We present the first volatile contents (H2O, CO2, Cl, F, S) of young (< 6 Ma) submarine basaltic glasses from the Phoenix and West Scotia mid-ocean ridges and the Bransfield Strait back-arc of the South Shetland subduction zone in the Antarctic Peninsula.The volatile contents of the MORB glasses correspond well with those of published Pacific MORB and reflect covariations in source enrichment and extent of melting. Our results support the hypothesis that decreasing spreading rates at the Phoenix Ridge resulted in preferential melting of less abundant enriched MORB mantle, due to its greater fertility and higher volatile contents, relative to the more abundant depleted MORB mantle.The volatile contents of the Bransfield Strait back-arc glasses correlate with geochemical indicators of subduction processes and reveal an along-axis spatial distribution consistent with a toroidal inflow of sub-slab asthenosphere around the edges of the subducting plate into the mantle wedge. This inflow should be considered when assessing spatial and geochemical variability at subduction zones, particularly those with slab windows and tears.A small group of Bransfield Strait samples have volatile contents that do not correlate with geochemical signals of subduction influence. We speculate that these samples reflect flux melting of residual enriched mantle brought beneath the Bransfield Strait via corner flow following recent alkaline magmatism in the far eastern regions of the Antarctic Peninsula.Our new data on lavas from the W7 segment of the West Scotia Ridge reveal their source was significantly affected by subduction processes. Unexpectedly, these lavas have CO2-H2O pressures of vapor saturation that suggest they were collected in-situ and erupted relatively recently (∼6 Ma), at odds with previous interpretations of their origins. We suggest they originated from a subduction-modified mantle (lithosphere or asthenosphere) left behind by the eastward-migrating South Sandwich subduction zone sometime over the past ∼30 Myr. These lavas demonstrate the long-lasting effects of subduction processes on the upper mantle and their potential to influence melt compositions even in non-subduction environments today.We use the compositions of lavas from the Phoenix Ridge and Bransfield Strait to estimate mantle potential temperatures; our results agree well with global estimates for mid-ocean ridges and subduction zones, respectively.