Sangay volcano, Ecuador: structural development, present activity and petrology

Journal of Volcanology and Geothermal Research 90 Ž1999. 49–79

Sangay volcano, Ecuador: structural development, present activity and petrology Michel Monzier a,) , Claude Robin b,1, Pablo Samaniego c,2 , Minard L. Hall Jo Cotten d,4 , Patricia Mothes c,5, Nicolas Arnaud e,6

c,3

,

a

IRD (preÕiously ORSTOM), Apartado 17-12-857, Quito, Ecuador IRD, UniÕersite´ Blaise Pascal, 5 rue Kessler, 63038, Clermont-Ferrand Cedex, France Instituto Geofisico, Escuela Politecnica Nacional (IG-EPN), Apartado 17-01-2759, Quito, Ecuador d UMR 6538, UniÕersite´ de Bretagne Occidentale, B.P. 809, 29285, Brest, France e UMR 6524 CNRS, Lab. ‘Magmas et Volcans’, 5 rue Kessler, 63000 Clermont-Ferrand, France b

c

Received 7 October 1998; accepted 27 January 1999

Abstract Sangay Ž5230 m., the southernmost active volcano of the Andean Northern Volcanic Zone ŽNVZ., sits ; 130 km above a ) 32-Ma-old slab, close to a major tear that separates two distinct subducting oceanic crusts. Southwards, Quaternary volcanism is absent along a 1600-km-long segment of the Andes. Three successive edifices of decreasing volume have formed the Sangay volcanic complex during the last 500 ka. Two former cones ŽSangay I and II. have been largely destroyed by sector collapses that resulted in large debris avalanches that flowed out upon the Amazon plain. Sangay III, being constructed within the last avalanche amphitheater, has been active at least since 14 ka BP. Only the largest eruptions with unusually high Plinian columns are likely to represent a major hazard for the inhabited areas located 30 to 100 km west of the volcano. However, given the volcano’s relief and unbuttressed eastern side, a future collapse must be considered, that would seriously affect an area of present-day colonization in the Amazon plain, ; 30 km east of the summit. Andesites greatly predominate at Sangay, there being few dacites and basalts. In order to explain the unusual characteristics of the Sangay suite—highest content of incompatible elements Žexcept Y and HREE. of any NVZ suite, low Y and HREE values in the andesites and dacites, and high NbrLa of the only basalt found—a preliminary five-step model is proposed: Ž1. an enriched mantle Žin comparison with an MORB source., or maybe a variably enriched mantle, at the site of the Sangay, prior to Quaternary volcanism; Ž2. metasomatism of this mantle by important volumes of slab-derived fluids enriched in soluble incompatible elements, due to the subduction of major oceanic fracture zones; Ž3. partial melting of this metasomatized mantle and generation of primitive basaltic melts with NbrLa values typical of the NVZ, which are parental to the entire

) 1 2 3 4 5 6

Corresponding author. E-mail: [email protected] E-mail: [email protected]. E-mail: [email protected]. E-mail: [email protected]. E-mail: [email protected]. E-mail: [email protected]. E-mail: [email protected].

0377-0273r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 0 2 7 3 Ž 9 9 . 0 0 0 2 1 - 9

50

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Sangay suite but apparently never reach the surface and subordinate production of high NbrLa basaltic melts, maybe by lower degrees of melting at the periphery of the main site of magma formation, that only infrequently reach the surface; Ž4. AFC processes at the base of a 50-km-thick crust, where parental melts pond and fractionate while assimilating remelts of similar basaltic material previously underplated, producing andesites with low Y and HREE contents, due to garnet stability at this depth; Ž5. low-pressure fractionation and mixing processes higher in the crust. Both an enriched mantle under Sangay prior to volcanism and an important slab-derived input of fluids enriched in soluble incompatible elements, two parameters certainly related to the unique setting of the volcano at the southern termination of the NVZ, apparently account for the exceptionally high contents of incompatible elements of the Sangay suite. In addition, the low CrrNi values of the entire suite—another unique characteristic of the NVZ—also requires unusual fractionation processes involving Cr-spinel andror clinopyroxene, either in the upper mantle or at the base of the crust. q 1999 Elsevier Science B.V. All rights reserved. Keywords: Northern Andean Volcanic Zone; Sangay volcano development; Sector collapse; Primitive arc basalt; Mantle processes; Crustal processes

1. Introduction Sangay Ž2.008S, 78.348W; 5230 m., located on the eastern edge of the eastern cordillera of Ecuador, is the most active volcano in the Northern Volcanic

Zone of the Andes ŽNVZ; Thorpe et al., 1982; Figs. 1 and 2A–B.. As permanent explosive activity has been continuously observed in its crater since 1628, it is one of the most active andesitic volcanoes in the world. Large eruptions occurred in 1628, 1728,

Fig. 1. Sangay III in December 1995 viewed from the base camp 4.3 km SW from the summit. The recent pyroclastic-flow deposit on which the campsite is located is in the foreground Žsee Fig. 3A., among the badlands corresponding to Sangay II edifice. At the summit can be seen the W lava dome and its now inactive lava tongues; behind this dome, a steam plume rises from the main crater. Photo by M. Monzier ŽIRD..

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

51

Fig. 2. ŽA. Distribution of the active volcanic zones along the Andean Cordillera Žafter Simkin and Siebert, 1994.; NVZs Northern Volcanic Zone, CVZ s Central Volcanic Zone, SVZ s Southern Volcanic Zone, AVZ s Austral Volcanic Zone. ŽB. Relationships between the Cocos–Nazca plates and the Colombian–Ecuadorian subduction zone Žfrom Lonsdale, 1978 and Lonsdale and Klitgord, 1978, modified.; black arrows indicate the Nazca–South America relative motion according to DeMets et al. Ž1990.; 1 s Northern Volcanic Zone with Chimborazo ŽCh., Tungurahua ŽTu. and Sangay volcanoes, 2 s trench, 3 s age of the oceanic crust in Ma. The box corresponds to Fig. 2C. ŽC. Intermediate seismicity Ž1968–1996. under the southern termination of the Northern Volcanic Zone. Seismicity data are from the USGSrNEIC Ž1968–1992. Ževents recorded by 20 stations or more, from May 1968 to June 1992., the National Ecuadorian Network Ževents recorded by 15 stations or more, from 1988 to 1996. and the Lithoscope Experiment ŽDecember 1994 to May 1995.. Depths to selected foci are given in kilometers. Volcanoes: Ch s Chimborazo, Ca s Carihuayrazo, Pu s Punalica, Hu s Huisla, C s Calpi, Ig s ˜ Igualata, Tu s Tungurahua, T s Tulabug, Al s Altar, S s Sangay. In grey, the sharp N558E boundary between the southern seismogenic slab and the northern, weakly seismogenic slab.

1738–1744, 1842–1843, 1849, 1854–1859, 1867– 1874, 1872, 1903, 1934–1937 and 1941–1942 ŽLewis, 1950; Hall, 1977; J. Egred, pers. commun.,

1998. and were characterized by Strombolian activity with block and ash explosions, ash falls, frequent lava flows, pyroclastic flows, and lahars. An old

52

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Spanish document mentioned by Lewis Ž1950. reports the disastrous effects of the 1628 eruption, during which abundant ash fell upon Riobamba, 50 km NW of the volcano, and mantled pastures with sufficient ash so that cattle starved. In the geodynamic context of the northern Andes, Sangay volcano occupies a special position. It is the southernmost Quaternary edifice of the NVZ, which extends northwards over 900 km to Guadalupe volcano in Colombia. To the south, active volcanism is absent along a 1600-km-long segment of the Andean Cordillera ŽFig. 2A–B.. Contrary to other volcanoes of southern Colombia and Ecuador, beneath which the intermediate seismicity Ž80–220 km in depth. is almost lacking, a strongly seismogenic slab is present under the Sangay area ŽPennington, 1981.. Updated seismological data show that this slab dips ; 358 towards N558E and is ; 130 km beneath the volcano ŽFig. 2C.. The N558E-oriented boundary between the seismogenic slab to the south and the weakly seismogenic slab to the north, beyond El Altar and Tulabug volcanoes, probably reflects a sharp change in the thermal characteristics of the subducted oceanic crust. This boundary clearly coincides with a set of major oceanic faults which separate the young Ž- 22 Ma. oceanic crust being subducted under Colombia and Ecuador, from older oceanic crust Ž) 32 Ma. beneath southern Ecuador and northern Peru ŽFig. 2B–C; Lonsdale, 1978; Lonsdale and Klitgord, 1978.. According to Stein and Stein Ž1992., heat-flow values are high for oceanic crusts younger than 20–25 Ma, but very low and relatively stable for older oceanic crusts. Moreover, the collision of the Carnegie Ridge—the trace of the active Galapagos Hotspot—with the continent

may be responsible for an increase in heat flow in front of northern Ecuador. Thus, contrary to what occurs under most Ecuadorian volcanoes, the slab beneath Sangay volcano is thought to be older, relatively cold, and more seismogenic. Lastly, Hall and Wood Ž1985. attributed Sangay’s persistent activity and monotonous petrography to its location near the segment boundary that separates these two crusts. Despite the continuous eruptive activity observed at Sangay, no detailed study has been done on this large edifice before the present work, due to the difficult access to the mountain and its harsh rainy climate. The present paper is mainly based upon field observations made on the SW half of the volcano in the 1970’s and in December 1995 ŽMonzier et al., 1996., and subsequent petrological studies. In this paper we attempt to reconstruct the overall development of the Sangay volcanic complex and present a tentative petrogenetic model for the magmas.

