When galaxies burst: enhanced shot-noise for line-intensity mapping in the JWST era

Abstract

Recent JWST observations indicate that star formation at $z\sim 4-6$ is more stochastic than previously assumed, with rms log-SFR scatter $\sim 0.6$ dex at $M_h\sim 10^{11}{M}_{\odot }$, growing toward smaller halos and time-correlated on $\sim 25$ Myr. This is significantly higher than the typical $\sim 0.3$ dex phenomenological lognormal scatter assumed in standard line-intensity mapping (LIM) forecasts. We propagate the JWST-era burstiness through to the LIM shot-noise power spectrum and show that the result is a simple multiplicative correction: the deterministic shot noise multiplied by a line-dependent boost factor $B_{\lambda }$ derived in closed form by convolving the SFR correlation function with the stellar-population-synthesis kernel of each line. At $z\sim 6$, we find $B_{\mathrm{H}\alpha }\simeq 7$ and $B\sim 2.5$-$3.5$ for longer-window tracers ([CII], CO, UV) - factors of $\sim 2$-$5$ above the standard prescription, and growing further toward higher redshift. The enhancement transforms the LIM landscape: it improves auto-spectrum detectability and suppresses lower-redshift interloper contamination, but degrades cosmological applications such as BAO that rely on a clean clustering measurement. Crucially, it also opens a new use of LIM as a diagnostic of high-redshift star-formation physics beyond the regime of individually resolved galaxies: redshift tomography of a single line constrains the amplitude and mass dependence of the burstiness, while cross-line shot-noise correlations probe its time coherence.

6 pages, 5 figures (with appendices: 11 pages, 7 figures)