The 2dF Galaxy Redshift Survey: constraints on cosmic star-formation history from the cosmic spectrum

We present the first results on the history of star formation in the Universe based on the `cosmic spectrum', in particular, the volume-averaged, luminosity-weighted, stellar absorption line spectrum of present day galaxies from the 2dFGRS. This method is novel in that unlike previous studies it is not an estimator based on total luminosity density. The cosmic spectrum is fitted with models of population synthesis, tracing the history of star formation prior to the epoch of the observed galaxies, using a method we have developed which decouples continuum and spectral-line variations and is robust against spectrophotometric uncertainties. The cosmic spectrum can only be fitted with models incorporating chemical evolution and indicates there was a peak of star-formation rate in the past of at least three times the current value and that the increase back to z=1, assuming it scales as (1+z)^beta, has a strong upper limit of beta<5. We find in the general case there is some model degeneracy between star formation at low and high redshift. However, if we incorporate previous work on star formation at z<1 we can put strong upper limits on the SFR at z>1: e.g., if beta>2 then the SFR for 1<z<5 scales as (1+z)^alpha with alpha<2. This is equivalent to stating that no more than 80% of stars in the Universe formed at z>1. Our results are consistent with the best-fit results from compilations of cosmic SFR estimates based on UV luminosity density, which give 1.8<beta<2.9 and -1.0<alpha<0.7, and are also consistent with estimates of Omega_stars based on the K-band luminosity density.