2. Overall structure and development of the volcanic complex Three successive edifices ŽSangay I, II and III. comprise the volcanic complex, the former two largely destroyed by huge sector collapses. 2.1. Sangay I Remnants of the oldest edifice form a wide amphitheater open to the east, which encircles the Sangay II and III edifices and is defined by a major crestline, locally 4000 m in elevation. Numerous

Fig. 3. ŽA. Geological sketch map of Sangay volcano from fieldwork, aerial photographs and 1:50,000 topographic maps of the Instituto Geografico Militar ŽQuito.. Contour interval is 1000 m. 1 s metamorphic formations Žoften blanketed by airfall deposits from Sangay. with the main topographic crests. Successive volcanic edifices: 2 s Sangay I with its topographic crests; 3 s approximate limit of Sangay I avalanche scar; 4 s collapsed megablock during the Sangay I avalanche event; 5 s Sangay I avalanche deposits; 6 s Sangay II with its topographic crests; 7 s limit of Sangay II avalanche scar; 8 s Sangay II avalanche deposits Žthin lines s topographic highs, mainly radial, which probably represent the spreading directions of the avalanche products.; 9 s Sangay IIIA Ža parasitic center.; 10 s active Sangay IIIB; 11 s N408–458E fractures and probable dyke. The main campsite ŽLa Playa s LP s white square. is shown as well as sample sites Žsmall black dots.. NRS s Northern Rio Sangay; PC s parasitic center ŽSangay IIIa.; L s small lake; CN s Cerro Negro; Q s Quilloyacu; RVo s Rio Volcan; SRS s Southern Rio Sangay; RVi s Rio Victoria. ŽB. Sketch of the summit ridge with the summit dome, the western crater ŽWC., the central crater ŽCC. and the northeastern crater ŽNEC., from 1974 and 1979 vertical aerial photographs. ŽC. Map of samples at same scale as ŽA.; samples with Ar–Ar ages are numbered in bold, and samples corresponding to the three stratigraphic sections of Fig. 4 are distinguished.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

53

54

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

radial and secondary ridges mark the outer flanks of Sangay I, which was 15–16 km in diameter and whose summit was probably located 2–3 km to the SE of the present summit ŽFig. 3.. The C-shape of the main crestline cannot be explained simply by erosion, even if one takes into account the role of previous glaciations and the heavy rains that characterize the area. We interpret this depression as the result of a large sector collapse whose distal deposits apparently reached the Amazon lowlands. Within the avalanche depression are megablocks of volcanic units whose stratigraphy was not disturbed during sliding. For example, a large piece of the southern caldera wall slid as a megablock into the amphitheater, leaving an obvious embayment in the C1 crestline in the Quilloyacu-Cerro Negro area ŽFig. 3A..

The internal wall of this megablock was sampled ŽFig. 3A–C.. It consists of a ; 400-m-thick lava sequence with subordinate intercalations of breccias, and pyroclastic flow and lahar deposits ŽFig. 4A.. Acid andesites Ž57%–59% SiO 2 ; all analyses recalculated to 100% on a LOI-free basis. dominate the whole sequence, but some basic andesites, down to 53.6% SiO 2 , appear in the median and upper parts. This sequence is overlain by siliceous andesites Ž61% SiO 2 ; samples SAN 70, 18 and 17., but field observations show that they probably belong to Sangay II. Lavas from the lower and upper parts of this sequence have been dated by the Ar–Ar method and give ages of 380 " 70 ka and 310 " 100 ka, respectively ŽSAN 34 and 64, Fig. 4A and Appendix A.. Thus, given the thickness and continuity of the se-

Fig. 4. Representative stratigraphic sections of volcanic formations Žthese are not vertical sections, but rather composites that extend laterally up to several kilometers.: A s the Sangay I megablock, which collapsed during the first avalanche event; B s the Sangay II avalanche scar; C s the Quebrada Surayhuaycu near La Playa basecamp; see Fig. 3 for the location of these sections. Silica content Ždata normalized to 100% on a LOI-free basis. of analysed samples is shown as well as Ar–Ar ages obtained on selected samples.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

quence, and the position of the dated samples, it is thought that Sangay I volcano was probably constructed some 500–250 ka ago. 2.2. Sangay II Sangay II apparently had an E–W elongate shape. Remnants of its southern and eastern parts are well preserved within the Sangay I amphitheater ŽFig. 3.. West of the volcanic complex, thick ash deposits and few intercalated lava flows that form badlands also pertain to Sangay II. Northwards, some remnant flanks are here thought to belong to Sangay II, although it is difficult to distinguish clearly between Sangay I and Sangay II in aerial photographs. Similar to Sangay I, the history of Sangay II ended with an important sector collapse directed to the east, which resulted in a 5-km-wide C-shaped avalanche depression. The avalanche deposits form a huge fan in the Amazon lowlands, easily recognized on topographic maps and aerial views. At 20 km from the present summit, subandean hills consisting of faulted Cretaceous formations ŽLitherland et al., 1994. locally diverted or blocked the avalanche. A ; 230-m-thick continuous sequence consisting of lava flows, lahars, breccias, and pyroclastic-flow deposits has been sampled on the interior wall of the amphitheater ŽFigs. 3 and 4B.. Andesites with 56%– 59% SiO 2 largely predominate, except at the top where more siliceous andesites Ž60%–61% SiO 2 . are present. Other Sangay II deposits in the southwestern badlands and in the Quebrada Surayhuaycu near La Playa basecamp ŽFigs. 3 and 4C. were also sampled. Here, the only basalt flow of the complex ŽSAN 20B, 49.8% SiO 2 . was found. The Sangay II edifice Ž; 9 km = 16 km. was less voluminous than Sangay I and its top was probably located near the present summit. Ar–Ar dating ŽFig. 3C and 4B; Appendix A. gives an age of - 100 ka for a sample from the caldera wall ŽSAN 39B. and 60 " 20 ka for a thick, acid andesitic lava flow interbedded in the badland deposits ŽSAN 4.. Thus, the range 100–50 ka is proposed for the construction of Sangay II volcano. 2.3. Sangay III Sangay III is an almost perfect cone, with ; 358 slopes, slightly elongate on a NE–SW axis at the

55

4000-m contour ŽFig. 3A.. The summit ridge, roughly oriented W–E, is comprised of three active craters and an active dome ŽFig. 3B.. Eastwards, Sangay III deposits are largely refilling the Sangay II avalanche depression, whereas at the southern foot of the cone, they have covered the caldera wall ŽC2. and flowed down into the Sangay I amphitheater. On the west side, lava and scree deposits of Sangay III are progressively invading the eroded ash tablelands formed during Sangay II activity. Northwards, they partly cover the remnants of the two previous edifices. In addition, an important but apparently extinct parasitic vent is observed on the Sangay II amphitheater rim, located about 3 km ENE of the summit near the 3800-m contour line; unfortunately its remoteness prohibited its sampling. Within the Sangay II amphitheater, near site 65 ŽFig. 3A–C., at 3700 m elevation, moraine deposits conceal earlier Sangay III lava flows. These moraines certainly correspond to one of the peaks of the Late Glacial Stade, dated between 32 and 14 ka ŽClapperton, 1993., giving a minimum age of 14 ka for the beginning of the activity of Sangay III. As a whole, Sangay III is smaller in size Ž; 9 km = 10 km. than Sangay II. Noteworthy, is the fact that throughout its approximate 500-ka history, Sangay’s magma conduit has not moved significantly.

3. Volcanic activity of the present cone 3.1. LaÕa flows Lava and scree fans derived from fragmented lava flowed in all directions, filling the Sangay II avalanche caldera. To the ENE, these flows now overlie lavas that issued from the parasitic vent, suggesting that this vent formed shortly after the avalanche II event. To the west and north, some flows entered deep gullys carved into the ash tablelands of Sangay II. A spectacular unit is the ‘spaghetti’ lava field ŽHall, 1977., located 2.5 km NW of the main crater, between 3600 and 4200 m, which formed by the accumulation of numerous narrow and contorted lava flows, readily observed in the 1965 and 1979 aerial photographs. Its origin is attributed to two small vents located on a radial N1108E fracture observed near the 4800–5000-m

56

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

contour line, WNW of the main crater. At present, this lava field is inactive and covered by brush. 3.2. Pyroclastic flows Though apparently not as numerous as lava flows, pyroclastic flows have frequently swept the flanks of Sangay III. Pyroclastic-flow deposits are well exposed at sites 2, 3, 46, 59, 60, 65 ŽFig. 3A–C.. At site 46, a typical deposit, 10 m thick, contains juvenile bombs Ž10 vol%. and blocks Ž45 vol%., up to 1.5 m in size, in a coarse ash matrix Ž45 vol%.. Charcoal is present in some deposits Žsite 65., whereas in others, wood is barely carbonized Žsite 3. or not carbonized at all Žsite 59.. This suggests that some pyroclastic flows may be relatively cool when they reach the vegetation line at 3700–4000 m, probably due to heat loss by icersnow melting and the incorporation of steam. The La Playa campsite is located on the flat surface of a 4-km-long, 10–15m-thick pyroclastic-flow deposit that was emplaced sometime between 1956 and 1965. This deposit Žsites 2–3. is monogenetic, unsorted, and contains numerous breadcrust glassy bombs and chilled blocks. It is interpreted as the deposit of a pyroclastic flow initiated by the partial destruction of the summit dome, during an eruption similar to that of the 1979 St. Vincent event ŽShepherd et al., 1979.. In addition, this deposit is capped by a 10–30-cms thick layer of fine reddish ash, corresponding to the ash-cloud deposit. At site 3, two additional deposits of this type were observed, indicating the high frequency of explosive events at Sangay and the hazards that threaten the La Playa area, especially if dome growth and lava flow collapse persist. 3.3. Growing summit dome and resulting rockfalls An active dome Žabout 200 m = 130 m wide. forms the western end of Sangay’s summit ŽFig. 3B.. This dome is a recent structure. In 1974–1976, lava slowly issued from a fracture located at this site, but by 1983 a small dome had taken its place. Its growth during the last decade has produced constant rockfalls and occasional lava flows ŽSEAN Bulletin, July 1976, August 1983; Hall, 1977; Raeburn and Atkins, 1987; Durieux and Heiken, 1988.. This prominent dome is not glowing at night but is the source of

fumarolic activity. Its growth rate, although diminished, persists, as demonstrated by numerous rockfalls—sometimes continuing for many hours—which make the western and southwestern slopes very hazardous. 3.4. Debris flows Given the abundant rainfall and snow cover of the cone, lahar deposits are ubiquitous. They are apparently derived from pyroclastic flows generated high on the edifice that transformed to debris flows, depending upon the amount of water available. In fact, there is a continuum of characteristics from pyroclastic-flow to debris-flow deposits. The constant rockfalls from the dome as well as the accumulation of bomb and ash ejecta from the other craters provide an additional supply of material. Debris-flow deposits vary greatly in composition, from 30% clasts in an indurated ash matrix to nearly 100% clasts. Sequences of lahar deposits are readily seen in all the valleys that lead away from the cone, extending many kilometers downvalley. 3.5. Fall deposits Ash and lapilli produced during great eruptions fell mainly to the west of the cone. However, thin sequences of fall deposits apparently related to Sangay III activity are observed from place to place on the cone, but are difficult to differentiate from older sequences formed during Sangay II activity, especially in the western ash badlands. Recent Sangay III fallout sequences, only a few meters thick, are usually made up of 10–30-cm-thick layers of ash and scoria lapilli, andesitic in composition, with notable but rare white pumice lapilli beds of dacitic composition Žalso observed in Sangay II fall deposits.. Accretionary lapilli were observed once, in a thin ash layer intercalated in the recent sequence at site 23 ŽFig. 3A–C.. 3.6. Crater actiÕity The summit of Sangay has frequently changed its shape, being modified continuously by explosive activity. During the 18th and 19th centuries an apparent summit crater was responsible for the occa-

