Lyα-emitting galaxies at z=3.1:: L* progenitors experiencing rapid star formation

We studied the clustering properties and multiwavelength spectral energy distributions of a complete sample of 162 Ly alpha- emitting (LAE) galaxies at z similar or equal to 3: 1 discovered in deep narrowband MUSYC imaging of the Extended Chandra Deep Field-South. LAEs were selected to have observed frame equivalent widths >80 angstrom and emission line fluxes >1.5 x 10(-17) ergs cm(-2) s(-1). Only 1% of our LAE sample appears to host AGNs. The LAEs exhibit a moderate spatial correlation length of r(0) = 3.6(-1.0)(+0.8) Mpc, corresponding to a bias factor b = 1.7(-0.4)(+0.3), which implies median dark matter halo masses of log(10)M(med) = 10.9(-0.9)(+0.5) M-circle dot. Comparing the number density of LAEs, 1.5 +/- 0.3 x 10(-3) Mpc(-3), with the number density of these halos finds a mean halo occupation similar to 1%-10%. The evolution of galaxy bias with redshift implies that most z 3: 1 LAEs evolve into present-day galaxies with L < 2.5L*, whereas other z > 3 galaxy populations typically evolve into more massive galaxies. Halo merger trees show that z 0 descendants occupy halos with a wide range of masses, with a median descendant mass close to that of L*. Only 30% of LAEs have sufficient stellar mass (>similar to 3 x 10(9) M-circle dot) to yield detections in deep Spitzer IRAC imaging. A two-population SED fit to the stacked UBVRIzJK+[3.6, 4.5, 5.6, 8.0] mu m fluxes of the IRAC-undetected objects finds that the typical LAE has low stellar mass (1.0(-0.4)(+0.6) 10(9) M-circle dot), moderate star formation rate (2 +/- 1 M-circle dot yr(-1)), a young component age of 20(-10)(+30) Myr, and little dust (A(V) < 0: 2). The bestfit model has 20% of the mass in the young stellar component, but models without evolved stars are also allowed.