Kormendy & Bender (2012) provide an ARA&A-style review of physical processes that heat galaxy disks and that use up or remove gas and so quench star formation. They suggest that the following processes all are important in transforming Im+S disks into Sph+S0 disks:
Internal process: Dekel & Silk (1986; see also Larson 1974; Saito 1979) "suggest that both the dIs and the [dSphs] have lost most of their mass in winds after the first burst of star formation, and that this process determined their final structural relations. The dIs somehow managed to retain a small fraction of their original gas, while the [dSphs] either have lost all of their gas at the first burst of star formation or passed through a dI stage before they lost the rest of the gas and turned [dSph]." Kormendy and Bender conclude that "the Sph+Im sequence of decreasing surface brightness with decreasing galaxy luminosity is a sequence of decreasing baryon retention" (emphasis in both originals). 2pt Figure 2 shows that surface brightnesses in the Sph+S0 sequence start to decrease at MV,disk ≃ -18, just where bulges disappear and where we therefore change galaxy classifications from S0 to Sph (i.e., where plot symbols change from dark to light green). Figure 4 independently finds the S0 → Sph transition from rotation curve decompositions. As dark matter Vcirc decreases, bulges decrease in importance relative to disks until they disappear at Vcirc ≃ 104 ± 16 km s-1. The Tully-Fisher (1976) relation shows that this corresponds to MV ≃ -19 for spirals (Courteau et al. 2007) and MV ≃ -18 for S0s (Bedregal et al. 2006). It is probably not an accident that baryons start to be lost roughly where bulges stop to contribute to the gravitational potential.
Figure 4. Maximum rotation velocity of bulge (Vcirc,bulge:red points) and disk Vcirc,disk (black points) components given in bulge-disk-halo decompositions of observed rotation curves V(r) whose outer, dark matter rotation velocities are Vcirc (Kormendy & Freeman 2014; references to the V(r) decomposition papers are given there). The dotted line indicates that rotation velocities of the visible and dark matter are equal. Every red point has a corresponding black point, but many late-type galaxies are bulgeless, and then the plot shows only a black point. The lines are symmetric least-squares fits; the disk fit is Vcirc,disk = (1.16 ± 0.03)(Vcirc - 200) + (183 ± 3) km s-1; Vcirc,bulge = (1.73 ± 0.29)(Vcirc - 200) + (166 ± 9) km s-1 is the bulge fit. The bulge correlation is steeper than the disk correlation. Bulges disappear at Vcirc ≃ 104 ± 16 km s-1.
(Continued) The transition in Figures 2 and 3 to (I suggest) a baryon retention sequence in smaller dwarfs corresponds within errors to the transition in edge-on galaxies from giants that do to dwarfs that do not have well-defined dust lanes in their disk midplanes (Dalcanton et al. 2004). They argue that the transition is not caused by changes in gas density. Rather, they argue that it is caused by a transition in giant galaxies to a regime in which disk instability leads to lower gas turbulence, smaller gas scale heights, and enhanced star formation. It is plausible that star formation is less efficient in smaller dwarfs. However: The observation that dwarf Im and Sph galaxies show the same decrease in surface brightness with decreasing luminosity shows that the dominant effect is not one that depends on the presence of gas. This favors the suggestion that the μe(MV) relation in Figures 2 and 3 is a baryon retention sequence at MV > -18. See also Figure 6.
It may be surprising that Sph+S0 and Im+S galaxies have similar surface brightnesses, because the latter are star-forming and so presumably have smaller mass-to-light ratios. However, (1) at high luminosities, internal absorption partly compensates for smaller mass-to-light ratios, and (2) at low luminosities, star formation in Im galaxies is gentle. Nevertheless, it is fair to emphasize that more work is needed to understand the detailed engineering that results in Sph+S0 and Im+S correlations that look so nearly identical.
Environmental processes I: Ram-pressure stripping-long underestimated in importance- is now An Idea Whose Time Has Come. Gunn & Gott (1972) suggested, based on the detection of X-ray-emitting, hot gas in the Coma cluster (Meekins et al. 1971; Gursky et al. 1971) that "a typical galaxy moving in it will be stripped of its interstellar material. We expect no normal spirals in the central regions of clusters like Coma. The lack of such systems is, of course, observed" (emphasis in the original). Calculations persuaded some people to neglect ram-pressure stripping even while many observations provided indirect evidence for its importance. Spirals near the center of the Virgo cluster are deficient in HI gas (Cayette et al. 1990, 1994; Chung et al. 2009). Also, Faber & Lin (1983); Lin & Faber (1983); Kormendy (1987), van den Bergh (1994b), and Kormendy et al. (2009, hereafter KFCB) suggested that Sph galaxies are ram-pressure-stripped dS+Im galaxies, based in part on observations (Einasto et al. 1974; van den Bergh 1994a, 1994b; Mateo 1998) that - with a few (understandable) exceptions - close dwarf companions of Local Group giant galaxies are almost all spheroidals, that distant companions are irregulars, and that galaxies with intermediate (Sph/Im) morphologies live at intermediate distances. Ever since van den Bergh (1976), these ideas provided the interpretation of a parallel sequence classification (Figure 1 here) that was constructed operationally to encode the full range of S0+Sph bulge-to-total luminosity ratios B / T from almost 1 to exactly 0.
