No Luminous Little Red Dots: A Sharp Cutoff in Their Luminosity Function

Yilun Ma, Jenny E. Greene, Marta Volonteri, Andy D. Goulding, David J. Setton, Marianna Annunziatella, Eiichi Egami, Xiaohui Fan, Vasily Kokorev, Ivo Labbe, Xiaojing Lin, Danilo Marchesini, Jorryt Matthee, Themiya Nanayakkara, Luke Robbins, Anna Sajina, Marcin Sawicki

First listed 2025-09-02 | Last updated 2025-09-02

Abstract

One of the most surprising results of early James Webb Space Telescope (JWST) observations is the discovery of an abundance of red, compact, broad-line objects dubbed "little red dots" (LRDs) at $z>4$. Their spatial density ($\sim10^{-4}$-$10^{-5}\,\mathrm{cMpc^{-3}}$) is 100 times more abundant than UV-selected quasars at those redshift if one extrapolates the quasar luminosity function (QLF) down to the LRD regime. However, whether LRDs dominate black hole accretion at quasar-like luminosities ($L_\mathrm{bol}\gtrsim 10^{45-46}\,\mathrm{erg\,s^{-1}}$) remains unanswered, as probing the bright end of the LRD luminosity function requires a much larger area than those able to be surveyed by JWST. In this work, we present our search for the brightest LRDs ($K<23.7$) at $4.5<z<4.9$ using wide-area multiwavelength imaging surveys from the near-UV to the infrared bands. With over 15 square degrees of sky coverage, we only identify one single LRD candidate at $z_\mathrm{phot}\approx4.6$, which translates into a spatial density of $n(M_{5100}<-23.5)\approx10^{-8}\,\mathrm{cMpc^{-3}}$ -- this is nearly 10 times less abundant than the UV-selected quasars at similar optical luminosity. When combined with the LRD sample identified by JWST at the same redshift range, we find a sharp cutoff in the optical luminosity function at $λL_{5100}\approx2.5\times10^{44}\,\mathrm{erg\,s^{-1}}$, while the QLF turnover occurs at $\gtrsim20$ times higher luminosity. We therefore confirm the exclusively low-luminosity nature of LRDs, ruling out that LRDs are the counter parts of quasars. Furthermore, we speculate that, if the shape of the luminosity function holds up, it points to LRDs being powered by low-mass black holes with a narrow range of Eddington-level accretion rates.

Short digest

Wide-area imaging over 15.3 deg^2 targets the bright end of little red dots at 4.5<z<4.9 with K<23.7 and finds only one candidate (ECOSMOS-LumLRD-z5, z_phot≈4.6), implying n(M5100<-23.5)≈10^-8 cMpc^-3—about ten times below UV-selected quasars at the same optical luminosity. Combining with JWST-identified LRDs, the optical LF shows a sharp cutoff at λL5100≈2.5×10^44 erg s^-1, whereas the quasar LF turns over at ≳20× higher luminosity. The result confirms LRDs are exclusively low-luminosity rather than quasar counterparts. If the LF shape holds, it favors low-mass black holes accreting within a narrow Eddington-ratio range.

Key figures to inspect

• Figure 1: Check that stacked, redshift-shifted LRD spectra matched to K=23.7 mag keep all selection-band photometry above survey depths—this tests volume completeness—and see where ECOSMOS-LumLRD-z5 falls inside the color-selection box versus the full parent catalog.
• Figure 2: Compare the photometric SED of ECOSMOS-LumLRD-z5 to the NIRSpec/PRISM spectrum of UNCOVER-45924 to verify the LRD-like V-shaped SED and Balmer-limit turnover; inspect the HST F814W cutout and the EAZY p(z) for robustness of the z_phot≈4.6 solution.
• Figure 3: Inspect the LRD optical luminosity function points and upper limit from this work alongside JWST LRDs; note the sharp cutoff near λL5100≈2.5×10^44 erg s^-1 and the contrast with extrapolated quasar LFs at the same epoch.
• Figure 4: Use the schematic to see how a low-mass BH population with high, narrowly distributed Eddington ratios can reproduce the observed LRD LF cutoff, in contrast to the broader quasar BH mass/Eddington distributions.