A study of one galaxy's dynamics backs up previous claims that surprisingly compact galaxies existed in the early Universe. But how such objects blew up in size to form present-day galaxies remains a puzzle.
Giant red elliptical galaxies are the oldest and most massive assemblies of stars in the nearby Universe. Large optical telescopes have tracked their evolution back through 11 billion years — about 80% of the Universe's lifetime — by observing them at large cosmological redshifts1,2,3. Observations seemed to indicate that nothing much had happened to them over this time, except that they grew rarer in the more distant past. It was thus a surprise when astronomers discovered recently that these galaxies have grown in size by a factor of five over this period while barely changing in mass. This is like suddenly discovering that Roman Londinium had the same population as Greater London does today. This extreme size and density evolution was not predicted by theories of galaxy formation, and remains difficult to explain.
One possibility is that measurements of the galaxies' masses that are based on their luminosities are flawed, not only because they might be missing starlight but also because they are not sensitive to the galaxies' invisible dark-matter component (which does not contribute significantly to mass in the luminous regions, at least for nearby ellipticals). On page 717 of this issue, van Dokkum and colleagues4 report the first 'dynamical' mass measurement — which is sensitive to both the visible and the dark-matter component — for an individual red compact galaxy, known as 1255–0, that is seen 10.7 billion years ago and is less than 1 kiloparsec (∼3,000 light years) in size. This galaxy is about four times more massive and five to six times smaller than the Milky Way spiral. Distant red compact objects are widely considered to be ancestors of today's ellipticals owing to their similar stellar populations and morphologies. However, nearby ellipticals of similar mass have sizes of 3–10 kpc; none is of similar size and mass to 1255–0 (ref. 5).
A dynamical mass measurement is definitive but requires resolution of the internal velocity structure of the galaxy. If high-redshift ellipticals really are small but massive, it follows from Newton's law that their stars' velocities should be very high. Because stars in ellipticals generally move on eccentric orbits, this is measured using the stellar-velocity dispersion, which quantifies the average spread of velocities and can be determined from the subtle Doppler broadening of absorption lines in the galaxies' spectra.
Measuring the velocity dispersion for 1255–0 was a tour de force4. Most of the light of such a high-redshift galaxy is redshifted into the near-infrared waveband (1–2 µm), where the airglow emission from the night sky contributes an enormous, noisy background. Van Dokkum et al.4 sustained a heroic 29-hour exposure time using the 8-metre Gemini South Telescope's Near-Infrared Spectrograph. This was just enough to detect the absorption lines in the galaxy's spectrum — a feat in itself for such a high-redshift galaxy — and to measure the velocity dispersion. The measurement was partly helped by the large value of the final result: 510 km s−1, the largest ever measured for any galaxy, but a value it had to have if it was as massive and as compact as previous measurements of its stellar mass had suggested6.
Van Dokkum and colleagues' result4 is especially surprising because an earlier velocity-dispersion measurement7, based on a composite spectrum of several distant ellipticals (an interesting technique, albeit subject to its own problems), had measured a much lower value, sparking debate about the reliability of the stellar-mass measurements and the role of dark matter. The new measurement is significant because it is for a single object and so implies unequivocally that 1255–0 does have a density much higher than any nearby galaxy.
So why are these compact, high-redshift galaxies such a theoretical conundrum? Elliptical galaxies have long been seen as the endpoint of galaxy formation: when a star-forming spiral or irregularly-shaped galaxy, full of young blue stars, has its star formation quenched by some astrophysical-feedback process, it quickly ages to become a 'red and dead' elliptical. As the overall star-formation rate of the Universe winds down with time, it seems natural to find an increasing number of these galaxy fossils full of old red stars. But such red galaxies are found to be more massive than any star-forming galaxy, so something extra is needed to provide the mass.
The consensus has been that elliptical galaxies have also assembled through mergers of smaller galaxies, a process naturally expected in current galaxy-formation theories8. The most massive ellipticals would be the result of major mergers of smaller ellipticals — with these progenitors having been of roughly equal mass. Elliptical galaxies are observed to follow tight scaling relationships between size, mass and velocity, which one might think would be seriously disturbed by mergers. However, computer simulations show that mergers simply 'move' the galaxies along the relationships without making them significantly less tight9.
But this picture breaks down when size evolution is taken into account: if you merge enough elliptical galaxies at high redshift to account for the size change, you also make many more high-mass galaxies than are observed in the nearby Universe. An alternative is that if mergers are predominantly minor — those in which a low-mass object merges into one of much larger mass — size growth can be achieved without a substantial increase in mass10. However, low-mass galaxies generally contain a lot of young stars, so this seems inconsistent with the observed old stellar populations of the high-redshift compact galaxies and their nearby descendants.
This 'lack of fit' with the standard picture of elliptical-galaxy formation has driven a search for ways other than mergers by which the size of these galaxies could have blown up. For example, feedback processes such as an energy injection from a supernova11 or quasar12 could achieve that by expelling gas from the galaxy slowly (or rapidly), making the galaxy's gravitational potential well shallower and moving stars into larger orbits. But these processes require a level of star or quasar activity that has not been observed. A more exotic explanation could involve the yet unknown nature of dark matter.
Any successful explanation of the size evolution must solve what I call the synchronization problem, which in my view is the most fundamental. The size–mass scaling relationship is tight in the nearby Universe, and possibly also at high redshift. It is just the normalization of this relationship that evolves. There are no massive compact elliptical galaxies today. Therefore, the high-redshift (early-epoch) compact galaxies must be growing in size with time (Fig. 1). But, at the same time, the Universe is making new elliptical galaxies, and somehow both the growing and the newly formed galaxies fall within the same tight size–mass relationship at all epochs. Their evolution is 'synchronized' through some process that is either a coincidence or an important new piece of astrophysics.
A lot hinges on the interpretation of van Dokkum and colleagues' single velocity-dispersion measurement4 of 1255–0. As large telescopes acquire new multi-object, near-infrared spectrographs, we can expect to see many hundreds of such velocity-dispersion measurements in the next few years. We can also expect to see improved measurements of the structural and environmental properties of these compact galaxies, which will help us to figure out how bad the problems we have in explaining these objects really are. It remains to be seen whether we need conventional or novel explanations for their astounding growth into the most massive elliptical galaxies we see today.
Abraham, R. G. et al. Astrophys. J. 669, 184–201 (2007).
Cimatti, A. et al. Nature 430, 184–187 (2004).
van Dokkum, P. G. et al. Astrophys. J. 638, L59–L62 (2006).
van Dokkum, P. G., Kriek, M. & Franx, M. Nature 460, 717–719 (2009).
Trujillo, I. et al. Astrophys. J. 692, L118–L122 (2009).
Kriek, M. et al. Astrophys. J. 700, 221–231 (2009).
Cenarro, A. J. & Trujillo, I. Astrophys. J. 696, L43–L47 (2009).
De Lucia, G., Springel, V., White, S. D. M., Croton, D. & Kauffmann, G. Mon. Not. R. Astron. Soc. 366, 499–509 (2006).
Boylan-Kolchin, M., Ma, C.-P. & Quataert, E. Mon. Not. R. Astron. Soc. 369, 1081–1089 (2006).
Bezanson, R. et al. Astrophys. J. 697, 1290–1298 (2009).
Damjanov, I. et al. Astrophys. J. 695, 101–115 (2009).
Fan, L., Lapi, A., De Zotti, G. & Danese, L. Astrophys. J. 689, L101–L104 (2008).