2003AJ....126..706W

We have constructed a sample of quasar candidates by comparing the FIRST radio survey with the 16 deg² Deeprange I-band survey carried out by Postman and coworkers. Spectroscopic follow-up of this magnitude-limited sample [I<20.5, F_ν(20cm)>1 mJy] has revealed 35 quasars, all but two of which are reported here for the first time. This sample contains some unusual broad absorption line (BAL) quasars, including the first radio-loud FR II BAL previously reported by Gregg and coworkers. Comparison of this sample with the FIRST Bright Quasar survey samples selected in a somewhat bluer band and with brighter magnitude limits reveals that the I-band-selected sample is redder by 0.25-0.5 mag in B-R and that the color difference is not explained by the higher mean redshift of this sample but must be intrinsic. Our small sample contains five quasars with unusually red colors, including three that appear very heavily reddened. Our data are fitted well with normal blue quasar spectra attenuated by more than 2.5 mag of extinction in the I band. These red quasars are only seen at low redshifts (z<1.3). Even with a magnitude limit I<20.5, our survey is deep enough to detect only the most luminous of these red quasars at z≲1; similar objects at higher redshifts would fall below our I-band limit. Indeed, the five most luminous objects (using dereddened magnitudes) with z<1.3 are all red. Our data strongly support the hypothesis that radio quasars are dominated by a previously undetected population of red, heavily obscured objects. Unless highly reddened quasars are preferentially also highly luminous, there must be an even larger, as yet undiscovered, population of red quasars at lower luminosity. We are likely to be finding only the most luminous tip of the red quasar iceberg. A comparison of the positions of the objects in our sample with the catalog of Deeprange cluster candidates reveals that five of our six z<1 quasars are associated with cluster candidates of similar estimated redshifts. This association is very unlikely to be the result of chance. It has some surprising implications, including the possibility that up to half of the Deeprange clusters at z∼1 have associated quasars.