Current models for the origin of MORB and OIB invoke different degrees of partial melting of a homogeneous lherzolitic source, and a heterogeneous deep mantle source, respectively. In the ocean basins, MORBs are only part of a spectrum of geochemically diverse depleted to enriched basalts that erupt at or near ridges, off-axis seamounts and large igneous provinces. Even at ridges, mantle is locally enriched (e.g. E-MORB). The gradation in compositions from MORB to slightly less depleted tholeiites at LIPS, to variably enriched tholeiitic and alkalic basalts, basanites and olivine nephelinites of many ocean islands requires only differences in depth and degree of partial melting of shallow mantle lherzolite upon which trace-element and isotopic heterogeneity are superimposed. Alkalic basalts and differentiates in the oceans occur along nearly every seamount ridge rising >2000 m above the seafloor, a distribution too extensive to be explained by any number of plumes; this makes a plume origin for similar lavas on linear island chains questionable. Tapping along fractures of a shallow asthenospheric layer of variably enriched and fertile mantle that develops beneath the lithosphere through time is more likely. The long-term differentiation of the Earth, magmatism, recycling, continental rifting, and subduction insure that the upper mantle cannot be well mixed and homogeneous, a common but fallacious assumption in much petrogenetic theory. Mantle major-element and isotopic heterogeneity and variable temperature is a consequence of plate tectonics. Every association of ultramafic rocks in the ocean crust, ophiolites, and xenolith suites demonstrates significant bulk heterogeneity that survives partial melting. Thus sources of modern abyssal tholeiites must be variably fertile with respect to a basaltic melt fraction, and range from average harzburgite to fertile lherzolite, on both local and regional scales. In addition, subduction guarantees that most abyssal basalt and gabbro is returned to the mantle, there transforming to eclogite of diverse lithology and mineralogy. Eclogite distribution in the upper mantle cannot be uniform owing to the localized occurrence of trenches. Eclogite-rich mantle can be an important source of magma where the crust is thick (LIPs, Iceland) or volcanic structures very large (Hawaii). There is no need for plume-induced heterogeneity of the shallow mantle or for high temperatures. The absence of heat-flow and thermal anomalies at hotspots implies the presence of athermal mechanisms to explain melting and geochemical anomalies. Fertility spots, wetspots and lithospheric stress heterogeneity are a natural result of plate tectonics and may explain 'hotspots' without assuming deep-mantle thermal plumes. The non-fixity of hotspots, the occurrence of erratic age progressions and 'hotlines' and the orientations of volcanic chains have a natural explanation if lithospheric architecture, subduction and asthenospheric heterogeneity control magma volumes and chemistry. The case for a depleted eclogitic MORB-like source for Icelandic tholeiites will be discussed as an example.