Temporal changes in mantle wedge geometry and magma generation processes in the Central Andes: towards linking petrological data to thermomechanical models

Temporal and spatial patterns of Neogene ignimbrite magmatism in the Central Andes were analyzed using GIS and geostatistical modeling. We compiled a comprehensive ignimbrite database using available literature, satellite imagery and geochemical data. 203 individual ignimbrite sheets were digitized in GIS, for which geochemical, isotopic (partly), and geochronological data are available from literature sources and own data (http://www.arcgis.com/home/item.html?id=47038ddc0628473f9f0ce67aa2eff8be). Based on this analysis, we estimate composition, volumes and sources of erupted ignimbrite magmas through space and time for five segments of the Central Andes.
The total erupted ignimbrite magma volume is estimated to be about 31,000 km3 for the past 30 Ma, with the highest average eruption rates (rm) for the northern Puna segment (3.4 km3 Ma-1 km (arc)-1), followed by 0.7 km3 Ma-1 km (arc)-1 for the Altiplano. For Southern Peru, rm is smaller (0.5 km3Ma-1km (arc)-1), which might be due to the lack of knowledge about intra-caldera volumes. Furthermore there is a clear N-S “younging” of ignimbrite pulses. Major pulses of high magma eruption rate occurred at 19-24 Ma (e.g. Oxaya, Nazca Group), 13-14 Ma (e.g. Huaylillas ignimbrites), 6-10 Ma (Altiplano and Puna ignimbrites, e.g. Vilama ignimbrite) and 3-6 Ma (e.g. Atana, Los Frailes, Toconao). In contrast, small volume young ignimbrites from 0-3 Ma (e.g. Lauca-Perez, Purico) do not show the spatio-temporal pattern of eruptions that are documented for the large- volume ignimbrite flare-ups.
Compositional and Sr-O isotopic data indicate that ignimbrite magmas are more crustally derived in younger flare-ups in the Southern Central Andes (up to 50 % crustal melts) compared to older ignimbrites in the north (only up to 20 % crustal melts). This suggests that thermal conditions, juvenile magma production in the mantle, thickness, and/or composition of the crust must have been different along the Central Andes at the time when ignimbrite flare up magmas formed.
The amount of juvenile magmas that entered the crust and the degree and volume of partial melting within the crust can thus in principle be constrained in time and space by these data. Such data are essential in order to understand the thermal evolution of the Andean crust in space and time.
The Miocene large-volume, plateau forming ignimbrites always overly a pronounced unconformity and occur after a time with no magmatism. They are followed, however, by andesitic arc magmatism characterized during the Late Miocene by low angle, large-volume (~2.2 km3 per lava flow) volcanic shields with long single lava flows up to 20 km. These shields are succeeded by younger and more evolved steeply-sided strato-cones that characterize much of the CVZ active volcanic front for Pliocene-Quaternary times. Andesites in such young stratovolcanoes (~0.7 km3 per lava flow) are often characterized by amphibole phenocrysts.
In principle, the transition between these andesite regimes could be due to:
(1) a change in the mantle melting regime from decompression (hot and dry?) to flux melting (wet and lower T?),
(2) different rates in magma production and effusion, and
(3) different P-T-regimes of magma evolution within the crust as is shown by the depletion in HREE and Y from Miocene to Pleistocene volcanic rock caused by a residual garnet after crustal assimilation in a thickened crust.