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

sionally witnessed explosive activity that distributed ash over populated areas to the W and NW. Lava flows sometimes issued from this vent Žde la Condamine, 1751.. According to Whymper Ž1892., supersonic explosions occurred at this crater in 1880: ‘‘The part of the cone within sight was about 4000 feet high; the jets rose to about once and a half this height, in less than three seconds, and they were consequently projected in the air at the rate of about twenty-two miles per minute’’. In 1946 a large single crater that occupied the entire summit and was breached to the north was the only active vent; its eruptive materials mainly descended the north flank of the cone ŽG.E. Lewis, pers. commun., 1978.. In aerial photos taken in 1956 and 1965, two unbreached craters marked the summit, a smaller NE crater Ž50 m wide. and a large central crater Ž100–150 m wide.. At that time, eruptive materials issued from both craters and descended a NNE gully. The NE crater was still steaming on 1974 photos, but was inactive by 1976. By 1975 another crater Ž50 m wide. had formed immediately to the W of the central crater and was responsible for the frequent explosions that resulted in the 1976 Sangay accident ŽSEAN Bulletin, 1976.. Concurrently a lava flow issued from a fracture located tens of meters to the W of this new explosion crater and descended 400 m down the western flank of the cone. By 1995, only small explosions, constant release of light blue gas, and occasional red glow at night occurred at the central crater, illustrating the low level of activity of the volcano at this time. The western crater was in repose and was partially covered by lava generated by dome growth from the 1975 fracture. The formation of the western crater, the extrusion of the western lava dome, and the concomitant inactivity of the northeastern crater show that the summit vents have changed frequently in recent times.

57

source edifice ŽSangay I, II or III; Table 1.. In summary, phenocryst assemblages vary from OL q CPX in the only basalt found ŽSiO 2 s 49.8%., to OL q CPX q PLAG in basic andesites Ž53 - SiO 2% - 57., and to CPX q OPX q PLAG in acid andesites Ž57 - SiO 2% - 63.. Amphibole appears in rocks with SiO 2 0 58.9% and biotite only in rocks with SiO 2 0 62.7%. In addition, Cr-spinel is a frequent inclusion in olivine, whereas titanomagnetite occurs in all rocks but the basalt. Phenocryst Žand a few microlite. compositions have been determined using a Camebax electron microprobe operating at 15 kV Ž12 nA.y1 10 sy1. Average mineral analyses for the basalt and selected mineral analyses for the andesites are given in Table 2. Chemical composition plots for olivine, plagioclase, and the pyroxenes are presented in Fig. 5. 4.1.1. Basalt SAN 20B The phenocryst and microphenocryst assemblage in this rock is dominated by euhedral Mg-rich olivine ŽFo 90 – 86 ; up to 1 mm; est. vol. ; 7%. showing a slight normal zonation. Mean CaO and NiO contents are 0.13% and 0.33% respectively. Olivine frequently includes euhedral Cr-spinel with about 21% Al 2 O 3 , 38% Cr2 O 3 , and a Cra Žs 100 CrrCrq Al; atomic ratio. of about 55. Thus, olivine crystals are not inherited xenocrysts, but true magmatic phenocrysts in equilibrium with the host basaltic melt ŽSimkin and Smith, 1970; Nye and Reid, 1986.. Clinopyroxene Žup to 0.9 mm but usually 0.4–0.5 mm; est. vol. ; 3%. is euhedral, Ca-rich ŽWo 47 – 44 ., with 0.43% Cr2 O 3 , and mainly plots on the diopside-salite boundary. Zoning is highly variable, either normal or inverse. The fine-grained matrix of basalt SAN 20B consists of intersertal and interstitial clinopyroxene and plagioclase. Microlites of clinopyroxene have compositions similar to those of the phenocrysts and microphenocrysts. Plagioclase microlites range from An 71 to An 63 in composition.

4. Petrological data 4.1. Petrography and mineral compositions The mineralogy of Sangay’s rocks is quite constant for a given silica content, independent of the

4.1.2. Andesites Andesites from Sangay are commonly porphyritic, with a variable glassy matrix. Olivine phenocrysts are present in rocks with SiO 2 contents( 57%, occasionally in those ranging from 57 to 60.5%

58

SAN

20B 21 69 8 46A 40 37B 41 31 30D 20A 55 70 48

Sangay

II? II? I III III II I II I III II? II I II?

SiO 2 Ž%.

MgO Ž%.

Cr-SPIN a ŽCr2 O 3 %.

OLIVINE ŽFo %.

49.8 53.6 54.7 55.9 56.2 57.0 57.0 58.4 58.9 60.4 60.6 61.3 61.3 62.7

11.0 5.0 4.9 4.4 4.9 4.2 4.7 3.8 3.2 2.9 2.4 2.8 2.3 1.9

31–40 Ž25.

86–90 Ž61. 69–81 Ž4. 72–78 Ž3. 74–78 Ž7. 80–82 Ž13. 81–84 Ž9. 75–83 Ž27.

14–31 Ž4. 14–32 Ž6. 11–25 Ž7.

75–76 Ž2. 78 Ž1.

Ti-MAGN ŽTiO 2 %. 10–16 c Ž3. 9–10 c Ž3. 8–10 c Ž3. 8–9 c Ž2. 9 a Ž2. 8 c Ž1. 11c Ž1. 4–5c Ž4. 7 c Ž1. 7–8 c Ž3. 9–13 c Ž5. 6 c Ž1.

CPX

OPX

En %

Wo %

Fs %

39–46 43–48 42–48 43–46 43–47 42–45 42–50 43–46 44–47 42–47 45–48 44–48 41–45

44–47 40–46 39–45 40–44 37–45 41–45 40–45 39–44 39–44 37–44 41–43 39–44 39–41

9–14 Ž21. 10–16 Ž8. 9–17 Ž11. 11–16 Ž11. 9–17 Ž12. 13–17 Ž12. 9–15 Ž29. 11–16 Ž11. 10–14 Ž14. 12–16 Ž9. 11–13 Ž5. 12–14 Ž8. 13–18 Ž7.

En %

76–78 69–71 72–76 71–75 74–77 69–78 62–72 72–73

PLAG ŽAn %.

Wo %

Fs %

2–3 3–4 2–3 2–4 3 2–3 1–4 2

63–71b Ž10. 45–69 Ž12. 50–63 Ž11. 43–72 Ž21. 50–72 Ž14. 42–57 Ž14. 19–21 Ž2. 44–73 Ž15. 26–27 Ž4. 41–67 Ž12. 21–26 Ž3. 42–59 Ž11. 21–27 Ž9. 38–61 Ž14. 20–24 Ž2. 35–73 Ž13. 19–29 Ž11. 41–56 Ž8. 25–37 Ž12. 46–77 Ž12. 25–26 Ž4. 35–79 Ž12.

Rocks are ordered by increasing silica content Žwhole-rock SiO 2 and MgO contents are given., independent of the source edifice ŽSangay I, II or III.. a In OLIVINE. b Microlites. c In CPX or OPX or free. Values in parenthesess number of analyses.

AMPH ŽMgO %.

BIOT ŽMgO %.

14–16 Ž8. 15 Ž3. 15–16 Ž8. not anal. 14–15 Ž5.

16 Ž5.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Table 1 Summary of phenocryst mineralogy Žplagioclase microlite range is shown for basalt SAN 20B as phenocrysts are absent in this rock.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

59

Table 2 Selected analyses of phenocrysts

Averages and standard deviations are given for the basalt SAN 20B Žin which plagioclase appears only as microlites.. For each mineral, compositions are ordered by increasing silica content in the host rocks; c s core, i s intermediate zone, r s rim.

60

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Fig. 5. Mineralogy of Sangay olivine, clinopyroxene, orthopyroxene and plagioclase phenocrysts Žsee Table 1Table 2.. For basalt SAN 20B: Ž1. plagioclase microlites are employed, since plagioclase phenocrysts are absent, Ž2. a few clinopyroxene microlites have been included with the clinopyroxene phenocrysts, their composition being similar.

SiO 2 , and absent in more siliceous rocks. Compositions range from Fo 84 to Fo 69 , but no obvious tendency towards less magnesian olivine is noted with increasing silica content of the rocks. Normal zoning is common, as are Cr-spinel inclusions with variable Cr2 O 3 contents Ž32%–11%.. Rare titanomagnetite inclusions occur in some olivines from basic andesites. Titanomagnetite microphenocrysts are ubiquitous ŽTiO 2 contents 16%–4%., either free or included in pyroxenes. Clinopyroxene Žaugite; Wo 46 – 37 . is present in all andesites from Sangay, except the most silica-rich acid andesites ŽSAN 48 with 62.7% silica.. It frequently includes apatite crystals and shows highly variable zoning. Orthopyroxene Žnormally zoned bronzite; Fs 19 – 30 . appears in the acid andesites. A few unusual hypersthene compositions are also observed ŽFs 34 – 37 .. Plagioclase is present in all andesites from Sangay and have compositions ranging from An 79 to An 35 . Compositional zoning, either normal, reverse or complex, is important, varying up to 38 An units in acid andesites, in which, surprisingly, the more calcic cores are found Ževidence of magma mixing..

