Annu. Rev. Astron. Astrophys. 2004. 42: 603-683 Copyright © 2004 by Annual Reviews. All rights reserved

8. CAVEATS

8.1. Stellar Populations in Classical Bulges and Pseudobulges

Published studies of stellar populations and metallicities promote quite a different view of bulges than the one discussed in this paper. They emphasize that the stellar populations are generally old, metal-rich, and similar to those of elliptical galaxies. On the basis of such evidence, many papers suggest that observations of stellar populations are inconsistent with secular evolution and instead point to early and rapid formation. Do these results reveal a problem with the present picture?

The themes of this section are as follows: Most stellar population studies concentrate on early-type galaxies; their results are consistent with our conclusion that these generally contain classical bulges. In contrast, many Sbc and later-type bulges, plus a few early-type objects with independent evidence for pseudobulges, do contain young stars. However, some galaxies with clearcut evidence for pseudobulges certainly have old stellar populations. At present, we see no compelling collision between stellar population and secular evolution studies. But the situation is not clearcut, and a collision is possible. More than any other subject discussed in this review, the stellar populations of late-type (pseudo)bulges need further work to make sure that there is no fundamental problem or else to uncover it. We concentrate on a few seminal papers that illustrate these points.

Peletier et al. (1999) studied the B - I versus I - H color-color diagram of bulges. They have shown convincingly that dust absorption dominates the central colors in many galaxies, and that colors at the effective radius r_e of the bulge are much less affected by dust. Their conclusions are as follows:

For 13 bulges of S0-Sab galaxies, colors at r_e mostly show little scatter and are consistent with the colors of Coma cluster ellipticals. The authors conclude that the age spread of these bulges is small, at most 2 Gyr, and that the stars formed 12 Gyr ago. Two early-type galaxies have blue colors indicative of younger ages. NGC 5854 is classified Sa, but Sandage & Bedke (1994) note: "The inner spiral pattern [of two] is in the bulge (it forms the bulge) and has the form and the sense of the opening of a stubby S." This spiral is more consistent with a pseudobulge than with a classical bulge. Spiral structure requires a dynamically cold system, and indeed, the weighted mean of the best velocity dispersion measurements is 102 ± 4 km s^-1 (Simien & Prugniel 1997; Vega Beltrán et al. 2001; Falcón-Barroso, Peletier & Balcells 2002). The second "young" bulge is in the S0 galaxy NGC 7457; Kormendy (1993) showed that this has an unusually small velocity dispersion indicative of a pseudobulge. Thus, of 13 early-type bulges discussed in Peletier et al. (1999), the only two that have young stellar populations also are pseudobulges.

Of four Sb bulges, one (NGC 5443) is young; it is also barred. The other three are made of old stars. It is important to note that these include the boxy pseudobulge NGC 5746 (Figure 15).

Peletier et al. (1999) noted that their three Sbc bulges are "considerably bluer, have lower surface brightness, show patchy dust and star formation together, and are rather different from the rest of the galaxies." So the sample is small, but the results in Peletier et al. (1999) are consistent with the present picture that secular evolution dominates late-type bulges but that early-type bulges, on the whole, are like ellipticals. One important new result is that at least one pseudobulge is made of old stars. Generally similar results were found by Bica & Alloin (1987).

Another result that is consistent with and even suggestive of secular evolution is a correlation between bulge color and the color of the adjacent part of the disk (Peletier & Balcells 1996, Gadotti & dos Anjos 2001). Bulges and disks both show large ranges in colors, but "bulges are more like their disk than they are like each other" (Wyse et al. 1997). Also, some bulge colors found in the above studies are indicative of young ages, especially for Sc-Sm galaxies (de Jong 1996c). Bulge and disk scale lengths and surface brightnesses also correlate (see the above papers; Courteau, de Jong, & Broeils 1996, Courteau 1996b). Courteau and collaborators have interpreted these correlations as products of "secular dynamical evolution...via angular momentum transfer and viscous [gas] transport."

Finally, Trager (2004) reviewed recent work which shows that bulges of S0/a-Sbc spirals span a large range of ages and [-element/Fe] abundance enhancements. The latter are a particularly important indicator of star-formation history because -elements are ejected by massive stars when they explode as supernovae of type II; their abundances are diluted with Fe when type I supernovae become important ~ 1 Gyr after a starburst. After that, [-element/Fe] can never again be much greater than the solar value. So overabundances of the -elements indicate that almost all of the star formation occurred quickly (e.g., Terndrup 1993; Bender & Paquet 1995; Thomas, Greggio & Bender 1999; Thomas, Maraston & Bender 2002; Worthey, Faber & Gonzalez 1992). Trager reviewed evidence that bulges with high luminosities or velocity dispersions show -element enhancements, but those with low luminosities or velocity dispersions do not. For example, many S0 bulges, which tend to be high in luminosity, tend to have -element enhancements indicative of rapid formation (e.g., Bender & Paquet 1995; Fisher, Franx, & Illingworth 1996). These results do not ring alarm bells, but they would be a more decisive test of secular evolution if they included more late-type galaxies.

It is well known that a few S0 bulges have post-starburst spectra, but these are more likely to be the result of galaxy accretions than processes that form either classical or pseudo bulges.

In summary, stellar population data appear reasonably consistent with the conclusion of previous sections that S0-Sb galaxies tend to have classical bulges, and that Sbc-Sm galaxies usually have pseudobulges. However, (a) the galaxy samples studied are too heavily weighed toward early-type galaxies to be a decisive test, and (b) there certainly exist galaxies that are difficult to understand. For example, while the boxy bulge of the S0 galaxy NGC 7332 is likely to be younger than its disk (Bender & Paquet 1995), the even more boxy bulge of NGC 5746 appears to be old (Peletier et al. 1999).