To understand the shift from andesite shields to stratovolcanoes we studied Miocene to modern Central Andean volcanic rocks that represent different ages but are similar in petrography and composition in order to test differences in processes of magma generation. Based on a survey of ~1300 chemical analyses of lava samples (http://andes.gzg.geo.uni-goettingen.de/) we selected three representative sample types: (1) most mafic samples (50-55 % SiO2), (2) intermediate andesites representing 63 % of the data (55-60 % SiO2), and (3) felsic samples (60-65 % SiO2), all of which were identified before as important endmember magma type in the Central Andes. Using a range of geothermometers, hygrometers and MELTS modelling we show that the P-T parameters at the time of eruption, for a given composition, remained surprisingly constant trough time and throughout the Central Andes (e.g. 975 °C to 985 °C for 2-px thermometry). Moreover, the depth of the last (phenocryst) crystallization of Miocene to Present magmas took place between 9 and 3.5 km throughout Andean history. These observations clearly indicate that estimated temperatures only reflect the late crystallization history at shallow levels and that any distinct regimes of magma formation in the mantle wedge that may have existed are entirely dampened out during the passage through the crust. Density, viscosity and degassing of andesite magmas control the latest stages of ascent and crystallization and these parameters are independent of crustal conditions, subduction geometry and mantle wedge conditions. Therefore, the thickened upper crust not only serves as a chemical filter for mantle wedge magmas but also controls (and synchronizes) P-T conditions of crystallization as recorded in erupted products.
Deep evolution at the level of the magma sources and lower crust, where assimilation and magmatic differentiation takes place, is thus completely decoupled from the shallow processes of late crystallization. Therefore only the rate of effusion, and by implication, magma production and upper crustal stress regime remain as primary factors that may have influenced differences between Miocene and Recent magmatic products.
Since the sequence of distinct magmatic regimes (plateau-ignimbrites, shield andesites and evolved stratovolcanoes) is diachronous during the past 26 Ma of Andean evolution with ages getting younger from N to S. This suggests control by “deeper” processes guided by the geometry of the slab and the thermal evolution of the upper plate during Andean orogeny. As patterns, timing of events, subduction parameters and magma production rates in the mantle wedge change regionally and temporally during ongoing thickening of the Central Andean crust, the upper plate reacts at any given location individually to these changes according to its present thermal state, crustal composition, magmatic history and tectonic stress conditions at that time and space.
We propose that large-volume ignimbrite eruptions occurred in the wake of subduction of the Juan-Fernandez ridge that passed below the Central Andes from N to S during the past 25 Ma. This event resulted in compression, uplift, low angle subduction (flat slab) and fluid release in a first stage, followed by massive inflow and melting of asthenospheric mantle after the passing of the ridge when the slab again steepened rapidly. This in turn caused massive melting within the crust aided by advective heat transport shortly after slab steepening. Differences in chemical and isotopic composition of the large-volume ignimbrites are related to changes in crustal thickness, and its “preconditioning” during the Anden orogeny over time.
The change in effusion rate during the Miocene to Pliocene/Quarternary may be the only parameter that relates to changing angles and/or convergence rates of the slab. Since only convergence rates changed during the last 26 Ma (Sérbier and Solar, 1991), this parameter likely controls magmatic activity (Cagnoicle et al., 2007). In southern Peru, Miocene voluminous magmatic activity correlates with high convergence rates, both decreasing in the last 10 Ma (Sébrier and Soler, 1991).
Previous model predictions for arc magmatism (Sobolev et al., 2006) follows from the comparison between the evolution of tectonic shortening and the evolution of the mantle temperature beneath the magmatic arc. Processing the delamination material throught the asthenospheric wedge by corner flow results in an increasing shortening rate. Simultaneously, the temperature of the asthenospheric wedge beneath the magmatic arc decreases, which in turn should lead to reduced magmatic activity. However, apart from the Puna (Kay and Kay, 1993) and Northern Altiplano (Back and Zandt., 2002; Yuan et al., 2002)) no evidence exists for mantle lithosphere delamination in the Central Andes. Therefore a new model based on our volumes, eruption rates, petrological constraints and the movement of the Juan Fernández ridge should give a better understanding of the controlling factors of arc magmatism.

Back SL, Zandt G (2002) The nature of orogenic crust in the Central Andes. J Geophys Res 107:n doi 10.1029/2000JB000124
Cagnioncle AM, Parmentier EM, Elkins-Tanton LT (2007) Effect of solid flow above a subducting slab on water distribution and melting at convergent plate boundaries. J Geophys Res 112: B09402
de Silva, S., Gosnold, W.D. (2007) Episodic construction of batholiths: Insights from the spatiotemporal development of an ignimbrite flare-up. J. Volcanol. Geotherm. Res. 167: 320–335.
Kay RW, Kay SM (1993) delamination and delamination magmatism. Tectonophysics 219:177 - 189
Sébrier, M., Solar, P. 1991. Tectonics and magmatism in the Peruvian Andes from late Oligocene time to the Present. Geological Society of America 265, 259–278.
Sobolev SV, Babeyko AY, Koulakov I, Oncken O (2006) Mechanism of the Andean orogeny: insight from numerical modeling. In: Oncken O, Chong G, Franz G, Giese P, Götze H-J, Ramos VA, Strecker MR, Wigger P (eds) The Andes – active subduction orogeny. Frontiers in Earth Science Series, Vol 1. Springer-Verlag, Berlin Heidelberg New York, pp 513–536, this volume
Yáñez, G., Cembrano, J., Pardo, M., Ranero, C., Selles, D., 2002. The Challenger–Juan Fernández– Maipo major tectonic transition of the Nazca–Andean subduction system at 33–34 S: geodynamic evidence and implications. Journal of South American Earth Sciences 15, 23‐38.