A collection of Quaternary, high-MgO (≤13.4 wt%) basanite and minette cinder and lava cones, with an enhanced arc geochemical signature, are located along the northern margin of the N-S Colima rift in western Mexico. The Colima rift overlies the lithospheric suture between the Jalisco block and Guerrero terrane, as well as the tear between the Rivera and Cocos subducting oceanic plates. From the literature, volatile analyses of olivine-hosted melt inclusions in the Colima cone samples document notably high concentrations of dissolved H2O in the melt (≤ 7.0 wt%) as well as degassing-induced phenocryst growth over a range of depths ≤ 25 km. In this study, it is shown that the high-MgO character of the Colima suite reflects liquid compositions, consistent with evidence for their rapid transit to the surface, without stalling in a crustal magma chamber. The most Mg-rich olivine analyzed in each sample matches the equilibrium composition at the liquidus based on olivine-melt Mn-Mg and Fe2+-Mg exchange coefficients. Application of both a Mg- and Ni-based olivine-melt thermometer, calibrated on the same experimental data set, to the Colima cone suite provides the temperature and dissolved H2O content at the liquidus. Because the Ni-thermometer is insensitive to dissolved H2O in the melt, it gives the actual temperature at the onset of olivine phenocryst growth. For the nine Colima samples that range from 13.4-9.2 wt% MgO, resulting temperatures range from 1221-1056 (± 6-11) °C. In contrast, the Mg-thermometer is sensitive to dissolved H2O in the melt, and its application (without a correction of H2O) gives the temperature of olivine crystallization under anhydrous conditions. When the Mg- and Ni-based temperatures are paired, the depression of the liquidus (∆T = TMg-TNi) due to dissolved H2O can be obtained. For the high-MgO (>9 wt%) Colima samples, ∆T values range from 188-109 °C. Corrections for the effect of pressure (i.e., evidence that phenocryst growth began at ~700 MPa), increase ∆T by ~21°C. An updated model calibration (on experiments from the literature) that relates ∆T with dissolved H2O in the melt shows that the best fit (R2= 0.95) is linear, wt% H2O = 0.047*∆T, with a standard error of ± 0.5 wt%. Although the experimental data set spans a wide range of melt composition (e.g., 47-58 wt% SiO2, 4.4-10.2 wt% MgO, 1.3-4.9 wt% Na2O, 0.1-5.0 wt% K2O, 0.3-5.3 wt% H2O), no dependence on anhydrous melt composition is resolved. Application of this updated model to the Colima suite gives H2O contents of 5.1-8.8 wt% H2O, consistent with those analyzed in olivine-hosted MIs from the literature. When the thermometry and hygrometry results for the Colima cone suite are compared to those for the adjacent calc-alkaline basalts from the Tancítaro Volcanic Field (TVF) in Michoacán, two distinct linear trends in a plot of wt% H2O vs. temperature are found, indicative of different mantle sources. It is proposed that the high-MgO (>11 wt%) Colima cone melts were derived from a phlogopite-bearing harzburgitic mantle at the base of the Jalisco block lithosphere, whereas both TVF and Colima melts with ≤10 wt% MgO were derived from the asthenosphere (i.e., arc mantle wedge). In both mantle sources, slab-derived fluids were an important source of H2O.
