Back to basics: Little Red Dots as galaxies and dust-obscured AGNs in a synthetic NIRCam sky simulated with L-GalaxiesBH

TL;DR

This work presents a JWST-motivated, wide-area mock sky built with the L-GalaxiesBH semi-analytic model to study Little Red Dots (LRDs) at $z\geq4$. By generating realistic galaxy+MBH spectral energy distributions, including dust attenuation and torus reprocessing, and applying photometric cuts analogous to observations, the authors quantify how LRDs arise from the balance of stellar continua and MBH emission. They find LRDs are predominantly massive galaxies with modest black holes, show V-shaped rest-frame SEDs driven by Balmer breaks and limited AGN contamination, and that heavy-seed MBH origins are not required to reproduce LRD demographics. The study demonstrates LRDs occupy a significant, evolving fraction of high-$z$ detections and provides a framework to interpret LRDs within the co-evolution of galaxies and MBHs, while acknowledging model limitations and the need for deeper, multi-wavelength tests.

Abstract

The enigmatic Little Red Dots (LRDs) discovered by the James Webb Space Telescope (JWST) exhibit properties challenging their interpretation as common galaxies or Active Galactic Nuclei (AGN). Understanding their nature is key to placing them within our picture of early galaxy and massive black hole (MBH) evolution. To this aim, we build a realistic comparison between LRD observations with photometric properties of galaxies and AGN simulated by the L-GalaxiesBH model in a NIRCam mock sky. We model stellar continua and emission lines, the MBH emission from accretion disk, infrared radiation from dusty torus, and lines from narrow and broad line regions, accounting for dust attenuation and obscuration. Using realistic photometric cuts, we select a population of LRDs including both AGN and galaxies. The LRD fraction peaks at 40% ($\sim10^{-4}\rm Mpc^{-3}$) at $z\sim4$. Our LRDs are central galaxies spanning $M_*=10^8-10^{10.5}\rm M_\odot$. A population of galaxies with $M_*<10^9\rm M_\odot$ appear as LRDs due to older stellar populations. At higher masses, LRDs dominate the halo and stellar mass functions ($M_{\rm vir} > 10^{11.5}\rm M_\odot$, $M_* > 10^{9.5}\rm M_\odot$), and the interplay between AGN and galaxy emission drives the LRD selection. AGN dominate rest-frame UV-optical emission, while dust obscuration is secondary. LRDs host lighter MBHs ($\sim 10^{6.5}\rm M_\odot$) than non-LRDs ($\sim 10^{7.5}\rm M_\odot$), with fainter emission unable to balance their hosts Balmer breaks. We find no evidence for dominant heavy-seed origin of MBHs. LRD Galaxies (97% hosting MBHs) and LRD AGNs are disk-dominated, with LRD AGNs showing larger bulges formed mainly via disk instabilities.

Figures (19)

• Figure 1: Distribution of galaxies in redshift ($z$) and right ascension (RA) for a thin slice in declination (DEC) of the JWST lightcone used in this work. Each point is color-coded by its F220W - F277W color to provide a qualitative idea of the color shift induced at $z\,{\sim}\,6$ by different spectral features affecting the F220W and F277W filters.
• Figure 2: Population of galaxies and MBHs in the simulated lightcone at $4\,{\leq}\, z\,{<}\,6.5$ (left column) and $6.5\,{\leq}\, z \,{\leq}\, 8.5$ (right column). Upper row: Comparison between the predicted stellar mass function (black line) and the observational constraints of Gonzalez2011Song2016 and Akins2025. Middle row: Predicted AGN bolometric luminosity function (black line) and observations from shen20_agnsathighredshiftgreene2024_lrdsAkins2025kokorev24_lrdsBarlowHall2025 and Greene2025. Lower row: Predicted black hole mass function. The results are compared with matthee2024_lrdskokorev24_lrds. Colored literature data relate to LRDs.
• Figure 3: Schematic illustration of the SED components in our galaxy and AGN modelling. On the bottom left we show the SEDs of two galaxies corresponding to different lines of sight ($\alpha_{\rm LOS}$): one directly towards the ISM and another with a more direct view of the disk plane. On the right, we show a zoom into the nuclear MBH configuration adopted. Different colours in the spectra correspond to different regions of the AGN structure: blue for the accretion disk, red for the torus, and green for the BLR & NLR. In black we show the total observed SED (i.e the MBH emission obscured and/or attenuated). Two $\alpha_{\rm LOS}$ are provided, corresponding to the regime of obscuration ($\varphi\leq\alpha_{\rm LOS}$) and another to non-obscuration ($\varphi>\alpha_{\rm LOS}$). On the bottom right, we show a zoom-in to the line profiles of ${\rm H{\alpha}}\,$ (6563 Å) and [NII] (6583Å) lines.
• Figure 4: Color-color (upper and middle panels) and compactness versus color (lower panel) diagrams used to define our photometrically selected LRDs. The blue contours correspond to all detected objects. The color map show the distribution of simulated objects from our lightcone (the color code represents the number of selected objects in each bin). The red dots represent the sample from kokorev24_lrds, with the vertical and horizontal red lines indicating the photometric cuts we apply to replicate their selection.
• Figure 5: Upper panel: Redshift distribution (red histograms) and number density (black points) of the detected LRDs. Lower panel: Fraction of LRDs over the total number of detected systems (black points), fraction of LRDs classified as AGN over the total number of detected objects classified as AGN (red points) and fraction of LRDs classified as AGN over the number of LRDs (green points). In all the plots, the error bars correspond to the Poisson error.