The evolution of quasar clustering is investigated with a new sample of 388 quasars with 0.3 < z \leq 2.2 , B \leq 20.5 and M _ { B } < -23 , selected over an area of 24.6 deg ^ { 2 } in the South Galactic Pole .
Assuming a two-point correlation function of the form \xi ( r ) = ( { r / { r _ { o } } } ) ^ { -1.8 } , we detect clustering with r _ { 0 } = 6.2 \pm 1.6 h ^ { -1 } comoving Mpc and \bar { \xi } ( r = 15 ~ { } h ^ { -1 } Mpc ) = { 3 \over r ^ { 3 } } \int _ { 0 } ^ { r } x ^ { 2 } \xi ( x ) dx = 0.52 \pm 20 at an average redshift of < z = 1.3 > .
We find a 2 \sigma significant increase of the quasar clustering between z = 0.95 and z = 1.8 , independent of the quasar absolute magnitude and inconsistent with recent evidence on the evolution of galaxy clustering .
If other quasar samples are added ( resulting in a total data-set of 737 quasars ) the increase of the quasar clustering is still favoured although it becomes less significant .
With a parameterization of the evolution of the type \xi ( r,z ) = ( { r / { r _ { 0 } } } ) ^ { - \gamma } ( 1 + z ) ^ { - ( 3 - \gamma + \epsilon ) } we find \epsilon \simeq - 2.5 .
Evolutionary parameters \epsilon > 0.0 are excluded at a 0.3 % probability level , to be compared with \epsilon \sim 0.8 found for galaxies .
The observed clustering properties appear qualitatively consistent with a scenario of \Omega = 1 CDM in which a ) the difference between the quasar and the galaxy clustering can be explained as a difference in the effective bias and redshift distributions , and b ) the quasars , with a lifetime of t \sim 10 ^ { 8 } yr , sparsely sample halos of mass greater than M _ { min } \sim 10 ^ { 12 } -10 ^ { 13 } ~ { } h ^ { -1 } M _ { \sun } . We discuss also the possibility that the observed change in the quasar clustering is due to an increase in the fraction of early-type galaxies as quasar hosts at high z .