As discussed above, we now know that superwinds are ubiquitous in actively-star-forming galaxies in both the local universe, and at high-redshift. The outflows detected in the high-z Lyman Break galaxies are particularly significant, since these objects may plausibly represent the production sites of much of the stars and metals in today's universe (Steidel el al. 1999). Even if the sub-mm SCUBA sources turn out to be a distinct population at high-z, their apparent similarity to local "ultraluminous galaxies" suggests that they too will drive powerful outflows (Heckman et al. 1996, 2000). With this in mind, let me briefly describe the implications of superwinds for the evolution of galaxies and the inter-galactic medium.
Martin (1999) and Heckman et al. (2000) showed that the estimated outflow speeds of the neutral, warm, and hot phases in superwinds are 400 to 800 km s-1, and are independent of the rotation speed of the "host galaxy" over the range vrot = 30 to 300 km s-1. This strongly suggests that the outflows selectively escape the potential wells of the less massive galaxies. This would provide a natural explanation for the strong mass-metallicity relation in present-day galaxies (e.g. Lynden-Bell 1992; Tremonti et al. 2001).
As summarized above, the mass-outflow rate in entrained interstellar matter in a superwind is similar to the star-formation rate in the starburst. The selective loss of gas-phase baryons from low-mass galaxies via supernova-driven winds is an important ingredient in semi-analytic models of galaxy formation (e.g. Somerville & Primack 1999). It is usually invoked to enable the models to reproduce the observed faint-end slope of the galaxy luminosity function by selectively suppressing star-formation in low-mass dark-matter halos.
A different approach is taken by Scannapieco, Ferrara, & Broadhurst (2000), who have argued that starburst-driven outflows can suppress the formation of dwarf galaxies by ram-pressure-stripping the gaseous baryons from out of the dark-matter halos of low-mass companion galaxies. The NGC 3073 / 3079 interaction (Irwin et al. 1987) may represent a local example.
A direct consequence of a galactic-wind origin for the mass-metallicity relation in galactic spheroids is that a substantial fraction of the metals today should reside in the inter-galactic medium. This has been confirmed by X-ray spectroscopy of the intra-cluster medium (e.g. Finoguenov, Arnaud, & David 2001). The mean metallicity of the present-day inter-galactic medium is not known, but the presence of warm/hot metal-enriched intergalactic gas is demonstrated by the abundant population of OVI absorption-line clouds (Tripp, Savage, & Jenkins 2000). If the ratio of ejected metals to stellar spheroid mass is the same globally as in clusters of galaxies, then the present-day mass-weighted metallicity of a general intergalactic medium will be of-order 10-1 solar (e.g. Renzini 1997; Heckman et al. 2000). Early galactic winds have been invoked to account for the wide-spread presence of metals in the Ly forest at high-redshift (e.g. Madau, Ferrara, & Rees 2001).
There is now a vigorous debate as to whether and by what means the inter-galactic medium might have been heated by non-gravitational sources at relatively early epochs (e.g. Ponman, Cannon, & Navarro 1999; Pen 1999; Tozzi & Norman 2001; Voit & Bryan 2001; Croft et al. 2001). As a benchmark, consider the maximum amount of energy per inter-galactic baryon that can be supplied by galactic winds. Star-formation with the local initial mass function (Kroupa 2001) produces about 1051 ergs of kinetic energy from supernovae per 30 M of low-mass stars ( 1 M). The present ratio of baryons in the intra-cluster medium to baryons in low-mass stars is 6 in clusters, so the amount of kinetic energy available in principle to heat the intra-cluster medium is then 1051 ergs per 180 M, or 3 keV per baryon. A similar value would apply globally. While this upper bound is based on an assumption of unit efficiency for the delivery of supernova energy, I have emphasized above that the observed properties of superwinds demand high efficiency.
The physical state of much of the inter-galactic medium is regulated by the meta-galactic ionizing background. QSOs alone appear inadequate to produce the inferred background at the highest redshifts (e.g, Madau, Haardt, & Rees 1999). In principle, star-forming galaxies could make a significant contribution to the background, provided that a significant fraction of the ionizing radiation can escape the galaxy ISM. Steidel, Pettini, & Adelberger (2001) have have reported the detection of substantial amounts of escaping ionizing radiation in Lyman Break galaxies and have speculated that galactic superwinds clear out channels through which this radiation can escape. We (Heckman et al. 2001b) have considered the extant relevant data on present-day starbursts, and have concluded that galactic winds may be necessary but not sufficient for creating a globally porous interstellar medium.
Heckman et al. (2000) have summarized the evidence that starbursts are ejecting significant quantities of dust. If this dust can survive a trip into the intergalactic medium and remain intact for a Hubble time, they estimated that the upper bound on the global amount of intergalactic dust is 10-4. While this is clearly an upper limit, it is a cosmologically interesting one (Aguirre 1999). Dust this abundant is probably ruled out by the recent results by Riess et al. (2001), but intergalactic dust could well complicate the interpretation of the Type Ia supernova Hubble diagram.
Acknowledgments. I would like to thank my principal collaborators on the work described in this contribution: L. Armus, D. Calzetti, M. Dahlem, R. Gonzalez-Delgado, M. Lehnert, C. Leitherer, A. Marlowe, C. Martin, G. Meurer, C. Norman, K. Sembach, D. Strickland, and K. Weaver. This work has been supported in part by grants from the NASA LTSA program and the HST, ROSAT, ASCA, and Chandra GO programs.