The color-magnitude relation of early-type galaxies differs slightly but significantly from a pure power-law , curving downwards at low and upwards at large luminosities ( M _ { r } > -20.5 and M _ { r } < -22.5 , respectively ) .
This remains true of the color-size relation , and is even more apparent with stellar mass ( M _ { * } < 3 \times 10 ^ { 10 } M _ { \odot } and M _ { * } > 2 \times 10 ^ { 11 } M _ { \odot } , respectively ) .
The upwards curvature at the massive end does not appear to be due to stellar population effects .
In contrast , the color- \sigma relation is well-described by a single power law .
Since major dry mergers change neither the colors nor \sigma , but they do change masses and sizes , the clear features observed in the scaling relations with M _ { * } , but not with \sigma > 150 { km~ { } s } ^ { -1 } , suggest that M _ { * } > 2 \times 10 ^ { 11 } M _ { \odot } is the scale above which major dry mergers dominate the assembly history .

We discuss three models of the merger histories since z \sim 1 which are compatible with our measurements .
In all three models , dry mergers are responsible for the flattening of the color- M _ { * } relation at M _ { * } > 3 \times 10 ^ { 10 } M _ { \odot } – wet mergers only matter at smaller masses .
At M _ { * } > 2 \times 10 ^ { 11 } M _ { \odot } , the merger histories in one model are dominated by major rather than minor dry mergers .
In another , although both major and minor mergers occur at the high mass end , the minor mergers contribute primarily to the formation of the ICL , rather than to the stellar mass growth of the central massive galaxy .
This model attributes the fact that \alpha < 1 in the scaling M _ { * } \propto M _ { dyn } ^ { \alpha } , to the formation of the ICL .
A final model assumes that the bluest objects today were assembled by minor dry mergers of the bluest ( early-type ) objects at high redshift , whereas the reddest objects were assembled by a mix of major and minor dry mergers .
In this model , the scatter of the color-magnitude relation should increase with redshift , and the dependence on environment should also be more pronounced at higher redshift : more clustered objects should be redder .
Similar measurements of these relations at high redshift will provide further valuable constraints on the mass scale at which major dry mergers dominate the assembly history .