As a reminder, globular clusters are tyically composed of 104 to 106 stars clustered within a few parsecs. They are old, although young globular-cluster-like objects are seen in mergers, and their metallicity can vary between [Fe/H] -2.5 dex and [Fe/H] > 0.5 dex. We refer to a globular cluster system as the totallity of globular clusters surrounding a galaxy.
2.1. Why study globular cluster systems?
The two fundamental questions in galaxy formation and evolution are: 1) When and how did the galaxies assemble? and 2) When and how did the galaxies form their stars? A third question could be whether, and to what extend, the two first points are linked.
Generally speaking, in order to answer these questions from an observational point of view, one can follow two paths. The first would be to observe the galaxies at high redshift, right at the epoch of their assembly and/or star formation. We will consider this to be the "hard way". These observations are extremely challenging for many reasons (shifted restframe wavelength, faint magnitudes, small angular scales, etc...). Nevertheless, they are pursued by a number of groups through the observations of absorption line systems along the line of sights of quasars, or the detection of high-redshift (e.g. Lyman break) galaxies, etc... (see e.g. Combes, Mamon, & Charmandaris 1999, Bunker & van Breugel 1999, Mazure, Le Fevre, & Le Brun 1999 for recent proceedings on the rapidly evolving subject).
The second path is to wait until a galaxy reaches very low redshift and try to extract information about its past. This would be the "lazy way". This is partly done by the study of the diffuse stellar populations at 0 redshift and the comparison of its properties at low redshift. Such studies on fundamental relations (fundamental plane, Mg-, Dn-) tend to be consistent with the stellar populations evolving purely passively and having formed at high redshifts (z > 2). Alternatively, one can study merging events among galaxies at low- to intermediate-redshift (e.g. van Dokkum et al. 1999) in order to understand the assembly of galaxies.
How do globular clusters fit into these picture? Globular cluster studies could be classified as the "very lazy way", since they reach out to at most redshifts of z = 0.03. However, globular clusters are among the oldest objects in the universe, i.e. they witnessed most, if not all, of the history of their host galaxy. The goal of the globular cluster system studies is therefore to extract the memory of the system. Photometry and spectroscopy are used to derive their ages and chemical abundances which are used to understand the epoch(s) of star formation in the galaxy. Kinematic information obtained from the globular clusters (especially at large galactocentric radii) can be used to help understanding the assembly mechanism of the galaxies.
2.2. Advantages of using globular clusters
Figure 1 illustrates extragalactic globular cluster studies. We show the galaxy NGC 1399 surrounded by a number of point sources. If these point sources could be resolved, they would look like one of the Galactic globular clusters (here shown as M 15). However, even with diffraction limited imaging from space, we cannot resolve clusters at distances of 10 to 100 Mpc into single stars, and have to study their integrated properties. The study of a globular cluster system is therefore equivalent to studying the integrated properties of a large number of globular clusters surrounding a galaxy, in order to derive their individual properties and compare them, as well as the properties of the system as a whole, with the properties of the host galaxy.
Figure 1. The figures shows a montage of what we are observing. The galaxy NGC 1399 of the Fornax galaxy cluster is shown on the left. It is located at a distance of ~ 18 Mpc from the Milky Way. Many of the point source surrounding it are globular clusters. If we could resolve these clusters into stars (which we cannot), we would see clusters similar to the Galactic clusters M 15, shown on the right.
The purely practical advantages of observing objects at z = 0 is the possibility to study the objects in great detail. Very low z observations are justified if the gain in details outbalances the fact that at high z one is seeing events closer to the time at which they actually happened. One example that demonstrates that the gain is significant in the case of extragalactic globular clusters, is the discovery of several sub-populations of clusters around a large number of early-type galaxies. The presence of two or more distinct star-formation epochs/mechanisms in at least a large number if not all giant galaxies was not discovered by any other type of observations.
The old age of globular clusters is often advanced as argument for their study, since they witnessed the entire past of the galaxy including the earliest epochs. If this would be the whole truth, globular clusters would not be suited to study the recent star formation epochs. Nor would they present a real advantage over stars, which can be old too. What are the advantages of observing globular clusters as tracers of the star-formation / stellar populations instead of studying directly the diffuse stellar population of the host galaxies?
Globular clusters trace star formation
A number of arguments support the fact that globular clusters indeed trace the star formation in galaxies. However, we know that some star formation can occur without forming globular clusters. One example is the Large Magellanic Cloud which, at some epochs, produced stars but no clusters (e.g. Geha et al. 1998). On the other hand, we know that major star formation episodes induce the formation of a large number of star clusters. For example, the violent star formation in interacting galaxies is accompanied by the formation of massive young star clusters (e.g. Schweizer 1997). Also, the final number of globular clusters in a galaxy is roughly proportional to the galaxy luminosity, i.e. number of stars (see Harris 1991). This hints at a close link between star and cluster formation. Additional support for such a link comes from the close relation between the number of young star clusters in spirals and their current star formation rate (Larsen & Richtler 1999).
In summary, globular clusters are not perfect tracers for star formation, as they will not form during every single little (i.e. low rate) star formation event. But they will trace the major (violent) epochs of star formation, which is our goal.
From a practical point of view
Globular clusters exist around all luminous (MV > 17) galaxies observed to date. Their number, that scales with the mass of the galaxy, typically lies around a few hundreds to a few thousands.
Furthermore, globular clusters can be observed out to ~ 100 Mpc. This is not as far as the diffuse light can be observed, but far enough to include many thousands of galaxies of all types and in all varieties of environments.
The study of globular clusters is therefore not restricted to a specific type or environment of galaxies: unbiased samples can be constructed. From this point of view, diffuse stellar light and globular clusters are equally appropriate.
The advantages of globular clusters over the diffuse stellar population
Globular clusters present a significant advantage when trying to determine the star formation history of a galaxy: they are far simpler structures. A globular cluster can be characterized by a single age and single metallicity, while the diffuse stellar population of a galaxy needs to be modeled by an unknown mix of ages and of metallicities. Studying a globular cluster system returns a large number of discrete age/metallicity data points. These can be grouped to determine the mean ages and chemical abundances of the main sub-populations present in the galaxy.
Along the same line, and as shown above, globular clusters form proportionally to the number of stars. That is, the number of globular clusters in a given population reflects the importance of the star formation episode at its origin. Counting globular cluster in different sub-populations indicates right away the relative importance of the different star formation events. In contrast, the different populations in the diffuse stellar light appear luminosity weighted: a small (in terms of mass) but recent star formation event can outshine a much more important but older event that has faded.
As for stellar populations, kinematical informations can be derived from the spectra originally aimed at determining ages and metallicities. Globular clusters have the advantage that they can be traced out to galactocentric distances unreachable with the diffuse stellar light. The dynamical information of the clusters can be used to study the assembly of the host galaxy. In Section 6, we will come back to this point.
The bottom line is that globular clusters are good tracers for the star formation history of their host galaxies, and eventually allow some insight into their assembly too. They present a number of advantages over the study of stellar populations, and complement observations at high redshift. Their study allows new insight into galaxy formation and evolution.