However, the plagioclase compositions show a definite tendency on average, towards less calcic compositions as the silica content of the host rock increases. Amphibole is also present in the acid andesites. According to the classification of Tindle and Webb Ž1994., it is a Žtitanian-. magnesio-hastingsite or a Žtitanian-. magnesio-hastingsitic hornblende Ž16%–14% MgO.. Biotite microphenocrysts contain 16% MgO and are restricted to rocks with SiO 2 contents ) 62%. 4.2. Whole-rock geochemistry Silica contents in Sangay rocks vary from 49.8% to 68.1% ŽTable 3., covering the entire compositional range from basalts to dacites. However, a histogram of silica contents ŽFig. 6. shows that basic to acid andesites predominate Ž53% to 63% SiO 2 ., with a frequency maximum at 55%–57%. 4.2.1. Basalt SAN 20B This basalt Ž49.8% SiO 2 and 1.3% K 2 O. plots on the boundary between the medium-K and high-K compositional fields of orogenic volcanic rocks

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79 Table 3 Whole-rock geochemistry of Sangay I, II and III

61

62 Table 3 Žcontinued.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

ŽFig. 6.. Its high Mg, Ni, and Cr contents Ž11.0% MgO, 273 ppm Ni, 410 ppm Cr., its low FeOU rMgO ratio Ž0.78., and its Al 2 O 3 content Ž14.7%. correspond to quite primitive characteristics. Olivine phenocryst compositions ŽFo 90–86. and the bulk rock FeOUrMgO suggest that this basalt was roughly in equilibrium with mantle material, and not affected by olivine accumulation ŽRoeder and Emslie, 1970; Nye and Reid, 1986.. Such a Mg-rich basalt is exceptional for the Andean Quaternary volcanic arc. Indeed, all Quaternary basalts known in the Southern Volcanic Zone ŽSVZ. of the Andes have MgO contents less than 7.5% ŽDeruelle, 1982; Futa and Stern, 1988; HickeyVargas et al., 1989; Lopez-Escobar et al., 1993., whereas the most basic andesites from Quaternary volcanism of southwestern Colombia ŽDroux and Delaloye, 1996. and Ecuador Žour unpublished data. do not exceed 7% MgO. However, Pleistocene lavas erupted from cinder cones of San Francisco volcano in the southernmost Central Volcanic Zone ŽCVZ. have primitive characteristics quite comparable to those of SAN 20B with 8.9% MgO, 1.9% K 2 O, and Mg-rich olivine ŽFo 88 – 89 ., even though their SiO 2 contents Ž53.0%. are unusually high for such primitive characteristics, an anomaly interpreted as the result of mixing between primitive basalts and silicic magmas ponded in the crust ŽKay et al., 1996.. By removing the silicic component, these authors obtain a primitive magma fairly similar to SAN 20B. The other Andean basic rock that is comparable to basalt SAN 20B, though older and clearly less siliceous, is the alkali basalt of Chiar Kkollu, a Miocene sill on the Bolivian Altiplano behind the volcanic front ŽDavidson and de Silva, 1995; olivine phenocrysts Fo 53 – 84 , 45.2% SiO 2 , 14.2% Al 2 O 3 , 1.3% K 2 O, 9.3% MgO, 129 ppm of Ni, 433 ppm of Cr, FeOU rMgOs 1.11..

63

Incompatible element abundances for Sangay rocks normalized to an N-type MORB ŽFig. 7. display a sloping pattern that increases progressively from low HREE to high Ba and Rb levels. However, several relative anomalies are evident, in particular a pronounced negative Nb anomaly Žaccompanied by weak negative Zr and Ti anomalies., and positive anomalies in Sr, Ba Žand, to a lesser extent, K.. Isotopic ratios of 87 Srr86 Sr s 0.704097 and 143 Ndr144 Nd s 0.512778 have been determined for this basalt ŽE. Bourdon, pers. commun., 1998.. These data will be discussed below. For comparison, N-MORB normalized patterns for San Francisco and Chiar Kkollu primitive lavas, those for a basalt from Aoba ŽNew Hebrides arc, Southwest Pacific. and for a Nb-enriched basalt from Kamchatka are also given in Fig. 7. These patterns are similar in shape to that of SAN 20B, but several differences are noteworthy: Ž1. Y and HREE are obviously low in the Sangay basalt; Ž2. the Nb content in SAN 20B is intermediate between that of an oceanic arc basalt ŽAoba. and a Nb-enriched arc basalt ŽKamchatka.; also, note that the alkali basalt from Chiar Kkollu does not show any depletion in Nb; Ž3. LREE abundances in SAN 20B are nearly similar to those in the oceanic arc basalt and Nb-enriched arc basalt, but nearly two times less than those in the primitive San Francisco basic andesite and Chiar Kkollu alkali basalt, in spite of quite similar major and trace element contents ŽY and HREE excepted.. 4.2.2. The Sangay magmatic suite Most analyzed rocks from Sangay are andesites Ž40 basic andesites and 37 acid andesites; Table 3.; only four dacites have been found, represented as pumice lapilli, having silica contents of 63.5%– 68.1% ŽFig. 6.. Most major and trace elements vs.

Notes to Table 3: Analyses by ICP–AES carried out at the University of Brest, France; analytical technique is detailed in Cotten et al. Ž1995.. All compositions recalculated to 100% on a LOI-free basis ŽLOI s loss on ignition at 10508C.. LOI and prenormalization total are given. A s Sangay I samples from the cross section of Fig. 4A; B s Sangay II samples from the cross section of Fig. 4B; C s Sangay II samples from the Quebrada Surayhuaycu cross section ŽFig. 4C.; D s other Sangay II rocks; E s Sangay III rocks; E1 s a subgroup of recent products; E2 s a homogeneous subgroup of very recent products from the summit area ŽA, B and C in stratigraphic order, E, E1 and E2 in increasing SiO 2% order.. LF s lava flow; B-PF s juvenile block or bomb in a pyroclastic flow deposit; B-BDFs block in a block-rich debris flow deposit; B-L s block in a lahar deposit; B-F s juvenile block in a fall deposit; WPL-Fs white pumiceous lapilli in a fall deposit; BSL-Fs black scoriaceous lapilli in a fall deposit; B s recently fallen dome block or bomb from the summit of Sangay III.

64

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

SiO 2% correlations display partially coherent trends, generally originating from the basalt, and apparently related to the fractionation of the observed mineral phases. As a whole, independent of the edifice they came from or their age, Sangay rocks have similar compositions and display a similar evolution, and therefore are considered to belong to a single high-K calc-alkaline suite. The Sangay suite has the highest K 2 O, Na 2 O, P2 O5 , Rb, Ba, La, Th, and Sr contents of all the presently known suites from the NVZ ŽFig. 6.. This is consistent with the general increase in K 2 O Žand other incompatible elements. observed away from the trench for the Quaternary Ecuadorian volcanics ŽBarberi et al., 1988; Kilian and Pichler, 1989; Monzier et al., 1997a., Sangay being one of the more distant volcanoes from the trench ŽFig. 2.. TiO 2 Žnot shown., Al 2 O 3 , and MgO abundances are normal for the NVZ. In contrast, Fe 2 OU3 and CaO contents Žthe latter not shown. are the lowest observed in the NVZ. Sr and Ba contents are unusually high in Sangay rocks. This is especially true for Sr whose abundances are almost two times that observed in other NVZ rocks, with the exception of Sumaco, an alkaline back-arc volcano in the area ŽBarragan et al., 1998.. However, such Sr values are common in andesites from the CVZ ŽDeruelle, 1982; Davidson et al., 1990; de Silva et al., 1993.. Nb contents in the Sangay suite are also the highest observed in the NVZ, moreso the unusually high abundance of this element in the basalt. Zr and Yb are high but not exceptional for the NVZ and, in spite of some scattering, the relative constancy of Yb in the whole Sangay suite is notable. More generally, the compatible behaviour of Yb in all suites from the NVZ is remarkable. Zr show a similar behaviour, but only for a part of the NVZ data set. Lastly, Ni contents in Sangay rocks are among the highest of the NVZ, whereas Cr abundances are more usual, or even low. In addition, Sc and V contents Žnot shown. are unusually low.

Incompatible element abundances in the basic andesites are, in general, slightly higher than those of the basalt SAN 20B, except for Nb whose values are always lower than that of the basalt ŽFig. 7.. Incompatible element abundances in the acid andesites are only slightly higher than those in basic andesites, and again, the Nb values are lower than that of the basalt. Once more, the unusually high content of this element in the basalt must be emphasized ŽSAN 20B has been analyzed two times to confirm the accuracy of this high Nb concentration.. Only the dacites show distinctly higher amounts of Rb, Ba, Th, K and Zr and lower amounts of Sr, P, Eu and Ti, these latter probably being related to an increase in plagioclase, apatite, and titanomagnetite fractionation.

5. Discussion 5.1. Repeated sector collapses Sector collapses of stratovolcanoes are commonly influenced by regional tectonic stress fields, which control dyke orientation and emplacement. Edifice collapse usually occurs in a direction perpendicular to these dykes, and is often triggered by weakening caused by repeated magma intrusion, hydrothermal alteration, and subsequent gravitational instability ŽSiebert, 1984; Johnson, 1987; Francis and Wells, 1988; Kokelaar and Romagnoli, 1995.. The two sector collapses that partially destroyed Sangay I and II were influenced by dominant NE–SW and lesser NW–SE fractures, both trends being very apparent in the morphology ŽFig. 3.. Sangay I’s avalanche must have been volumetrically important, given the size of the resulting amphitheater. In both cases, large megablocks slid a short distance downslope and stopped within the avalanche depression, forming ridges subparallel to the caldera rim. The NE–SW elliptical shape of Sangay III, as observed between the 4000 and 5000 m contours, can also be interpreted as the result of NE–SW-trending dykes that

Fig. 6. Selected diagrams for element content vs. SiO 2% for Sangay and Northern Volcanic Zone volcanic rocks Žall compositions recalculated to 100% on a LOI-free basis.. The subdivisions shown in the K 2 O vs. SiO 2 diagram are from Gill Ž1981.: B s basalt, BA s basic andesite, AA s acid andesite, D s dacite, R s rhyolite, LK s low-K, MK s medium-K, HK s high-K. The histogram of SiO 2 content is also shown. Data for the NVZ are from Droux and Delaloye Ž1996. for Colombia, and from IRD Ž378, largely unpublished, analyses. for Ecuador. Basalt sample SAN 20B is distinguished on all diagrams.

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

65

66

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Fig. 6 Žcontinued..