Spectacular observations of ram-pressure stripping in action now clearly demonstrate the importance of this process (see Sun et al. 2010; Kormendy & Bender 2012 for reviews). Kenney et al. (2004, 2008), Oosterloo & van Gorkom 2005, and Chung et al. (2007, 2009) find that many spiral galaxies near the center of the Virgo cluster show long Hα or HI tails interpreted to be cold gas that is being stripped by the ambient X-ray-emitting gas (Böhringer et al. 1994). Kormendy and Bender emphasize that these galaxies and the HI-depleted galaxies are substantially brighter than almost all Sphs: "If even the deep gravitational potential wells of still-spiral galaxies suffer HI stripping, then the shallow potential wells of dS+Im galaxies are more likely to be stripped." The most impressive recent example of ram-pressure stripping is the multi-wavelength (COgas+HII+X-ray) tail of the galaxy ESO 137-001 in Abell 3627 (Sun et al. 2006, 2007, 2010; Woudt et al. 2008; Sivanandam et al. 2010; Pavel et al. 2014; Figure 5 here).
For a recent treatment of ram-pressure stripping in the Galactic halo, see Gatto et al. (2013).
Figure 5. Composite Hubble Space Telescope (HST) and Chandra X-Ray Observatory image of galaxy ESO 137-001 in Abell 3627. The HST image is a I-, g-, and U-band color composite. Added in blue is the X-ray image; it extends the lighter blue optical streaks of ongoing star formation toward the lower-right. This material is interpreted to be ram-pressure stripped. The image source is http://www.spacetelescope.org/images/heic1404b/ and http://apod.nasa.gov/apod/ap140328.html. At a distance of ~ 64 Mpc, the absolute magnitude of ESO 137-001 is MV ~ -20.8. Conveniently, this field of view also shows a normal-looking Sph galaxy at upper-right; it is ~ 3 mag fainter than ESO 137-001. Again, if we see a giant galaxy caught in the process of undergoing ram-pressure stripping, it is not surprising that a much smaller Sph galaxy is thoroughly "red and dead."
Environmental processes II: Dynamical harassment results from many, high-speed encounters with other galaxies in a cluster and with the overall cluster potential. Simulations show that (1) it promotes gas flow toward galaxy centers, (2) it heats disks, especially vertically, and (3) it strips off the outer parts of galaxies (Moore et al. 1996, 1998, 1999; Lake et al. 1998). Even in poor environments like the Local Group, tidal stirring of dwarfs on elliptical orbits around the Galaxy or M31 should have similar effects (Mayer et al. 2001a, b, 2006). One success of this picture is that inflowing gas can feed star formation and help to explain why spheroidals, in which star formation stopped long ago, do not have lower surface brightnesses than current versions of S+Im progenitors (Figure 3). This process is clean and inescapable.
Kormendy & Bender (2012) conclude that dynamical harassment is much more important in the Virgo cluster than we thought. Edge-on S0s with close companions show warps in their outer disks that will phase-wrap around the center into structures that resemble Sphs in their shapes and density profiles (NGC 4762 and NGC 4452; their Figures 4 and 8, respectively). They identify S0/Sph transition objects: (1) NGC 4638 is an edge-on S0 with a Gaussian (i.e., radially truncated) inner disk embedded in a boxy, E3 halo that has the properties of a large Sph (their Figures 13-16). (2) VCC 2048 is an E6 Sph with an embedded edge-on, S0 disk (their Figures 10-12). Indeed, many Virgo S0s have Gaussian disk profiles. Most telling is the observation that several Virgo cluster S0s have bars embedded in steep, often Gaussian-profile lenses with no sign of a disk outside the bar (NGC 4340, NGC 4442, and NGC 4483, their Section A.12). We do not know how to form a bar that fills the whole disk, because an outer disk is generally required as an angular momentum sink to allow a bar to grow. Kormendy and Bender suggest that the outer disks in these galaxies have been heated or stripped off. Finally, they interpret the "new class of dwarfs that are of huge size and very low surface brightness" (Sandage & Binggeli 1984) as "spheroidals that have been harassed almost to death."
Environmental processes III: Starvation of late growth by cold-gas infall (Larson et al. 1980) seems inevitable in environments like the center of the Virgo cluster where the ambient gas is very hot. Even in environments like the Local Group, Sph galaxies that orbit around the Galaxy and M31 at velocities V ~ 200 km s-1 are unlikely to encounter cold gas slowly enough to be able to accrete it. And much of the gas in the Local Group may in any case be in a warm-hot intergalactic medium (WHIM: Davé et al. 2001).
So Kormendy & Bender (2012) "suggest that the relevant question is not `Which of these mechanisms is correct?' It is `How can you stop any of them from happening?' It seems likely that all of the above processes matter." Engineering details probably depend on environment.