Before HST, a few ground-based studies provided hints that the galaxies responsible for the Butcher-Oemler effect are prevalently spirals with disturbed morphologies, and/or mergers [73, 74]. Already the first HST images proved that indeed distant clusters contain a large number of spirals, many of which are disturbed [75, 76, 77, 78, 79, 55]. Moreover, in the MORPHS clusters, the high spiral fraction corresponds to a low S0 fraction suggesting that a significant fraction of the spirals evolve into at least some of the S0s that dominate rich clusters today ([78], also [94]). This result has been questioned by other studies [80, 81], but at least part of the discrepancy might be ascribed to the fact that these works have focused on the number ratio of S0 to elliptical galaxies. This latter quantity, more prone to errors than the simple fractions of each morphological type, has been found to be highly dependent on the cluster characteristics, in particular on the central concentration of ellipticals in the clusters [83]. When considering separately the E, S0 and spiral fractions as a function of redshifts, studies at z = 0.1-0.25 confirm the trend of evolution of S0s [83]. Also if the overall early-type fraction (E+S0) is examined, a strong evolution with redshift is found [16, 82] with high redshift clusters being composed of a much larger fraction of late type galaxies.
Signatures for quite recent star formation in S0 galaxies consistent with a transformation from spirals have been detected in several clusters, while such activity is found to be absent in ellipticals ([84, 85, 86, 87, 88], see also Katgert in these proceedings). However, not all studies find a difference among the stellar population ages of S0s and ellipticals (e.g. [19, 89, 38]). A possible explanation for the discrepant results are the magnitude limits of the various studies, since most of the S0s with recent star formation avoid the top end of the luminosity function, as expected evolving the luminosities of typical star-forming spirals at intermediate redshifts [86].
Interestingly, cosmological high-resolution simulations are able to account for the morphology-density relation of ellipticals, but they fail to reproduce the properties of the S0s, as if some additional mechanism for S0 formation, not currently included in the models, might be required [90, 91].
Finally, [44] argue for an unexpectedly large fraction of mergers in a cluster at z = 0.8. Since most of these pairs have red colors, we might be witnessing the formation of ellipticals through the merging of galaxies whose stars formed at high z, as expected in hierarchical models of structure formation.
Due to space limitations, I am forced to omit in these proceedings the discussion of the morphology-density relation [78, 83, 84, 92], and of the "star formation-density" relation [93, 94].