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

67

Fig. 7. N-MORB-normalized incompatible element abundance patterns for Sangay rocks. For comparison with basalt SAN 20B, several N-MORB normalized patterns are shown, including a primitive lava from San Francisco volcano ŽKay et al., 1996., an alkali basalt from Chiar Kkollu ŽDavidson and de Silva, 1995., a basalt from Aoba ŽMonzier et al., 1997b. and a Nb-enriched basalt from the Kamchatka ŽKepezhinskas et al., 1996.. Element incompatibility relative to a typical lherzolite decreases from left to right. Concentrations in N-MORB are from Sun and McDonough Ž1989..

supplied the magma during the construction phases of the cone. Inflation of such dykes, combined with

excessive loading and the chronic gravitational instability due to the volcano’s location at the top of the

68

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

steep and very high Ž3 to 4 km. eastern flank of the Cordillera Real, could certainly explain the two successive collapses of its unbuttressed eastern flank onto the Amazon plain. Frequent, large earthquakes occurring on active thrust faults along the foot of the Cordillera Real ŽYepes et al., 1996. may also have played a role in the triggering of these collapses. Because Sangay III is becoming very high, its gravitational instability is increasing, consequently another avalanche is probable in the future. 5.2. SAN 20B basalt source In a first approach, the straight-sloping line defined by normalized Nb, Zr, Ti, Y, and HREE abundances—elements mainly controlled by mantle wedge melting ŽDavidson, 1996. —could be considered as indicative of an enriched mantle source in comparison with a MORB source ŽFig. 7.. Moreover, the positive anomalies above this line might reflect the addition of a slab component able to supply large amounts of Ba, Rb, Th, K, and Sr, low amounts of LREE, and almost no Eu and Gd. Radiogenic Sr and Nd isotope values for SAN 20B place it within the mantle array, near the center of the OIB field ŽZindler et al., 1982., supporting the hypothesis of an enriched mantle source. However, additional data ŽPb isotopes for example. are required to conclusively identify a pre-subduction mantle composition and to assess the role and importance of crustal contamination. The unusually high Nb abundance in the basalt is obvious on a Nb vs. La diagram ŽFig. 8., two highly incompatible elements that display a linear trend with a high correlation coefficient for the whole Andean NVZ, except for the most acidic compositions. Considering this tight relationship, the Nb content in basalt SAN 20B should be 6 ppm instead of 9 ppm, suggesting a moderate Nb enrichment. In spite of this moderate Nb enrichment, it should be emphasized that SAN 20B still clearly shows the negative Nb anomaly relative to adjacent elements in the spider diagram ŽFig. 7., typical of arc rocks. This reflects considerable transfer of fluid-mobile elements from the slab to the mantle wedge ŽTatsumi et al., 1986; Davidson, 1996.. Metasomatism of an enriched mantle source by such a subduction component, followed by partial melting of this modified

Fig. 8. Nb vs. La diagram. Data as for Fig. 6. The relation Nbs 0.31 La is valid only for La - 25 ppm.

mantle material, would produce basalts like SAN 20B, with an unusually high Nb abundance but still with a negative Nb anomaly relative to adjacent elements. 5.3. Petrogenesis of the Sangay suite In spite of the strong compositional affinities between SAN 20B and Sangay’s andesites, these latter, given their lower Nb contents cannot be produced by common differentiation processes from SAN 20B. However, a parental basalt similar to SAN 20B, but having a lower Nb content, would be a good candidate for andesite generation. 5.3.1. Low-pressure fractionation Fig. 9 shows log–log diagrams for trace elements plotted against Th, assuming that Th acts as the most incompatible element in the entire Sangay suite. Disregarding the limited scattering, these diagrams display linear trends with variable slopes. Only Rb shows a similarly high degree of incompatibility in the Sangay suite, the other elements showing decreasing incompatibility in the order K 2 O ) Ba ) Zr ) Nb ) La ŽDi’s - 1., or increasing compatibility in the order Yb - Sr and Dy - Co and Sc - Ni and Cr ŽDi’s ) 1.. If most of these apparent distribution coefficients are fairly consistent with low-pressure fractionation processes involving mainly a PLAG q

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

69

Fig. 9. Log–log diagrams for trace elements plotted against Th.

OL q CPX q OPX q Ti-MAGN assemblage, the Di of others, such as Yb, Y, and Dy, are clearly inconsistent with such processes and would require the major intervention of phases like garnet and hornblende. The prevailing role of low-pressure fractionation processes during the petrogenesis of the Sangay suite is implied by the Al 2 O 3 vs. SiO 2 and CaO vs. MgO diagrams ŽFig. 10., where all whole-rock analyses and the full data set of mineral analyses are plotted together. On these diagrams, it is obvious that the removal of the observed mineral phases would explain most of the noted characteristics of the suite, with fractionating mineral assemblages evolving in the classical order OLIV q CPX, OLIV q CPX q PLAG, PLAG q CPX q OPX, PLAG q CPX q OPX q AMPH. Note also that the basalt SAN 20B plots close to the primary arc magma field proposed by Davidson Ž1996.. Thus, low-pressure fractionation plays an important role in the petrogenesis of the high-K calc-alkaline Sangay suite. However, some unusual characteristics of this suite need further explanation. 5.3.2. Y and HREE depletion When compared with rocks from an intraoceanic arc Žthe New Hebrides island arc, Southwest Pacific.

and from the 398–468S segment of the SVZ of the Andes, the Sangay suite as well as all suites from the Northern, Central, and Austral Volcanic Zones ŽAVZ., appears to be very depleted in Y and HREE ŽFig. 11, where interestingly, the only rock of the Sangay suite having normal Y and HREE contents is the basalt SAN 20B.. Furthermore, these low Y and HREE values remain roughly constant while SiO 2 increases, or even tend to decrease for some NVZ and CVZ suites, a compatible behaviour certainly explained by low-pressure amphibole fractionation at SiO 2 values above 57%. In contrast, Y and HREE behave clearly as incompatible elements for the suites from the 398–468S segment of the SVZ and from the New Hebrides arc. An adakitic model has been proposed by Stern and Kilian Ž1996. to explain the depletion in Y and HREE observed in AVZ rocks, especially in the Cook Island volcanics. As Sangay is located at ; 350 km from the trench and ; 130 km above a rather old Ž) 32 Ma. downgoing slab, such an adakitic model is unlikely to explain the Y and HREE depletion observed in this suite. In addition, the Sangay suite is a high-K, not a low-K suite typical of adakites. As a consequence, Sangay’s andesites and dacites are not adakites, even if they show some adakite-like geochemical characteristics.

70

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Fig. 10. SiO 2 vs. Al 2 O 3 and MgO vs. CaO diagrams showing Sangay phenocryst and whole-rock ŽWR, black diamonds. compositions. Phenocrysts ŽOLIV, CPX, Cr-SPIN. and plagioclase microlites from basalt SAN 20B are represented by black symbols, whereas phenocrysts from all other rocks are represented by open symbols. The primary arc magma field proposed by Davidson Ž1996. is also shown.

In contrast, the lack of Y and HREE enrichment in Quaternary andesites–dacites–rhyolites from the Andean CVZ is generally attributed to garnet and hornblende stability in the lower continental crust

whose thickness attains 60–70 km ŽDrummond et al., 1996.. Conversely, volcanic products in a thin crustal environment Ž; 30–35 km., like that of the New Hebrides island arc ŽMonzier et al., 1997b. or

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

71

Fig. 11. SiO 2 vs. Y ppm diagram. Data for the Andean Northern Volcanic Zone ŽNVZ. are from Droux and Delaloye Ž1996. for Colombia, and from IRD Ž378, largely unpublished, analyses. for Ecuador. Data for the Andean Central Volcanic Zone ŽCVZ. are from Deruelle Ž1982., Davidson et al. Ž1990. and de Silva et al. Ž1993., those for the 398–468S segment of the Andean Southern Volcanic Zone ŽSVZ. from Deruelle Ž1982., Futa and Stern Ž1988., Hickey-Vargas et al. Ž1989., and Lopez-Escobar et al. Ž1993., and those for the Andean Austral Volcanic Zone ŽAVZ., including Cook island adakites ŽCIA., from Stern and Kilian Ž1996.. Data for the New Hebrides Arc ŽNHA; Southwest Pacific. are from Monzier et al. Ž1997b..

that of the 398–468S segment of the Andean SVZ, show normal Y and HREE enrichment. The processes often mentioned to explain this non-enrichment in the CVZ suites include the AFC interaction of basaltic magma with garnet-bearing lower crust ŽFeeley and Hacker, 1995. or mixing of basaltic magma with lower crustal melts ŽDavidson et al., 1990.. In the same way, the decreasing HREE enrichment at the northern termination of the Andean SVZ, where crustal thickness increases from 30–35 km to 50–60 km, is attributed by Hildreth and Moorbath Ž1988. to lower crustal contributions that leave garnet-bearing residues. High-pressure fractionation of hornblende andror garnet could be another effective process to suppress HREE enrichment, but petrographic evidence is generally lacking, perhaps lost during magma ascent through thick crustal sections. However, in the Sangay suite, no marked negative anomaly is observed for MREE ŽGd to Er. compared to HREE ŽFig. 7., thus precluding a

dominant deep-crustal hornblende fractionation process ŽHildreth and Moorbath, 1988.. 5.3.3. High Ni and low Cr r Ni contents Ni contents in Sangay rocks are among the highest observed in the NVZ, whereas Cr abundances look normal, or even low. In fact, the entire Sangay suite has an unusually low CrrNi value of ; 1.5, whereas CrrNi in all other NVZ rocks range from 2 to 5 ŽFig. 12.. For comparison, Aoba picritic basalts ŽNew Hebrides island arc; Eggins, 1993. have CrrNi ranging between 2 and 3. The dominant Mg-rich olivine in the Sangay basalt, its high NiO content Ž0.33% on average., as well as the unusually high Ni and the low Fe 2 OU3 , CaO, Sc, V and Cr contents and the low CrrNi of the whole suite point toward a picritic assemblage with a deficit in clinopyroxenerelated elements. It seems that only petrogenetic processes favouring clinopyroxene restites are likely to produce such characteristics.

72

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

Fig. 12. Ni vs. Cr ppm diagram. Data as for Fig. 6.

5.3.4. Petrogenetic model A preliminary petrogenetic model for Sangay magmatism is proposed here, mainly based on Ni vs. Sr behaviour. Such a model should account for the interrelationship between the basalt and the andesites, the dominant low-pressure fractionation processes, and the unusual features of the Sangay suite. Furthermore, as the Sangay rocks—the basalt excepted—and most NVZ and CVZ Quaternary volcanic rocks appear to be enriched in SiO 2 for a given MgO content, as compared to volcanic rocks from the New Hebrides arc or to those of the 398–468S segment of the SVZ, relevant crustal processes must also be integrated into this model. The model assumes that the crust present under the Ecuadorian arc is thicker than 50 km ŽPrevot ´ et al., 1996.. The base of the crust is clearly in the garnet stability field and thus the Y and HREE depletion observed in Ecuadorian, as well as in all NVZ volcanics, can be attributed to the interaction between rising basaltic melts with either garnetbearing lower crust or with lower crustal melts that leave behind garnet-bearing restites. Such lower crustal processes have been proposed for andesites erupted through thick crusts; they apparently work under the whole Andean Cordillera, except for its southernmost segments. The model is also partly founded upon melting processes of basaltic rocks at the base of a thick crust. For example, the partial melting of altered

