Tracking extragalactic wildlife

Suppose you take a slice of the universe at a random moment in cosmic time, and look around you at all the different galaxies you see. Classifying them by how they look is like a Dr. Seuss story, there are big ones and small ones, red ones and blue ones, some alone and some in groups.  You see a beautiful array of galaxy 'ecosystems', from huge numbers of galaxies crowded together in galaxy clusters and superclusters, to galaxies lined up in extended filaments, to isolated galaxies living alone in large voids. But compare this view to a different slice at a different phase in the universe's history, say much earlier in time, and things look quite different. You'd see most of them are blue, small and living rather isolated lives. Take a slice later in time and they're redder, and in big groups or clusters.

In some sense, learning to track specific types of galaxies from one cosmic epoch to the next is one of the holy grails of extragalactic astronomy. If we could watch individual galaxies as they change, from 12 billion years ago to today, we could observe all the physics that goes into shaping them, changing them, and many of our questions would be answered. What turns galaxies from blue to red? How do they increase their size and change shape? What happens to them as they move under the influence of gravity from isolation into galaxy clusters?

How do you link galaxies with their progenitors and descendants?

This is not an easy problem to solve, and although astronomers have learned a lot about key processes and phases in a galaxy's life, a lot of questions remain unanswered. It would help a lot if astronomers could watch galaxies change, in the same way that biologists tag wild animals and come back periodically to check in on what they've been up to.

Unfortunately astronomers can't tag galaxies, but it is possible to infer tags based on some simple assumptions. One such tag is based on dark matter. Astronomers think they understand some things about the underlying distribution of dark matter in the universe, which, because it makes up almost 90% of the matter in the universe, it dominates the gravitational forces on galaxies. Those gravitational forces are what determine the distribution of galaxies in the universe, and thus, the pattern they make on the sky. The dark matter essentially decides the shape of the cosmic web (see image below). The most basic building block of the cosmic web is the dark matter halo. Dark matter halos are gravitationally bound sphereoids of mass, which formed soon after the big bang (before the formation of galaxies),  when little regions of the universe with higher density of dark matter collapsed under their own gravity. These regions of higher mass became the future sites of galaxy formation. So every galaxy lives in a dark matter halo and is gravitationally 'stuck' in its dark matter halo. Wherever the halo goes, whatever the halo does, however many other halos it merges with, the galaxy is along for the ride.

Distribution of dark matter halos through cosmic time.
Image Credit: Springel et al. (Virgo Consortium), Max-Planck Institute for
Astrophysics, (composite: Christina Williams)

 Because of gravity, these dark matter halos tend to form a characteristic pattern, based on a mass hierarchy. As you might intuitively expect, the most massive ones tend to attract the less massive ones to cluster around them. As the universe evolves, the pattern of this web of dark matter changes with time, due to gravity, as halos attract and get closer and closer together. Astronomers use data including the Cosmic Microwave Background to figure out how the dark matter should behave, and then use physics and computer simulations to see how it changes from the big bang until now. The picture above shows snapshots of the cosmic web for 4 different epochs in time, calculated using one such computer simulation called the Millennium Simulation. And its easy to see the structure of dark matter grow over cosmic time.

Although we can't see these dark matter halos directly, we know the galaxies are living inside them. This means, by looking at the pattern that galaxies make, and how they cluster together at any given moment in cosmic history, you get a snapshot glimpse of the backbone structure holding the universe together: the distribution of galaxies outlines the distribution of dark matter, including the mass hierarchy (see image below). Essentially this means we can look at the pattern that galaxies make in the sky at any given time, and use it to figure out the mass of their dark matter halos. The simulations tell us how the patterns made by different halos of different mass change with time.

 The distribution of galaxies in the low-redshift universe.
Image Credit: M. Blanton and the SDSS

So the dark matter halo mass of the galaxy becomes the wildlife tag that tells us where the galaxy is going and where its been. All we have to do is measure the pattern (or, clustering) of galaxies in the sky, at different epochs of cosmic time! And better yet, split the galaxies by different colors, sizes, or shapes, and their clustering can tell us something about how their evolutionary sequences were different.

There is a catch: The pattern of galaxies used to measure the clustering must be precisely measured in order to distinguish it from other patterns. Clustering is better determined the more area of the sky that is used to image the pattern. For example if you have a large, complex pattern that repeats on very large scales, and only map a small region of it, you can't tell if you've missed a bigger pattern. You have to make sure your image of the sky samples a large enough area to adequately classify it. By mapping only a small portion of the pattern, you are limited by a phenomenon known as cosmic variance: because of the patterns, any small view of the sky will differ from another, and its impossible to reconstruct the pattern without mapping large areas.

Efforts to map the clustering of nearby galaxies in the sky has enjoyed great successes, for example with the Sloan Digital Sky Survey (SDSS). Galaxies are brighter the closer to Earth they are, and so imaging them takes less time. The majority of the telescope time can then be used to make big maps that cover significant fractions of the night sky. SDSS for example mapped more than a quarter of the entire low-redshift universe. Astronomers are still a long way from doing this on similar scales at high redshifts. At high-redshifts, galaxies are very faint and it takes a long time to image them, so the telescope time has to be spent on making deep images in order to detect the galaxies, not wide area.  Efforts to study clustering of high redshift galaxies have been notoriously plagued by the small sizes of high-redshift surveys, with the largest areas (COSMOS and DEEP2 surveys for example) being on the order of 2 -3 square degrees. This covers about one ten-thousandth of the sky, painfully tiny in comparison to SDSS's coverage of the sky. (Although the high-redshift surveys are small on the sky, they do probe a larger volume of space at high-redshift than a similar sized area would at low-redshift, for a discussion of this in COSMOS see blog post here.) These surveys detected lots of galaxies out to redshift 1-1.5 (when the universe was about half its current age), but missed key epochs during their earlier phases of evolution because they were limited to optical imaging. On this front, CANDELS is poised to make some significant advances in studies of high-redshift galaxy clustering, because it is a near-infrared survey, and therefore will be probing galaxies at the highest redshifts we can observe. (see discussion in this previous post). Because CANDELS will detect hundreds of galaxies out to redshift 7 and even higher, we may now be able to start making some initial measurements of the clustering of galaxies at the earliest stages of galaxy evolution, and hopefully gain a more complete picture of their life cycles. Where do the first galaxies in the universe end up? What shape did spheroidal red galaxies have 8 billion years ago? For now, we anxiously await the completion of the full CANDELS dataset.