A number of theoretical studies indicate that bars redistribute the disc gas content, driving it through torques and dynamical resonances into spiral arms, rings and dust lanes, where it usually changes phase from neutral to molecular as it is submitted to higher pressures. Gas lying beyond the bar ends is driven outwards, whereas gas lying within the bar ends is driven to the central regions (e.g. [81, 18, 2, 33, 75]; see also [82, 51]). This rearrangement of the gas can erase chemical abundance radial gradients [35]. Accordingly, barred galaxies have flatter O/H gradients than unbarred galaxies [97], and the stronger the bar the flatter the gradient [61]. Furthermore, it has also been observed that the central concentration of molecular gas in barred galaxies is higher than in unbarred galaxies [80].
What happens to the gas funnelled by the bar to the centre? Several observational results point out that star formation activity is enhanced in the central regions of barred galaxies as compared to unbarred galaxies, for spirals of early and intermediate types [83, 84, 47, 15, 46, 1, 49], arguing that the gas driven to the centre by the bar is relatively efficiently transformed in young stars. Consequently, barred galaxies also show flatter colour gradients than unbarred galaxies, due to bluer central regions [41]. The average global star formation, however, seems to be comparable in barred and unbarred galaxies of intermediate types, since their integrated colours are very similar. In addition, bars in late type spirals show star forming regions all along the bar, whereas bars in early type spirals have star formation concentrated at the centre and/or at the bar ends [74]. There are suggestions that the former bars are also dynamically younger (i.e. formed later) than the latter ([34, 60]; see also [95, 73, 94]).
Theoretical work indicates that not only the disc gas content is rearranged by the effects induced by a bar, but that the distribution of stars in the disc also changes (see e.g. [71, 70, 79]). Furthermore, due to an exchange of angular momentum between the disc and the dark matter halo, mediated by the bar, the disc becomes more centrally concentrated [7, 3]. The inflow of disc material (gas and stars) to the centre, and the star formation episodes associated with it, seem to often produce sub-structures such as nuclear rings, nuclear discs and spiral arms, and secondary bars, as shown by observations (see e.g. [31]). The increase in central mass concentration can also make stars migrate to orbits out of the disc plane [10]. These structures built at the centre of the disc by the bar might naturally generate an excess of luminosity in the central part of the disc luminosity profile, as compared to an extrapolation to the centre of its outer part. Likewise, classical bulges can also be defined as the extra light on top of the inner disc profile, and also extend out of the disc plane. Thus, to distinguish these different components, formed through distinct processes and having dissimilar physical characteristics, structures built through the inflow of disc material are called pseudo-bulges [51], or disc-like bulges [4]. There are also observational evidences that bars might affect the distribution of stars in the outer disc [30].
Our understanding of yet another type of bulge also benefits largely from a successful connection between theory and observation. Theoretical studies show that the vertical structure of a bar grows in time through dynamical processes, generating a central structure out of the plane of the disc that can have a boxy or peanut-shaped morphology (e.g. [19, 76, 62]). Several observational evidences, mostly based on theoretical expectations, argue that such box/peanut bulges are indeed just the inner part of bars seen edge-on (see [22, 52, 14, 16, 13]).
The gas brought to the centre by the bar can also end up fuelling a super massive black hole and AGN activity. The basic idea is that the primary bar brings gas from scales of ~ 5 kpc to ~ 1 kpc and a secondary bar instability brings gas further inwards to ~ 100 pc. The gas still needs to loose further angular momentum, and at that scales other physical processes, such as viscosity, come to help [89, 88]. However, studies comparing the fraction of barred galaxies in quiescent and active galaxies show contradictory results [46, 68, 50, 53, 57]. Such comparisons are plagued by issues such as AGN classification, sample selection and bar definition, which in turn depends on wavelength and spatial resolution. Should one expect to see a clear distinction in bar fraction in quiescent and active galaxies? Even though there are more clear evidences of bars at least building up a fuel reservoir at the galaxy centre for star formation and AGN activity ([80, 58, 20]; see also [96]), the answer to this question is, for several reasons, more likely, no. To begin with, we saw that bars are very often seen in disc galaxies, and thus any difference is likely to have low statistical significance, unless samples are large enough. More important are factors such as the availability of gas (quiescent barred galaxies might just lack gas to fuel the black hole) and the strength of the bar (or its ability to bring gas inwards). The issue gets more complicated when one considers in detail the role of inner spiral arms and rings, secondary bars and dynamical resonances near the centre. Inner rings might prevent gas to reach the centre, and inner spiral arms could either remove or give further angular momentum from (to) the gas if they are trailing (leading) [17]. Hydrodynamical simulations show that, at least in some cases, secondary bars might not help (and might even prevent) the gas inflow to the nucleus ([56]; but see [45]). Finally, one must keep in mind that the typical time-scale of AGN activity episodes is likely to be much shorter than the typical time-scale for funnelling the gas and the life time of bars. This means that even if a bar is bringing gas to the nucleus one has to be looking at the right time to see the black hole accreting the gas. Thus, one should not be too surprised (or rather confused) to see results like the similar fractions of secondary bars found in quiescent and active galaxies [63], or the fact that galaxies with the strongest bars are mostly quiescent [90, 55]. Things are just not that simple! The nuclear region of barred galaxies shows complex structures that follow several distinct patterns, which might be related to the strength of the bar and the final destiny of the gas collected by it [72].
From theoretical work, it is still unclear whether primary and secondary bars are long-lived or have short lives. Earlier studies suggested that primary bars could be destroyed by a central mass concentration, such as a super massive black hole. However, it turned out that the central densities needed are unrealistically high, and that at most a weakening of the bar can result (see [85, 5] and references therein). This suggests that primary bars are long-lived. Models that include accretion of gas from the halo to the disc, however, suggest that primary bars can be destroyed due to a transfer of angular momentum from the gas to the bar [11, 12], but this is contested (see [10]). In addition, a new bar could be formed after the previous one vanished, since the accreted gas could turn the disc unstable again, by replenishing the disc with new stars. A key point in models that predict the demise and rebirth of bars is the availability of external gas. They are thus just relevant to gas-rich galaxies. Still on theoretical ground, also the life of secondary bars remains uncertain, with results suggesting both long and short life times (see [64, 28, 21]).
A natural and direct way to assess the life time of bars is to study bars at different redshifts, but this is as yet also not free from opposing results. Early studies had pointed out a sharp drop in the fraction of barred galaxies at z ~ 1 (see [92]). Not long afterwards, some studies showed that the apparent lack of bars was only caused by band shifting and poor spatial resolution [93, 86, 48, 26]. The most recent studies, however, do not seem to agree on whether there is a rapid decline in the bar fraction with redshift to z ~ 1 [87], or if this fraction is fairly constant [9]. A constant fraction of barred galaxies from z ~ 1 to z ~ 0 might put models that predict bar dissolution and reformation in trouble, as this would require a fine tuning of the corresponding time-scales. There are recent indications that bars in early and intermediate type spirals are long-lived [27]. If this is correct, it indicates that, in fact, models of bar reformation might concern only late type spirals (later than Sc). This agrees with the findings in [39, 40], where a methodology to estimate the dynamical ages of bars is introduced. This will be discussed in more detail in the next section, including implications for secular evolution scenarios and AGN fuelling by bars.