Since the pioneering observations of Spite & Spite in 1982 , the constant lithium abundance of metal-poor ( [ Fe/H ] < -1.3 ) halo stars near the turnoff has been attributed to a cosmological origin .
Closer analysis , however , revealed that the observed abundance lies at \Delta ^ { 7 } Li \sim 0.4 dex below the predictions of Big Bang Nucleosynthesis ( BBN ) .
The measurements of deuterium abundances along the lines of sight toward quasars , and the recent data from the Wilkinson Microwave Anisotropy Probe ( WMAP ) , have independently confirmed this gap .
We suggest here that part of the discrepancy ( from 0.2 to 0.3 dex ) is explained by a first generation of stars that efficiently depleted lithium .
Assuming that the models for lithium evolution in halo turnoff stars , as well as the \Delta ^ { 7 } Li estimates are correct , we infer that between one-third and one-half of the baryonic matter of the early halo ( i.e . \sim 10 ^ { 9 } M _ { \odot } ) was processed through Population III stars .
This new paradigm proposes a very economical solution to the lingering difficulty of understanding the properties of the Spite Plateau and its lack of star-to-star scatter down to [ Fe/H ] = -2.5 .
It is moreover in agreement both with the absence of lithium in the most iron-poor turnoff star presently known ( HEÂ 1327-2326 ) , and also with new trends of the Plateau suggesting its low metallicity edge may be reached around [ Fe/H ] = -2.5 .
We discuss the role of turbulent mixing associated with enhanced supernovae explosions in the early interstellar medium in this picture .
We suggest how it may explain the small scatter , and also other recent observational features of the lithium Plateau .
Finally , we show that other chemical properties of the extremely metal-poor stars ( such as carbon enrichment ) are also in agreement with significant Population III processing in the halo , provided these models include mass-loss and rotationally-induced mixing .