The timing for the onset of plate tectonics, along with the secular changes in the tectonic settings of continental crust formation, continue to be debated. Recent interpretations based on the increase in zircon 176 Hf/ 177 Hf ratios at the time of crystallisation (expressed as εHf (t) with respect to chondritic evolution) have been used to ascertain changes in geodynamic settings in the early Earth, specifically in the 3.8-3.6 Ga interval. This increase is widely interpreted as a change in magma generation, from source(s) dominated by ancient crust to source(s) dominated by juvenile inputs from the mantle. At issue, Hf isotope variations remain limited in the early Earth due to the long decay of the Lu-Hf system. This feature, along with the scarcity of rocks and minerals of Eo/Mesoarchaean and Hadean ages, generate large uncertainties over the nature and the timing of the interactions between mantle and crustal reservoirs. The distinction between mantle and crustal sources becomes much clearer in the Palaeoproterozoic, and the study of ancient terranes with several billion years of protracted crustal evolution may hold the key to unlock complex mantle-crust interactions. Here we investigate high-grade metamorphic rocks from the Anabar shield, which contain zircons that crystallised between the Eoarchaean (oldest core at 3814 ± 16 Ma) and the Palaeoproterozoic (youngest core at 2251 ± 15 Ma and youngest rim at 1910 ± 21 Ma). The combination of in situ U-Pb and Hf isotope analyses in zircon indicates the formation of the continental crust in the Siberian Craton in the Eoarchaean, and a conspicuous metamorphic event at 2.0-1.9 Ga. We demonstrate that 2.0-1.9 Ga zircon ages reflect recrystallisation processes under subsolidus conditions, involving the breakdown of high-Lu/Hf minerals (i.e. garnet and pyroxene). The εHf (t) shift at 2.0-1.9 Ga towards more radiogenic values may not be related to a change in magmatic style and sources, but rather to resetting of the Lu-Hf system in response to heating and metamorphic reactions on a mineral scale. Our findings challenge the widely-evoked mechanism of changes in tectonic style and magma sources to account for vertical arrays in the εHf (t) versus crystallisation age space. This calls for considering alternative options, including those based on petrographic data, when interpreting Hf isotope variations in the Hadean/Archaean detrital zircon record.