Cretaceous MORB accreted tectonically to the lower crust is proposed by Kilian et al. Ž1995. to explain some characteristics of the Chimborazo volcanics. Another possibility is the partial melting of newly underplated basaltic material, leaving garnet, clinopyroxene, and amphibole restites, as proposed by Petford and Atherton Ž1996. for the genesis of the Mio-Pliocene Cordillera Blanca Batholith ŽPeru.. This batholith, located 300 km from the trench, intrudes the thickened keel of the Andes, where crustal thicknesses attain 50 km. Stage 1 of their model involves the creation of a mafic underplate by partial melting of slightly enriched upper mantle: during stage 2 this newly accreted material is itself partially melted. On the Ni vs. Sr diagram ŽFig. 13., Sangay compositions plot in a roughly triangular area, whose corners correspond respectively to basalt SAN 20B ŽSiO 2 s 49.8%, Sr s 970 ppm, Ni s 273 ppm., to andesite SAN 31 ŽSiO 2 s 58.9%, Sr s 1415 ppm, Ni s 43 ppm., and to dacite SAN 27A ŽSiO 2 s 68.0%, Sr s 620 ppm, Ni s 12 ppm.. The use of the Ni–Sr pair is preferred to others such as Ni–Rb or Ni–Ba, because of its very contrasted behaviour during high-pressure vs. low-pressure processes. The peculiar composition of SAN 31 should be emphasized, this andesite having by far the highest Sr content of any Sangay rock ŽFig. 6.. The model assumes the presence of an enriched mantle Žin relation to an MORB source. at the Sangay site Ž; 350 km from the trench., prior to Quaternary volcanism. The subduction of the Grijalva fracture zone and other related fractures ŽFig. 2B., with their presumed high volumes of water-rich serpentinized oceanic crust and trapped sediments, should have produced a slab with unusually high fluid contents ŽSinger et al., 1996.. Thus, in our tentative model, important amounts of slab-derived fluids, enriched in soluble incompatible elements, migrate to and react with the overlying mantle wedge. Subsequent partial melting of this metasomatized mantle could lead either to Ž1. basaltic melts similar to SAN 20B with NbrLa values typical of the NVZ that are parental to Sangay’s andesites or Ž2. by lower degrees of melting, to SAN 20B-like, high NbrLa, melts ŽNb being more incompatible than La.. An alternative would be to consider a variably enriched mantle at the Sangay site, prior to volcan-

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

73

Fig. 13. Preliminary petrogenetic model for the Sangay suite. Rayleigh fractionation equation used for fractional crystallization modeling, batch partial melting equation used for partial melting modeling, and DePaolo Ž1981. equation used for AFC process calculations. Curve FC represents fractional crystallization of basalt SAN 20B, extracting CPX q OLIV q Cr-SPIN phenocrysts in the respective proportions 55.5–25.9–18.7; percentage remaining liquid values are given. Curve AFC represents concurrent assimilation and fractional crystallization for a contaminant with Ni s 42 ppm and Sr s 3283 ppm corresponding to 10% of partial melting of SAN 20B, with a mass assimilationrcrystallization rate r s 0.5, and extracting CPX q OLIV q Cr-SPIN phenocrysts in the respective proportions 55.5–25.9–18.7; percentage of initial magma remaining is given. Curve PM represents batch partial melting of basalt SAN 20B with a residual mineralogy AMPHq GARNETq CPX in the respective proportions 49.2–32.9–17.9; percentage of partial melt is given. Crystalrmelt partition coefficients used for these calculations are from Martin Ž1987 and personal communication, 1997.. Calculated curves for low-pressure fractional crystallization paths, extracting PLAGq CPX q OPX q MAGN, between pairs of samples SAN 46A–SAN 70 Ž57.1–14.8–22.8– 5.3., SAN 31–SAN 20A Ž54.3–21.3–16.3–8.1. and SAN 20A–SAN 48 Ž71.1–7.9–14.8–6.2. are also shown; percentage of remaining liquid is depicted at 5% intervals. Values in parentheses correspond to the SiO 2% content of identified samples.

ism—as suggested by Thompson et al. Ž1984. for convecting mantle wedges above subducted slabs—a characteristic perhaps linked to the location of this volcano at the end of the NVZ. Metasomatism of such an heterogeneous source followed by partial melting could generate both basaltic melts parental to Sangay’s andesites and high NbrLa, SAN 20Blike, basaltic melts. Lower-crustal, high-pressure fractionation ŽCPX q OLIV q Cr-SPIN. of the parental basaltic magmas Ži.e., those with ‘normal’ NbrLa values for the

NVZ. may directly account for the formation of some andesitic magmas ŽFig. 13., but does not explain their low Y and HREE contents. Partial melting of arc basaltic material, underplated to the base of the crust, due to heating by new magma influx rising from the mantle, could be an alternative process for generating andesitic magmas with unusually high-Sr contents. Such remelting processes, leaving an AMPHq GARNETq CPX residue, would also explain the low Y and HREE contents of these andesitic magmas, but the high degree of partial melt-

74

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

ing required to obtain SAN 31-like compositions is not readily conceivable. More likely, lower crustal AFC processes, including the concomitant crystallization of parental Ži.e., near primitive. basalts and the incorporation of partial melts from compositionally similar basaltic materials, would explain most characteristics of the Sr-enriched, low Y and HREE andesitic magmas. The calculated AFC curve for a process involving a contaminant obtained by 10% partial melting of SAN 20B-type basalt and an assimilationrcrystallization rate of 0.5 seems in good agreement with the data ŽFig. 13.. Such an assimilationrcrystallization rate seems plausible for processes that are supposed to occur at the base of the crust in high heat flow regions. Andesite SAN 31 would be the result of such an AFC process, with about 90% of the initial magma remaining. For these calculations, minor variation of Ni and Sr in the basaltic source Že.g., "10% as shown in Fig. 13. may shift the curves significantly; the proposed AFC process may thus account for most of the andesitic compositions that plot in the arcuate area between SAN 20B and SAN 31, on the right side of the triangular field. Subsequent low-pressure fractionation and mixing processes would lead to the compositional range observed in the Sangay suite. Moreover, isotopic data ŽE. Bourdon, personal communication, 1998. show no upper crustal contamination, which implies that the underlying Palaeozoic–Jurassic formations ŽLitherland et al., 1994. do not participate in the genesis of this suite. The only basalt found at Sangay has a high NbrLa value. Why no basalt with ‘normal’ NbrLa values, parental to andesites, has been found, is certainly an intriguing question. In the hypothesis of a uniformly enriched mantle source, all parental basaltic melts might have been involved in subsequent deep crustal processes at work directly under Sangay, leading to the formation of andesites. Whereas high-Nb basaltic melts, resulting from lower degrees of melting at the periphery of the main zone of magma formation, could have infrequently reached the surface, bypassing the deep crustal zone where AFC processes were at work ŽSAN 20B being the only rock of the Sangay suite having normal Y and HREE contents; Fig. 11.. In the case of a variably enriched mantle source, this question is unanswer-

able and thus, until there is additional data, the hypothesis of an uniformly enriched mantle source seems the most reasonable. Lastly, the low CrrNi of the Sangay suite—an unique characteristic in the Andean NVZ—suggests unusual fractionation processes involving Cr-spinel andror clinopyroxene either at the base of the crust or in the upper mantle.

6. Conclusions The development of the Sangay volcanic complex saw the building of three successive edifices of decreasing volume, Sangay I, II and III. The construction of Sangay I occurred in the age range of 500–250 ka, whereas that of Sangay II is thought to have happened between 100–50 ka. Each of these former cones was partially destroyed by collapse of its unbuttressed eastern side, which resulted in a large avalanche that slid onto the Amazon plain. The active cone of Sangay III is at least 14 ka old and presently reaches 5230 m in elevation. During this 500-ka interval, Sangay’s magmatism has been dominated by andesitic compositions with only occasional dacitic activity, a significant aspect given the long time period. We attribute this behaviour to an almost constant supply of deep magma —well illustrated by the permanent eruptive activity observed ever since 1628—apparently related to Sangay’s location at the southern end of the NVZ and close to a major tear in the subducting slab. During historical times its activity has been mainly characterized by strombolian behaviour. A tentative five-stage petrogenetic model is proposed for the Sangay suite in order to explain its unusual chemical characteristics Žhighest K 2 O, Na 2 O, P2 O5 , Rb, Sr, Nb, Ba, La, Th contents of any NVZ suite, low Y and HREE contents of the andesites and dacites when compared to similarly differentiated rocks from thin crust arcs, and high NbrLa of the only basalt found.. Stage 1 considers the existence of an enriched mantle Žin comparison with a MORB source., or maybe a variably enriched mantle, at the site of the Sangay, prior to Quaternary volcanism. Stage 2 corresponds to the metasomatism of this mantle by important amounts of slab-derived fluids enriched in soluble incompatible elements, a

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

consequence of the subduction of major oceanic fracture zones under the area. Stage 3 corresponds to the melting of this metasomatized mantle, leading to primitive basaltic melts with ‘normal’ NbrLa values, that are parental to the entire Sangay suite but never reach the surface. Subordinate high NbrLa basaltic melts are possibly produced by lower degrees of melting at the periphery of the main zone of magma formation, and only infrequently reach the surface. Stage 4 corresponds to AFC processes working at the base of the 50-km-thick crust, where parental basaltic melts pond and crystallize while assimilating remelts of previously underplated, compositionally similar basaltic material. This step accounts for the generation of Sr-enriched andesitic magmas with low-Y and -HREE contents. Lastly, during stage 5, low-pressure fractionation of these magmas as well as mixing processes lead to the range of differentiated compositions observed at Sangay. Thus, the exceptionally high contents of incompatible elements Žexcept Y and HREE. in the Sangay suite, in comparison to other NVZ suites, appear to be related to the unique location of Sangay at the southern termination of the NVZ, because of both Ž1. an unusually enriched mantle at the Sangay site, prior to Quaternary volcanism, and Ž2. large inputs of slab-derived fluids enriched in soluble incompatible elements, due to the subduction of major fracture zones under the area. In addition, the low CrrNi values found in the whole suite—another unique characteristic for the Andean NVZ—also requires unusual fractionation processes involving Cr-spinel andror clinopyroxene, either at the base of the crust or in the upper mantle. Additional work on other volcanoes in this area, especially the Tulabug cones and the Altar caldera that are located immediately above the tear in the slab ŽFig. 2., is planned in order to better assess the regional variations in mantle characteristics, slab inputs, and deep crustal processes. Because Sangay III is widely surrounded by uninhabited badlands, only rare large eruptions with unusually high Plinian columns represent a major hazard for the inhabited areas located 30 to 100 km west of the volcano. During the 1628 eruption, notable ash falls reached the town of Riobamba, about 50 km NW of the volcano. In addition, as the Sangay III

75

cone attains a relief in excess of 4000 m, a third collapse becomes a greater possibility and must be considered. The corresponding avalanche should be smaller than the two previous ones, but it would seriously threaten the present-day colonization of the Amazon plain located at the foot of the cone. Detailed studies of the two previous collapse events are necessary for a good assessment of this hazard. During low activity, rockfalls and rolling stones are the principal danger for tourists, especially below the western dome. However, ash and scoria falls, pyroclastic flows, lahars, and intense rockfalls may represent very serious hazards in periods of medium to high activity.

