocene in age, thus constraining their ages to Pliocene or younger. The minor centers on the Altiplano to the east define two broad lineaments subparallel to the trend of the arc front (Fig. lb). The central Altiplano group comprises basaltic trachyandesites to dacite flows that form isolated plugs or mesa-like flows. The age relations of these again suggest a Pliocene to Pleistocene age range. Two small maars, Nekhe Kkota and Jayu Kkota (BC9014 and BC9015) are much more youthful and are probably Holocene in age (de Silva & Francis 1991). Phreatomagmatic eruptions have resulted in deposits consisting of olivine-phyric basalt with abundant xenoliths of granitic lithologies. Chiar Kkollu is a small isolated hilt formed by an alkali basalt sill. This lava, first analyzed by Thorpe et al. (1984), is clearly older than the other Quaternary centers. A 39Ar/4~ date o]1 K-rich groundmass gave an age of 22.5 Ma (Table 1). This is similar to the age of 25.2-+0.5 Ma obtained through K-Ar dating by Hoke et al. (1993). The older age of the Chiar Kkollu basalt is significant it is the most mafic sample collected, is free from petrographic evidence for contamination, and may represent little-modified basaltic compositions erupted prior to Miocene crustal shortening (Davidson and de Silva1993). Further east, a second lineament occurs along the southwestern edge of Lago Poopo. Lake terraces on these centers indicate that they too are older than the high stand of ancestral Lake Minchin ( - 1 7 ka) that included Lago Poopo (Bills et al. 1994). Rocks here vary from basalts to andesites, are microporphyritic and clearly contaminated. Quillacas (BC9022-24; Fig. lb) comprises two small hills where lava has erupted along two main SE-NW trending faults which have uplifted the local basement about 30 m. The northern, smaller hill comprises a plexus of subvertical feeder

Table I 4~

age data for Bolivia minor centers

Sample

Location and Description

Age (Ma)

BC9016A (K-rich groundmass) BC9022 (amphibole) BC9025 (K-fspr megacryst)

Chiar Kkollu, lava

22.51 +0.45 a

Quillacas, lava

1.45+0.15

Pampa Aullagas, lava

1.89-+0.0I c

b

a Laser fusion: (mean of 9 individual fusions) 63% radiogenic argon on average b Laser fusion: (one shot) 10% radiogenic argon c Isochron age: 11 of 19 steps representing 90% of released argon

dikes with brecciated margins, which in places turn over into a subhorizontal disposition suggesting effusion and flow at the surface. The southern hill consists of a series of flows emanating from a vent marked by agglomerate, at the summit of the hill. At least one of the lava flows is clastogenic. On the southeastern side another vent, about 100 m lower in elevation, erupted phreatomagmatic deposits. These bedded deposits are found some 30 m above the surrounding Altiplano, suggesting eruption during a recent lake highstand - probably the late Pleistocene. However, the northern part of Quillacas has lake terraces cut into it, suggesting that it may be older, and a 39Ar/4~ date of 1.2 Ma was obtained by laser fusion of hornblende separated from an andesite flow (Table 1). Cerro Aullagas consists of a large hill of uplifted local basement, again along a SE-NW trending fault. Two or three lava flows Were erupted from a common vent at the summit. A K-feldspar xenocryst gave an 39Ar/4~ age of 1.89 Ma (Table 1) with a flat age spectrum (indicating complete outgassing and resetting to the age of assimilation and eruption). A hornblende-bearing silicic andesite was collected from Cerro Gordo, the easternmost of the Bolivian Altiplano minor centers studied (BCg001; Fig. lb).

Petrology and mineral chemistry Texture and mineralogy varies with the degree of differentiation of the lavas. With the exception of Chiar Kkollu, the basalt and basaltic andesites are typically glassy. Several samples exhibit a seriate texture of olivine, plagioclase and augite embedded in a groundmass of minute plagioclase and more rarely augite (Fig. 3a). Hypersthene is rare and generally subordinate to augite except in the lavas of Quillacas and Cerro Aullagas. Olivine is the most common microphenocryst in these lavas and commonly contains inclusions of chrome spinel. Magnetite was the only oxide detected and is ubiquitous throughout the basalts and basaltic andesites. The lava of Chiar Kkollu sill is a fine- to medium-grained otivine-phyric basalt. Sub-ophitic patches are common, but more commonly augite and plagioclase exhibit interstitial and intergranular relationships (Fig. 3b). In the andesites, hornblende joins plagioclase and augite as the most common minerals and olivine is rare. Most of the more mafic

