Commented words, mainly from referred journals

2MASS miss dim galaxies and the outer halo of galaxies?

Yes, it does, I said in  2002A&A...382..495A and the galaxy luminosity function (and luminosity density) is biased low. No, it doesn't Norber et al. replied in 2002MNRAS.336..907N after my criticisms to their paper (2001MNRAS.326..255C). Yes, it does, as independently confirmed Blanton M. et al. (2003, which one of the many papers he wrote?) and Bell et al. (2003, ApJ in press, astro-ph/0302543). Kirby et al. (2008, ApJ, in press, arXiv:0808.2529 ) confirm it in the most straighforward way: by deep observations.

Cooling flow: cooling or not?

As early as 1995 (1995A&A...300..711A) I used the argument that no hot gaz can cool and forms stars in clusters, otherwise a color gradient would appear in the color profile of the brightest cluster galaxy (BCG) whereas none is seen. "It is easy to convince oneself that any stellar population different in age, metallicity  or mass function from the BCG one, induces a variation in the shape of the color profile at least if its spatial distribuition is not a fine tuned function of the BCG one". Five year later, no transitional levels characteristic of low temperature gaz were observed in high resolution spectra of cooling flow clusters, showing that some energetic source should compensate the energy radiated by the cooling flow and keep the gaz not cooling.

E/S0 ratio evolution, a very short story:

Galaxy evolution is one of the central topics of observational cosmology. When galaxies formed? And galaxies with different morphologies have different ages and assembly times? Dressler et al. (1997ApJ...490..577D) found that the ratio of the number of lenticulars and ellipticals decreases from fE/S0=2   at z=0 to about fE/S0~0.2 at z~0.5. In simpler terms, the morphology of early-type galaxies (ellipticals and lenticulars) evolves, or, equivalently, these galaxies have different average ages. Lenticulars should appear at z<0.5, yet there are conflicting observations: lenticulars are old and have homogeneous ages, share the same fundamental plane as ellipticals have larger bulge on disk ratio and mass that their supposed ancestors (spirals) (e.g. Dressler 1980, Simien & de Vaucoulers 1985). How is this possible? On a tiny sample at z=0 and z=0.4, I (1997A&A...323..337A) instead claim the ratio to be constant. In a later work I (1998ApJ...501..533A) claimed their result to be biased and their statistically significance overestimated, triggering a vigorous debate. According to my analysis, galaxy morphology does not evolve, solving the problem of conflicting observations. Fasano et al. (2000ApJ...542..673F) rebutted my claim, by studying a larger sample by the same biased analysis that underestimated errors and confirmed the original Dressler et al. claim, hence re-introducing the problem of conflicting observations. Fabricant et al. (2000ApJ...539..577F), co-authors of a paper (1998ApJ...500..714V) sharing my criticisms of Dressler et al. (1997), recognized the difficutly of the game and preferred left the "ring". To settle this issue Lubin et al. (1998AJ....116..584L) asked to the greatest living authority concerning morphological types, A. Sandage, to perform the morphological classification of the galaxies of two clusters at z~0.85. They find an E/S0 ratio consistent with local value. Later works by the same team on other clusters confirmed my original result: no evidence for evolution (yet). Burstein et al. (2005ApJ...621..246B) show that lenticulars are too bright/massive to be the faded descendent of spirals. Finally, Dressler (2003) in his magnificent review, suggests to recognize E from S0 using kinematics. If the proportion of E and S0 changes with redshift, then the ellipticiy distribution of E+S0 should change either, being S0 more elliptical than E. Holden et al. (2009ApJ...693..617H) show this is not the case.
The Butcher-Oemler effect:

The Butcher-Oemler effect is defined as the increase of the fraction of blue galaxies in the core of galaxy clusters, for a smart definition of blue fraction, one that account for the youger age of the universe at high redshift. In the recent literature authors use the term quenched fraction to indicate 1 minus the blue fraction. There are a number of issues to be addressed to successfully measure the blue fraction (or the quenched fraction) (Andreon 2006): a) the cluster apple vs orange issue (Andreon & Ettori 1999): the fraction of blue galaxies depends on the cluster mass and one should not select very massive clusters at high redshift and lower mass clusters at low redshift, as instead cluster samples naturally end up; b) clusters should be selected independenty on the quantity which is under measure, their blueness. It is not a great idea to select intermediate-high redshift clusters from images sampling the cluster blue rest-frame: clusters with a large blue fraction will be over-represented, biasing the blue fraction itself. c)  the blue fraction depends on the clusters radius. If you don't have the latter under control (say you took 1 Mpc for all clusters) you are putting yourself in a bad position (may  you exclude that more compact clusters are not selected at high redshift?). d) there is no doubt that the luminosity of the galaxies evolves. Dont count galaxies down to a fixed absolute luminosity, but to a fixed evolving luminosity, because the fraction  of blue galaxies does depend on luminosity, and a constant absolute magnitude would induce a bias. e) the  luminosity change of blue and red galaxies is different, don't evolve the two in the same way, say by taking M*+1 for both, with M* being the one of the whole population. f) finally, if your derived fraction or its errorbar is outside the 0 to 1 range, perhaps your data are asking a better statistical analysis, because fraction are bounded in the 0 to 1 range and you cannot observed them outside this interval!

