Résumé
The record of the continental lithosphere is patchy and incomplete; no known rock is older than 4.02 Ga, and less than 5% of the rocks preserved are older than 3 Ga. In addition, there is no recognizable mantle lithosphere from before 3 Ga. We infer that there was lithosphere before 3 Ga and that ∼3 Ga marks the stabilization of blocks of continental lithosphere that have since survived. This was linked to plate tectonics emerging as the dominant tectonic regime in response to thermal cooling, the development of a more rigid lithosphere, and the recycling of water, which may in turn have facilitated plate tectonics. A number of models, using different approaches, suggest that at 3 Ga the volume of continental crust was ∼70% of its present-day volume and that this may be a minimum value. The continental crust before 3 Ga was on average more mafic than that generated subsequently, and this pre-3 Ga mafic new crust had fractionated Lu/Hf and Sm/Nd ratios as inferred for the sources of tonalite-trondhjemite-granodiorite and later granites. The more intermediate composition of new crust generated since 3 Ga is indicated by its higher Rb/Sr ratios. This change in composition was associated with an increase in crustal thickness, which resulted in more emergent crust available for weathering and erosion. This in turn led to an increase in the Sr isotope ratios of seawater and in the drawdown of CO2. Since 3 Ga, the preserved record of the continental crust is marked by global cycles of peaks and troughs of U-Pb crystallization ages, with the peaks of ages appearing to match periods of supercontinent assembly. There is increasing evidence that the peaks of ages represent enhanced preservation of magmatic rocks in periods leading up to and including continental collision in the assembly of supercontinents. These are times of increased crustal growth because more of the crust that is generated is retained within the crust. The rates of generation of continental crust and mantle lithosphere may have remained relatively constant at least since 3 Ga, yet the rates of destruction of continental crust have changed with time. Only relatively small volumes of rock are preserved from before 3 Ga, and so it remains difficult to establish which of these are representative of global processes and the extent to which the rock record before 3 Ga is distorted by particular biases.