Acknowledgements The assistance of the Ecuadorian Army Forces, which provided helicopter transportation to and from the basecamp, is gratefully acknowledged. We wish to thank the local guides from Alao village, led by Roberto Caz Quillay. Thanks also to Michele ` Veschambre for his help during the electron microprobe work, and to Erwan Bourdon for the isotopic analyses of basalt SAN 20B, both carried out at the UMR 6524, University of Clermont-Ferrand ŽFrance.. We are grateful to Jon Davidson and Jim Luhr for their constructive reviews, which significantly improved the manuscript. Financial support for this study was provided by IRD Žpreviously ORSTOM. within the framework of the IRDrIGEPN cooperation.

Appendix A A.1. 4 0Ar r39Ar geochronology Whole rocks were crushed and sieved, and individual grains were selected under binocular microscope. All separates were irradiated in a single batch, at the Siloee ´ reactor of CEA ŽGrenoble, France.. The J factor was estimated at 0.01311 with 1% relative standard deviation using the Fish Canyon sanidine standard with an age of 27.55 " 0.08 ŽLanphere and Baadsgaard, 1997.. Interfering nuclear reactions on K and Ca were calculated by coirradiation of pure

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

76

salts. Samples were put in aluminium packets and placed in a frequency furnace whose temperature was calibrated by means of an optical pyrometer and step heated in a classical fashion from 7008C to 14008C. Each step lasted for 20 min. The gas was purified by means of cold traps with liquid air and Al–Zr getters. Once cleaned, the gas was introduced into a VG3600 mass spectrometer, and 2 min were allowed for equilibration before analyses were done statically. Signals were measured by means of a Faraday cup with a 10 11 -V resistor for 40Ar and 39Ar, while 39Ar, 38Ar, 37Ar and 36Ar were analysed with a

photomultiplier after collision on a Daly knob. The gain between both collectors was estimated by duplicate analysis of 39Ar during each analysis, and also by statistical analysis over a period of several years. This gain is an average of 95 and is known to be better than 1.5%. This error is included in the age calculation, along with analytical errors on each signal and errors on the blank values. We have analyzed age spectra Žweighted mean plateau ages including error on the J factor. and inverse isochron results ŽRoddick et al., 1980. combining errors from each point and linear regression by York’s method

A.2. Analytical data Temp. Ž8C.

40

Arr 39Ar

38

Arr 39Ar

37

Arr 39Ar

36 Arr 39Ar 39Ar F 39Ar % 40Ar ) Ž10y3 . Ž10 – 14 moles. released

SAN 34, Bas, J s 0.0131100 600 0.079 0.023 700 0.071 0.019 800 0.056 0.018 900 0.105 0.019 1000 0.438 0.025 1100 0.337 0.034 1200 0.000 0.030

0.066 0.263 0.865 1.548 1.144 5.942 39.244

0.293 0.294 0.485 0.895 1.902 3.740 11.103

2.56 10.62 21.88 17.47 6.07 1.82 0.28

4.23 21.75 57.83 86.60 96.61 99.58 100.00

SAN 64, Bas, J s 0.0131100 700 0.114 0.029 800 0.085 0.020 900 0.034 0.020 1000 0.088 0.024 1100 0.028 0.027 1200 0.000 0.025 1400 10.874 0.029

0.262 0.483 1.096 0.968 3.891 22.093 26.456

0.478 0.452 0.515 0.662 1.547 5.327 45.039

7.21 20.22 32.08 9.99 4.27 0.66 0.32

9.67 36.75 79.69 93.06 98.75 99.60 100.00

SAN 4, Bas, J s 0.0131100 700 0.082 0.019 800 0.106 0.019 900 0.061 0.021 1000 0.120 0.024 1100 0.277 0.028 1200 1.434 0.028 1300 2.853 0.028 1400 34.608 0.047

0.179 0.348 0.758 0.386 1.615 4.131 5.920 6.446

0.337 0.507 0.492 0.612 1.525 6.342 11.997 118.899

10.32 17.04 31.43 16.52 6.03 1.59 1.24 0.23

12.24 32.44 69.69 89.28 96.41 98.28 99.73 100.00

SAN 39 B, Plagioclase, J s 0.0131100 700 1.163 0.022 800 0.270 0.019 900 0.086 0.018 1000 0.133 0.018 1100 0.000 0.018 1200 0.000 0.018

4.379 6.533 6.577 8.103 7.409 7.181

6.262 4.102 3.617 3.768 2.894 0.498

0.50 2.67 7.07 4.38 2.35 0.77

2.82 17.88 57.79 82.45 95.66 100.00

40

Ar ) r

2.78 0.00 23.36 0.02 28.38 0.02 24.03 0.03 4.47 0.02 y17.45 y0.06 100.00 1.45

39

Ar Age ŽMa. "1s ŽMa.

0.05 0.40 0.38 0.60 0.47 0.00 33.92

0.07 0.03 0.02 0.03 0.11 0.00 13.43

0.01 0.01 0.01 0.01 0.03 1.07 0.93

0.13 0.23 0.27 0.21 0.73 25.03 21.91

0.09 0.05 0.04 0.14 0.69 6.11 24.58

5.87 0.00 y0.25 0.00 10.81 0.01 y10.74 y0.01 7.98 0.02 5.15 0.07 2.12 0.06 2.53 0.89

0.11 0.00 0.16 0.00 0.52 1.76 1.44 20.86

0.03 0.00 0.05 0.00 0.24 1.44 2.54 17.41

0.00 0.00 0.00 0.00 0.02 0.70

0.00 0.00 0.00 0.00 0.48 16.45

0.00 0.00 0.00 0.00 0.93 4.51

5.00 11.42 33.78 9.81 100.00 100.00 8.11

y12.46 y60.23 y231.49 y14.80 100.00 100.00

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

ŽYork, 1969. leading to a mean square weighted deviation ŽMSWD.. Only a summary is presented; detailed analytical results are available from the authors upon request ŽNA.. Sample SAN 34 shows a disturbed age spectrum associated with apparent alteration in the lowest extraction step Žhigher KrCa ratio.. When those first steps are excluded, an age plateau of 400 " 100 ka is indicated for a 90% gas release, which corresponds to an associated isochron age of 380 " 70 ka and no excess argon but a large MSWD Ž10.. Sample SAN 64 has a better, less discordant, age spectrum, but has higher individual errors and gives a plateau age of 230 " 30 ka associated with an isochron age of 310 " 100 ka Žthus identical within uncertainties. with a MSWD of 0.12 Žexpressing the relatively large errors on each point. and no apparent excess argon. Sample SAN 4 is largely dominated by atmospheric argon. The age plateau is very discordant because of the sensitivity to isotope corrections. The isochron approach leads to a cluster of points near the Y axis suggesting an age of 60 " 20 ka, no excess argon, and a MSWD of 0.6. Although the high grouping near the atmospheric intercept can cast doubt on the exact age, this one seems fairly robust. Finally, sample SAN 39 is very young, certainly younger than 100 ka Žminimum age calculated from analytical blank values on the day of analysis., but no precise age could be calculated because all steps yielded 99% atmospheric argon.

References Barberi, F., Coltelli, M., Ferrara, G., Innocenti, F., Navarro, J.M., Santacroce, R., 1988. Plio-Quaternary volcanism in Ecuador. Geol. Mag. 125 Ž1., 1–14. Barragan, R., Geist, D., Hall, M., Larson, P., Kurz, M., 1998. Subduction controls on the compositions of lavas from the Ecuadorian Andes. Earth Planet. Sci. Lett. 154, 153–166. Clapperton, C., 1993. Quaternary Geology and Geomorphology of South America. Elsevier, Amsterdam, 779 pp. Cotten, J., Le Dez, A., Bau, M., Caroff, M., Maury, R.C., Dulski, P., Fourcade, S., Bohn, M., Brousse, R., 1995. Origin of anomalous rare-earth element and yttrium enrichments in subaerially exposed basalts: evidence from French Polynesia. Chem. Geol. 119, 115–138. Davidson, J.P., 1996. Deciphering mantle and crustal signatures in

77

subduction zone magmatism. In: Bebout, G.E., Scholl, D.W., Kirby S.H., Platt, J.P. ŽEds.., Subduction: Top to Bottom. Geophys. Monogr. 96, Am. Geophys. Union, pp. 251–262. Davidson, J.P., de Silva, S.L., 1995. Late Cenozoic magmatism of the Bolivian Altiplano. Contrib. Mineral. Petrol. 119, 387–408. Davidson, J.P., McMillan, N.J., Moorbath, S., Worner, G., Har¨ mon, R.S., Lopez-Escobar, L., 1990. The Nevados de Payachata volcanic region Ž188Sr698W, N. Chile.: II. Evidence for widespread crustal involvement in Andean magmatism. Contrib. Mineral. Petrol. 105, 412–432. de la Condamine, C., 1751. Journal du voyage fait par ordre du Roi a` l’Equateur servant d’introduction historique a` la mesure des trois premiers degres Imprimerie Royale, ´ du meridien. ´ Paris. DeMets, C., Gordon, R.G., Argus, D.F., Stein, S., 1990. Current plate motions. Geophys. J. Int. 101, 425–478. DePaolo, D.J., 1981. Trace element and isotopic effects of combined wallrock assimilation and fractional crystallization. Earth Planet. Sci. Lett. 53, 189–202. Deruelle, B., 1982. Petrology of the Plio-Quaternary volcanism of the south-central and meridional Andes. J. Volcanol. Geotherm. Res. 14, 77–124. de Silva, S.L., Davidson, J.P., Croudace, I.W., Escobar, A., 1993. Volcanological and petrological evolution of volcan Tata Sabaya, SW Bolivia. J. Volcanol. Geotherm. Res. 55, 305–335. Droux, A., Delaloye, M., 1996. Petrography and geochemistry of Plio-Quaternary calc-alkaline volcanoes of southwestern Colombia. J. South Am. Earth Sci. 9 Ž1r2., 27–41. Drummond, M.S., Defant, M.J., Kepezhinskas, P.K., 1996. Petrogenesis of slab-derived trondhjemite–tonalite–daciteradakite magmas. Trans. R. Soc. Edinburgh: Earth Sci. 87, 205–215. Durieux, J., Heiken, G., 1988. Sangay volcano ŽEcuador.. Bull. Scientific Event Alert Network 13 Ž8., 10, Smithsonian Institution, Washington, DC. Eggins, S.M., 1993. Origin and differentiation of picritic arc magmas, Ambae ŽAoba., Vanuatu. Contrib. Mineral. Petrol. 114, 79–100. Feeley, T.C., Hacker, M.D., 1995. Intracrustal derivation of Narich andesitic and dacitic magmas: an example from volcan Ollague, Andean Central Volcanic Zone. J. Geol. 103, 213– 225. Francis, P.W., Wells, G.L., 1988. Landsat Thematic Mapper observations of debris avalanche deposits in the Central Andes. Bull. Volcanol. 50, 258–278. Futa, K., Stern, C.R., 1988. Sr and Nd isotopic and trace element compositions of Quaternary volcanic centers of the southern Andes. Earth Planet. Sci. Lett. 88, 253–262. Gill, J.B., 1981. Orogenic Andesites and Plate Tectonics. Springer, Berlin, 390 pp. Hall, M.L., 1977. El volcanismo en el Ecuador. Publicacion del Instituto Panamericano de Geografia e Historia, seccion nacional del Ecuador, Quito, 120 pp. Hall, M.L., Wood, C.A., 1985. Volcano-tectonic segmentation of the northern Andes. Geology 13, 203–207. Hickey-Vargas, R., Moreno Roa, H., Lopez Escobar, L., Frey, F.A., 1989. Geochemical variations in Andean basaltic and silicic lavas from the Villarrica-Lanin volcanic chain Ž39.58S.:

