Understanding and realistically simulating the coupling between land and atmosphere in global climate models (GCMs) is an ongoing research frontier. We explore the hypothesis that past attempts to investigate these physics using GCM mechanism denial experiments may have suffered systematic limitations stemming from an overly strong sensitivity of deep convection parameterizations to surface conditions. Taking the philosophy of the Global Land-Atmosphere Coupling Experiment (GLACE), we therefore compare the effects of breaking the soil-atmosphere feedback mechanism in the Super-Parameterized Community Atmosphere Model version 3.5 (SPCAM3.5) – which uses O(10k) embedded cloud resolving models to explicitly resolve moist convection – against the conventionally parameterized CAM3.5. This helps isolate the influence of explicit convection on land-atmosphere coupling. We find that soil moisture – precipitation coupling strength is reduced over northern Africa, northern South America and Arabian Peninsula due to superparameterization. Several geographically distinct coupling "hotspots" emerge in SPCAM3.5 located upstream of major topographic features in the Northern Hemisphere mid-latitudes.