Finally, the best-studied boxy bulge is the one in our Galaxy. The low-absorption field that has received the most attention is Baade's window. At a Galactic latitude of -4°, it is well up into the boxy part of the bulge revealed by COBE (see figure 1 in Wyse, Gilmore, & Franx 1997). Still, it is almost along the minor axis, so there is a small chance that the stars that define the boxiness are not completely the same as the ones in Baade's window. In any case, the observations imply that bulge stars far from the Galactic plane are old (Terndrup 1993, Ortolani et al. 1995, Feltzing & Gilmore 2000, Kuijken & Rich 2002, Zoccali et al. 2003, all of which also review earlier work; for further review, see Sandage 1986; Wyse, Gilmore, & Franx 1997; Renzini 1999; Rich 1999). The absolute age is uncertain but is approximately 11-13 Gyr. Moreover, moderate -element overabundances with respect to iron (McWilliam & Rich 1994, Barbuy et al. 1999) again imply rapid star formation over a period of 1 Gyr. Finally, the observed correlation that more metal-poor bulge/halo stars have more eccentric, plunging orbits continues to point to an early collapse with accompanying self-enrichment (Eggen, Lynden-Bell & Sandage 1962; Sandage 1986, 1990). The Galactic bulge is clearly older than a secular evolution picture can easily accommodate. If we try to solve this problem by postulating that the boxy structure was made by heating a pre-existing disk of old stars, then the fact that the bulge and the metal-poor halo are similar in age becomes a coincidence. Also, the Galactic center is currently forming stars at a rate that, if sustained for an appreciable fraction of a Hubble time, adds up to much of the stellar density observed there (Rich 1999). We argue in this paper for the importance of secular evolution, but we would be the last to suggest that the above results are easily understood. Clearly, solving them deserves high priority. O n the other hand, the Galactic bulge is clearly boxy. At present, the only model that we have for its origin is via secular processes.

8.2. Can Minor Accretion Events Mimic Pseudobulge Growth?

This is certainly possible. Kannappan, Jansen & Barton (2004) found a correlation between blue-centered, star-forming bulges and evidence of tidal encounters with neighboring galaxies. Well-known examples are M82 and NGC 3077, which are connected to M81 by H I tidal bridges (Yun, Ho & Lo 1994). Counterrotating gas and even stellar components in some galaxies also imply accretion. An example is NGC 4826 (Braun et al. 1994, Rubin 1994, Walterbos et al. 1994, Rix et al. 1995, Garcia-Burillo et al. 2003). However, NGC 4826's pseudobulge signature - a very low stellar velocity dispersion (Kormendy 1993) - is a property of corotating stars and therefore predates the accretion of counterrotating material and has not yet been affected by it. An example of a galaxy with pseudobulge characteristics that may instead be caused by a gas accretion is NGC 7457 (Kormendy & Illingworth 1983, Kormendy 1993, Peletier et al. 1999).

There are three reasons why we suggest that secular evolution accounts for more pseudobulges than do accretion events: (a) Many of the most recognizable pseudobulges occur in strongly barred and oval galaxies, especially in ones in which radial dust lanes imply that gas infall is ongoing now. (b) If galaxies approach each other closely enough to transfer gas, then their dark matter halos are likely already to overlap and they are likely to merge after a few more orbits. A configuration like the M81 - M82 - NGC 3077 encounter can last for a billion years but not for a significant fraction of a Hubble time and not at all without being recognizable. Most pseudobulge galaxies show no signs of tidal interactions in progress. (c) Inhaling a tiny, gas-rich dwarf does no damage to an existing disk, but a major merger heats a thin disk too much to be consistent with flat, edge-on galaxies.

Nevertheless, the relative importance of internal and externally driven secular evolution is not known and needs further study. It is likely that accretions create more than an occasional quasipseudobulge.

8.3. Disky Distortions in Elliptical Galaxies

Some ellipticals contain central disky distortions (see Bender et al. 1989, Kormendy & Djorgovski 1989, Bender 1990b, Kormendy & Bender 1996 for reviews). In the more extreme cases, the line-of-sight velocity distributions show asymmetries or even a two-component structure indicative of a cold, rotating nuclear disk embedded in a more slowly rotating elliptical host (e.g., Franx & Illingworth 1988; Bender 1990a; Bender, Saglia & Gerhard 1994; Scorza & Bender 1995). Because these disks are not self-gravitating, the processes discussed in this paper cannot operate. Therefore there must be some embedded nuclear disks in the earliest-type galaxies that are not related to the themes of this paper.

How are they produced? Minor accretion events in which an elliptical swallows a gas-rich dwarf almost certainly produce some of them, along with the central dust disks commonly observed in ellipticals (see Kormendy & Djorgovski 1989 for a review and Jaffe et al. 1994, van Dokkum & Franx 1995, and Martini et al. 2003 for HST images). There is evidence that dust disks gradually turn into small stellar disks (Kormendy et al. 1994, 2004). Alternatively, gas shed by dying stars in the elliptical may, in some cases, cool and fall to the center. These processes are quite different from secular evolution driven by nonaxisymmetries in dominant, self-gravitating disks.

We do not know whether the tiny nuclear disks seen, for example, in NGC 3115 (e.g., Lauer et al. 1995, Scorza & Bender 1995, Kormendy et al. 1996b) and NGC 4594 (e.g., Burkhead 1991, Kormendy 1988, Seifert & Scorza 1996, Kormendy et al. 1996a) are more nearly related to pseudobulges or to the disky distortions discussed above. The dividing line between the above processes and those that make pseudobulges deserves further investigation. This uncertainty affects only a minority of disky components embedded in the largest, earliest-type classical bulges. It is not a problem for the identification of most pseudobulges in Sb and later-type galaxies.