78

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79

an evaluation of source heterogeneity, fractional crystallization and crustal assimilation. Contrib. Mineral. Petrol. 103, 361– 386. Hildreth, W., Moorbath, S., 1988. Crustal contributions to arc magmatism in the Andes. Contrib. Mineral. Petrol. 98, 455– 489. Johnson, R.W., 1987. Large-scale volcanic cone collapse: the 1888 slope failure of Ritter volcano, and other examples from Papua New Guinea. Bull. Volcanol. 49, 669–679. Kay, S.M., Mpodozis, C., Coira, B., 1996. Central Andean mantle-derived basalts and Neogene mantle enrichment beneath the Puna plateau. Third ISAG, St. Malo ŽFrance., 17– 19r9r1996, Extended abstracts, ORSTOM edn., Paris, pp. ´ 583–586. Kepezhinskas, P., Defant, M.J., Drummond, M.S., 1996. Progressive enrichment of island arc mantle by melt–peridotite interaction inferred from Kamchatka xenoliths. Geochim. Cosmochim. Acta 60 Ž7., 1217–1229. Kilian, R., Pichler, H., 1989. The northandean volcanic zone. Zbl. Geol. Palaont. Teil I 5r6, 1075–1085. ¨ Kilian, R., Hegner, E., Fortier, S., Satir, M., 1995. Magma evolution within the accretionary mafic basement of Quaternary Chimborazo and associated volcanoes ŽWestern Ecuador.. Revista Geologica de Chile 22 Ž2., 203–218. Kokelaar, P., Romagnoli, C., 1995. Sector collapse, sedimentation and clast population evolution at an active island-arc volcano: Stromboli, Italy. Bull. Volcanol. 57 Ž4., 240–262. Lanphere, M.A., Baadsgaard, H., 1997. The Fish Canyon tuff: a standard for geochronology. AGU Spring Metting, EOS, Trans. AGU, p. S326. Lewis, G.E., 1950. El Sangay, fire-breathing giant of the Andes. The National Geographic Magazine, XCVII 1, 117–138. Litherland, M., Aspden, J. A. and Jemielita, R. A., 1994. The metamorphic belt of Ecuador. Overseas Memoir of the British Geological Survey, No. 11. Keyworth, Nottingham, 147 pp.q2 geological maps in color. Lonsdale, P., 1978. Ecuadorian subduction system. Am. Assoc. Petrol. Geol. Bull. 62 Ž12., 2454–2477. Lonsdale, P., Klitgord, K.D., 1978. Structure and tectonic history of the eastern Panama basin. Geol. Soc. Am. Bull. 89, 1–9. Lopez-Escobar, L., Kilian, R., Kempton, P.D., Tagiri, M., 1993. Petrography and geochemistry of Quaternary rocks from the X Southern Volcanic Zone of the Andes between 41830 and X 46800 S, Chile. Revista Geologica de Chile 20 Ž1., 33–55. Martin, H., 1987. Petrogenesis of Archaean trondhjemites, tonalites and granodiorites from eastern Finland: major and trace elements geochemistry. J. Petrol. 28, 921–953. Monzier, M., Robin, C., Hall, M.L., Mothes, P., Samaniego, P., 1996. Sangay volcano ŽEcuador.. Bull. Global Volcanism Network 21 Ž3., 2–4, Smithsonian Institution, Washington, DC. Monzier, M., Robin, C., Hall, M.L., Cotten, J., Mothes, P., ´ Eissen, J.-P., Samaniego, P., 1997a. Les adakites d’Equateur: donnees C.R. Acad. Sci. Paris 324 ŽIIa., 545– ´ preliminaires. ´ 552. Monzier, M., Robin, C., Eissen, J.-P., Cotten, J., 1997b. Geochemistry vs. seismo-tectonics along the volcanic New He-

brides Central Chain ŽSouthwest Pacific.. J. Volcanol. Geotherm. Res. 78, 1–29. Nye, C.J., Reid, M.R., 1986. Geochemistry of primary and least fractionated lavas from Okmok volcano, central Aleutians: implications for arc magmagenesis. J. Geophys. Res. 91 ŽB10., 10271–10287. Pennington, W.D., 1981. Subduction of the Eastern Panama Basin and seismotectonics of northwestern South America. J. Geophys. Res. 86 ŽB11., 10753–10770. Petford, N., Atherton, M., 1996. Na-rich partial melts from newly underplated basaltic crust: the Cordillera Blanca Batholith, Peru. J. Petrol. 37 Ž6., 1491–1521. Prevot, R., Chatelain, J.-L., Guillier, B., Yepes, H., 1996. Tomo´ graphie des Andes equatoriennes: evidence d’une continuite´ ´ ´ des Andes Centrales. C.R. Acad. Sci. Paris 323 ŽIIa., 833–840. Raeburn, S., Atkins, S., 1987. Cambridge high Andes ’86, A geosciences expedition to Ecuador. Final report, Cambridge University, UK, 107 pp. Roddick, J.C., Cliff, R.A., Rex, D.C., 1980. The evolution of excess argon in alpine biotites. Earth Planet. Sci. Lett. 48, 185–208. Roeder, P.L., Emslie, R.F., 1970. Olivine–liquid equilibrium. Contrib. Mineral. Petrol. 29, 275–289. SEAN ŽScientific Event Alert Network. Bulletin, July 1976, August 1983, Smithsonian Institution, Washington, DC. Shepherd, J.B., Aspinall, W.P., Rowley, K.C., Pereira, J.A., Sigurdsson, H., Fiske, R.S., Tomblin, J.F., 1979. The eruption of Soufriere ` volcano, St. Vincent, April–June 1979. Nature 282, 24–28. Siebert, L., 1984. Large volcanic debris avalanches: characteristics of source areas, deposits and associated eruptions. J. Volcanol. Geotherm. Res. 22, 163–197. Simkin, T., Siebert, L., 1994. Volcanoes of the World, 2nd edn. Geosciences Press, Tucson, AZ, USA, 349 pp. Simkin, T., Smith, J.V., 1970. Minor-element distribution in olivine. J. Geol. 78, 304–325. Singer, B.S., Leeman, W.P., Thirlwall, M.F., Rogers, N.W., 1996. Does fracture zone subduction increase sediment flux and mantle melting in subduction zones? Trace element evidence from Aleutian arc basalt. In: Bebout, G.E., Scholl, D.W., Kirby, S.H., Platt, J.P. ŽEds.., Subduction: Top to Bottom. Geophysical Monograph 96, American Geophysical Union, pp. 285–291. Stein, C.A., Stein, S., 1992. A model for the global variation in oceanic depth and heat flow with lithospheric age. Nature 359, 123–129. Stern, C.R., Kilian, R., 1996. Role of the subducted slab, mantle wedge and continental crust in the generation of adakites from the Andean Austral Volcanic Zone. Contrib. Mineral. Petrol. 123, 263–281. Sun, S.-S., McDonough, W.I., 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. In: Saunders, A.D., Norry, M.J. ŽEds.., Magmatism in the Ocean Basins, No. 42. Geological Society Special Publication, pp. 313–315. Tatsumi, Y., Hamilton, D.L., Nesbitt, R.W., 1986. Chemical characteristics of fluid phase released from a subducted litho-

M. Monzier et al.r Journal of Volcanology and Geothermal Research 90 (1999) 49–79 sphere and the origin of arc magmas: evidence from high-pressure experiments and natural rocks. J. Volcanol. Geotherm. Res. 29, 293–309. Thompson, R.N., Morrison, M.A., Hendry, G.L., Parry, S.J., 1984. An assessment of the relative roles of a crust and mantle in magma genesis: an elemental approach. Philos. Trans. R. Soc. London. A 310, 549–590. Thorpe, R.S., Francis, P.W., Hammil, M., Baker, M.C.W., 1982. The Andes. In: Thorpe, R.S. ŽEd.., Andesites: Orogenic Andesites and Related Rocks. Wiley, Chichester, pp. 187–205. Tindle, A.G., Webb, P.C., 1994. Probe-amph—a spreadsheet program to classify microprobe-derived amphibole analyses. Computers and Geosciences 20 Ž7r8., 1201–1228. USGSrNEIC ŽU.S. Geological Survey, National Earthquake Information Center., 1968–1992. Preliminary determination of epicenters, monthly listing, May 1968 to June 1992.

79

Whymper, E., 1892. Travels amongst the Great Andes of the Equator. New edition with introduction and photographs by Loren McIntyre, Gibbs M. Smith. Peregrine Smith Books, Salt Lake City, 1987. Yepes, H., Chatelain, J.-L., Guillier, B., Alvarado, A., Egred, J., Ruiz, M., Segovia, M., 1996. The M w 6.8 Macas earthquake in the subandean zone of Ecuador, October 3, 1995. Seismol. Res. Lett. 67 Ž6., 27–32. York, D., 1969. Least square fitting to a straight line with correlated errors. Earth Planet. Sci. Lett. 5, 320–324. Zindler, A., Jagoutz, E., Goldstein, S., 1982. Nd, Sr and Pb isotopic systematics in a three component mantle: a new perspective. Nature 298, 519–523.