Linking volcano morphology and composition to magma reservoir evolution at Parinacota Volcano, Chile

Mapping, stratigraphy, and 57 new 40Ar/ 39Ar ages describe the 163 k.y. history and volumetric evolution of Parinacota Volcano in the Andean Central Volcanic Zone (CVZ), constraining variability in eruptive flux over the lifetime of a stratovolcano built upon extremely thick (70 km) crust. Most of Parinacota's 46 km3 volume erupted in the last 52 k.y. in two cone-building episodes (old cone---52 to 20 ka, and young cone---15 ka to present), separated by major sector collapse. Peak eruptive flux of 1.0+/-0.3 km3/k.y. is similar to arc volcanoes on thinner crust. Major and trace element, 87Sr/86Sr, and 238U-230Th compositions tied to geochronology show that the old cone was initially composed of spatially distinct eruptive centers, fed by separate reservoirs, that coalesced prior to sector collapse. With increasing recharge and magma output lavas transition from low-Fe, strongly calc-alkaline, phenocryst-rich, silicic compositions to medium-Fe, near-tholeiitic, aphanitic and mafic characteristics. High La/Yb and 230Th excess in young cone lavas represent variable degrees of garnet-residual lower-crustal melting, and up to 30% eruption-age-corrected 230Th excesses indicate rapid <18 k.y. crustal transit times. In contrast, near-equiline old cone magmas have 230Th/ 232Th similar to uppermost crust and contain crystals up to 120 k.y. older than eruption, indicating protracted shallow assimilation and crystallization. These data support the hypothesis that the expression of low-Fe 'calc-alkaline' (CA) vs. medium-Fe 'tholeiitic' (TH) trends can be modulated by transitions from compartmentalized, stagnant, assimilation-prone 'dirty' systems (CA) to 'clean' systems (TH) where rapid magma throughput results in minimal opportunity for upper-crustal contamination. At Parinacota and elsewhere in the CVZ, biotite 40Ar/39Ar apparent ages are consistently up to 600 ka older than coeval sanidine, implying that biotite is retentive of extraneous 40Ar but sanidine is not. Because closure temperature (Tc) varies with pre-eruption cooling rate, use of plutonic cooling rates underestimates Tc for most volcanic systems. Recalculated Tc for biotite and hornblende are sub-equal to eruption temperatures for silicic lavas, thereby allowing preservation of non-atmospheric Ar partitioned into biotite prior to volcanic degassing and eruption. Young samples exacerbate the problem, calling into question application of biotite absolute ages to Neogene stratigraphic correlation.