Nature, Vol.573, No.7775, 578-+, 2019
Kimberlites reveal 2.5-billion-year evolution of a deep, isolated mantle reservoir
The widely accepted paradigm of Earth's geochemical evolution states that the successive extraction of melts from the mantle over the past 4.5 billion years formed the continental crust, and produced at least one complementary melt-depleted reservoir that is now recognized as the upper-mantle source of mid-ocean-ridge basalts(1). However, geochemical modelling and the occurrence of high He-3/He-4 (that is, primordial) signatures in some volcanic rocks suggest that volumes of relatively undifferentiated mantle may reside in deeper, isolated regions(2). Some basalts from large igneous provinces may provide temporally restricted glimpses of the most primitive parts of the mantle(3,4), but key questions regarding the longevity of such sources on planetary timescales- and whether any survive today-remain unresolved. Kimberlites, small-volume volcanic rocks that are the source of most diamonds, offer rare insights into aspects of the composition of the Earth's deep mantle. The radiogenic isotope ratios of kimberlites of different ages enable us to map the evolution of this domain through time. Here we show that globally distributed kimberlites originate from a single homogeneous reservoir with an isotopic composition that is indicative of a uniform and pristine mantle source, which evolved in isolation over at least 2.5 billion years of Earth history-to our knowledge, the only such reservoir that has been identified to date. Around 200 million years ago, extensive volumes of the same source were perturbed, probably as a result of contamination by exogenic material. The distribution of affected kimberlites suggests that this event may be related to subduction along the margin of the Pangaea supercontinent. These results reveal a long-lived and globally extensive mantle reservoir that underwent subsequent disruption, possibly heralding a marked change to large-scale mantle-mixing regimes. These processes may explain why uncontaminated primordial mantle is so difficult to identify in recent mantle-derived melts.