Mining the gap: evolution of the magnitude gap in X-ray galaxy groups from the 3-square-degree XMM coverage of CFHTLS

International audience; We present a catalog of 129 X-ray galaxy groups, covering a redshift range 0.04 \textless z \textless 1.23, selected in the similar to 3 deg(2) part of the CFHTLS W1 field overlapping XMM observations performed under the XMM-LSS project. We carry out a statistical study of the redshift evolution out to redshift one of the magnitude gap between the first and the second brightest cluster galaxies of a well defined mass-selected group sample. We find that the slope of the relation between the fraction of groups and the magnitude gap steepens with redshift, indicating a larger fraction of fossil groups at lower redshifts. We find that 22.2 +/- 6% of our groups at z \textless= 0.6 are fossil groups. We compare our results with the predictions of three semi-analytic models based on the Millennium simulation. The intercept of the relation between the magnitude of the brightest galaxy and the value of magnitude gap becomes brighter with increasing redshift. This trend is steeper than the model predictions which we attribute to the younger stellar age of the observed brightest cluster galaxies. This trend argues in favor of stronger evolution of the feedback from active galactic nuclei at z \textless 1 compared to the models. The slope of the relation between the magnitude of the brightest cluster galaxy and the value of the gap does not evolve with redshift and is well reproduced by the models, indicating that the tidal galaxy stripping, put forward as an explanation of the occurrence of the magnitude gap, is both a dominant mechanism and sufficiently well modeled.