As soon as the first submm galaxies were detected in 1997, it was clear that they made a significant contribution to the luminosity density in the high-redshift Universe, subject to the plausible assumptions that their SEDs were similar to those of luminous low-redshift dusty galaxies, and that their redshifts were not typically less than about z = 0.5. These assumptions remain plausible and have been confirmed to an acceptable level by subsequent observations (Smail et al., 2000, 2002). Despite an initial suggestion that 30% of the SCUBA galaxies could be at z < 1 (Eales et al., 1999), it now seems that a median redshift of submm-selected galaxies is of order 2-3 (Eales et al., 2000; Smail et al., 2002).
We have already discussed that evolution by a factor of about 20 in the value of L* is required to account for the properties of the submm source counts and background radiation intensity. That is the key result from submm surveys, but how can it be explained in models of galaxy evolution? In this section we will not describe the modeling process in great detail, but we highlight the key features of such analyses, and the most important future tests of our current understanding.
It is important is to be aware that there is still considerable
uncertainty in the exact form of evolution required to explain the submm
observations. While strong luminosity evolution of the dusty galaxy
population is required out to
z 
1 and
beyond, the detailed form of that evolution is rather loosely constrained
by count
and background data. Redshift distributions are an essential requirement
in order to determine the form of evolution accurately.