Conclusion: if you don't account for the above, a bias is there, a change is usually observed, but this do not imply evolution (in addition to the one related to the younger age of the universe). If you account for all the above, there is no BO effect (i.e. no more evolution than expected by simple stellar aging and star formation change), up to z~0.4-0.7 (Andreon et al. 2006; Andreon et al. 2004 ). Similarly, there is  no BO effect if you take a mass-selected sample of galaxies (De Propris et al. 2003, but point c is violated, hoewever). Loh et al. (2008) follow Andreon 2006  but for violating point b and c, and found no BO effect in the redshift range where their sampling of the Balmer break (used to define a galaxy as blue) is not extreme (0.6<z<0.8). No BO effect is found using 24 micron (in the case you worry optical because of dust) as indicator of star formation (Haines et al.  Jenam 2009 conference), but for a sample extending only to z~0.3. If you select the cluster sample independently on the quantity being investigated, no BO effect is there ( Smail et al. 1998), again for a sample extending only to z~0.3. There is perhaps some signal if you go up to z=1 (Andreon et al. 2008), but who know? It is just one hi-z cluster and point b is violated. As time goes, more and more works accounts for the above and found less and less BO effect. Raichoor and Andreon (2012) studied an X-ray selected sample formed by 25 clusters with 0 < z < 1 and one cluster at z ~ 1.8 and found no BO effect for massive galaxies in the cluster core. Evolution is instead found for less massive galaxies and less dense environments (not studied by Butcher & Oemler). What happen at high redshift is still under discussion: all massive galaxies in the most distant cluster and with deepest data, JKCS041 at z=1.803, are red (Newman et al. 2014, Andreon et al. 2014), while z=1.5 is the star formation epoch according to some authors having only incomplete (for red galaxies, for example) data of lower quality, or perhaps some, but not all, galaxies are quenched (Nantais et al. 2016, 1606.07832).

The deficit of faint red galaxies:

We known since a long time that the z=0 field environment is populated by dwarf irregular (faint star-forming, i.e. non-red, galaxies) while z=0 clusters have a large population of dE, i.e. faint quiescent (red) galaxies, and that massive galaxies stopped forming stars earlier than less massive galaxies (down-sizing). We also known that some faint red galaxies in Coma have a spiral morphology (Andreon 2008), i.e. were blue in the recent past. As a consequence the faint part of the red sequence was less populated in the past and ratio between the number of bright and faint red galaxies in cluster should evolve. But how much? Not much, because at z=0 the color-magnitude relation is linear (e.g. Andreon 2006) and with a constant scatter (the CM should bend or its scatter should increase if some galaxies become quiescent ad later times). A lot said Stott et al. (2007). De Lucia et al. (2007) and other works reiterate the point. However, the effect was not seen in one z=0.8 cluster with excellent data (Andreon 2006), and either in the best (largest number of clusters, larger redshift baseline, more rest-frame homogeneous photometry, etc) sample by Andreon (2008), who also explains why other people find a deficit instead:  the aimed measurement is intrinsically difficult, both astronomically and statistically. Astronomically, the claimed deficit of dwarfs disappear when the quality of the data increases (e.g. Lidman et al. 2008)  and it is only apparent in the faintest magnitude bin, where optimistic astronomers believe their data to be complete, but are not, as reiterated by De Propris et al. (2013). Crawford et al. (2009) simulations confirmed my  claims that other people statistical analysis is too simplified. Of course, persisting in make the same mistake do not remove the bias, and some authors for a while confirmed the strong evolution originally claimed. Hilton et al. (2009) claim that his hi-z cluster have too many dwarfs (or too few giants), the opposite of what many other authors claim (a deficit of dwarfs at high-z). In 2016 most (all?) authors concord that the ratio has not changed: Cerulo et al. 2016 (9 clusters, 0.84 < z < 1.45), Zenteno et al. 2016 (26 clusters, 0.1 < z < 1.2, SPT selected), De Propris et al. 2015 (< z >~1.25), De Propris et al. 2013 (11 clusters, 0.2 < z < 0.6). In the most distant cluster known, at z=1.803, the ratio is equal to the z=0 value (Andreon et al. 2014).
To summarize, we now know that quiescent massive and less massive galaxies assembled at almost the same time and that the population of less massive quiescent galaxies grew by a small amount, in agreement with fossil (i.e. z=0) evidence.

How many low surface galaxies there are?

Many, say several authors, starting from Disney (1976Natur.263..573D): if you seen only few of them, is only because the sky is bright! Not so many, replay Cross et al. (2001MNRAS.324..825C). O'Neil, Andreon & Cuillandre (2003A&A...399L..35O) show that the authors integrated the bivariate luminosity-brightness function on an insufficiently extended part of the luminosity (down only to MB=-16). Cross et al. (2003, astro-ph/0312317) then find a problem in the 2dF data, and considered the results of their previous work compromised.

Coma LF

Mobasher et al. (2003ApJ...587..605M) say: "Andreon & Cuillandre (2002ApJ...569..144A) carried out a wide-area (650 arcmin2) photometric survey of the Coma Cluster. Combining this survey with Hubble Space Telescope observations, they confirmed that unresolved globular clusters could be mistakenly classified as dwarf galaxies at faint magnitudes, causing the observed steepness of the LF faint-end slope. Correcting for this, they derived α = -1.4, with both the shape and slope of the LF depending on color. They also derived LFs in surface brightness intervals and found low surface brightness galaxies as the main contributor at fainter magnitudes ... . This result is in close agreement with that found in § 4.5, using a [shallower] spectroscopic sample". The agreement implies that no (telescope expensive) spectroscopy is needed!

My favorite referee report

S. Odewan, referee of the paper (2000MNRAS.319..700A) wrote in his referee report: "In all, I think this work represents a "raising of the bar" in the field of automated astronomical image classification. It greatly exceeds the level of past such works and should be published with only minor cosmetic changes"