Fig. 3a-h Photomicrographs showing charactertistic textures of the Bolivian minor centers: a Seriate textured basaltic lava with microphenocrysts of olivine (O) in a glassy groundmass with plagioclase microlites, augite, and magnetite (xp). Field of view 2 mm. BC9029 Cerro Pucara. b Chiar Kkollu basalt; phenocrysts of olivine (O) in a groundmass of intersertal and interstitial plagioclase and augite (XP). Field of view 4 mm. BC9016A. c Seriate textured andesite lava. Phenocrysts o f plagioclase (white laths), hornblende (/t) with opacite rims, biotite (B) in a finegrained groundmass of the same minerals (ppl). Field of view 2 mm. BC9001. Cerro Gordo. d Dacite: large zoned sieve-textured plagioclase phenocryst, with heavily oxidised biotite (B), and hornblende (H) phenocrysts (partially xp). Field of view 2 mm. BC9013. Cerro Jara. e Embayed quartz xenocryst rimmed by clinopyroxene (CPX). Host andesite lava has a trachytic texture defined by plagioctase microlites (xp). Field of view 0.5 ram. BC9021. Cerro San Martin. f Annealed plutonic xenolith donsisting of quartz and alkali feldspar (right) in a seriate textured basalt with microphenocrysts of olivine, augite and plagioclase, and xenocrysts (left). Disaggregatiou of the xenolith is occuring at margin. Note resorption of quartz grains and crenutate grain boundaries (xpL). Field of view 2 ram. BC9015. Jayu Kkota Maar. g Quartz-sillimanite gneiss xenolith (left) in andesitic lava (right). Note penetration of veinlet of magma into xenolith (arrow). (xp). Field of view 2 ram. BC9025. Cerro Aullagas. h Large kyanite xenocryst in andesite lava. Note resorption on edge and minor disaggregation (arrows). (xp). Field of view 2 mm. BC9025. Cerro Aullagas

391

392 100

I

I

I

I

I

1.20

I

a

t

80 60

A

0

9

40 .

< 0

i

85

9 9

9

0

0

o~R Z~ ~-~

i

a A i

,

~ A ~ Ai

0.20

80 . t 75 i

/,, 9 cores O rims ix groundmass

~~

50

I

44

48

52

I

I

I

56

,

,

~ 4 ~ * ~"

i

,

,

,

i

,

,

,

i

,

,

,

i

,

,

,

N

. . . . . . . . . . . . . . . . .

0.00

55

,

i ooo

D

i

o~@./~_ xc~' ~X9 ~0,-~

0.40

9 A

9

70 65

'

;~ ~ 0.60

A eAW~

~A

20

'

b

~ o.8o

O

q~

s

.. ~ 1.oo

'

I

60

I

64

W h o l e rock S i O 2 Fig. 4a, b Mineral chemistry of Bolivian minor centers, emphasizing disequilibrium, a Plagioclase and olivine composition vs whole rock SiO2. b Equilibrium Mg/Fe of liquid calculated from olivine compositions, compared with measured whole rock Mg/Fe

andesites are fine-grained with the major minerals displaying a seriate texture within a groundmass of glass and plagioclase microlites (Fig. 3c). Some of the larger plagioclase microphenocrysts display complex zoning, a feature not seen in the basalts. Trachytic textures are displayed by some of the andesites, such as the three hornblende microporphyritic samples collected from Lomas Uchimilles (BC9002-4). In the most silicic andesites and dacites, porphyritic textures predominate, biotite joins plagioclase and hornblende as the main phenocrysts, and pyroxene is subordinate. Plagioclase phenocrysts are large and zoned with sieve-textured cores and intermediate zones (Fig. 3d). The phenocrysts are set in a fine-grained matrix of the same minerals and glass. In samples from Quillacas (BC9022-24), magma mingling textures are observed on the mm-to-cm scale. The process manifests itself as highly vesicular, lighter, selvages with resorbed and altered phenocrysts within the poorly vesicular darker host lava. Vesiculated mafic inclusions fi-om Quillacas and Cerro Pucara (BC9028), have an acicular hornblende - plagioclase assemblage, common in magmatic inclusions. Xenocrystic material is a common feature in most of the more mafic lavas we sampled (Fig. 3e-h). Quartz is the most abundant xenocryst and is commonly resorbed and rimmed with clinopyroxene (Fig. 3e). Other xenocrysts include alkali feldspar, biotite, apatite and kyanite, all of which show advanced stages of reaction. Xenoliths of granitic and metamorphic lithologies were common in several samples, most notably from Quillacas and Cerro Aullagas (metamorphic) and the maars of Jayu Kkota and Nekhe Kkota (granitic). Plutonic xenoliths consist of quartz and feldspar displaying annealing textures with rounded and crenulate grain boundaries (Fig. 3f). The metamorphic xenoliths (garnet-biotite gneiss) commonly contain aluminosilicates (Fig. 3g, h). All these materials show strong evidence of disaggregation and assimilation into the lava. Plagioclase compositions vary from>An85 to