Why Little Red Dots Disappear at z < 3: Evolution of Number Density and Halo Mass

Chenxuan Zhang, Huanian Zhang, Qingwen Wu, Luis C. Ho, Jian-Min Wang

First listed 2026-06-03 | Last updated 2026-06-01

Abstract

A significant puzzle in extragalactic astronomy is the scarcity of Little Red Dots (LRDs) at $z < 3$, compared to their higher abundance at earlier epochs. To understand this transition, we investigate the cosmic evolution of LRD environments. We measure the overdensity for LRDs and the general galaxy population at $3<z<7$, and find that at $z > 4$, LRDs predominantly reside in under-dense regions relative to the general galaxy population. By $z \sim 3.5$, however, this environmental contrast roughly diminishes, and LRDs are found in regions of comparable density to typical galaxies. Simultaneously, the dark matter halo masses of LRDs, inferred from large-scale clustering, grow rapidly from $\lesssim 10^{10.1} \, M_{\odot}$ at $z \sim 7.5$ to $\sim 10^{11.3} \, M_{\odot}$ at $z \sim 3.5$, where the halo mass becomes close to that of normal galaxies at lower redshift. Applying an empirical stellar-to-halo mass scaling relation, we derive stellar masses for LRDs; these show that black hole masses remain over-massive relative to stellar mass at $z > 4$, but converge toward the local $M_* - M_{\rm BH}$ scaling relation by $z \sim 3.5$. The coherent evolution of LRDs' large-scale environments $-$ as expressed by their overdensity and halo mass $-$ points to a distinct evolutionary pathway from that of normal galaxies. The significantly increased halo masses of LRDs lead to larger galaxy sizes, driven primarily by the potential enhancement of halo spins. Consequently, these sources are no longer as compact as typical high-redshift LRDs. Meanwhile, the depletion of dense gas and/or elevated star formation in their host galaxies would also significantly alter the spectral energy distribution of LRDs.

Short digest

Using 98 spectroscopically confirmed little red dots across six JWST deep fields, this paper asks whether the disappearance of LRDs below z < 3 is tied to how their environments evolve rather than to selection alone. The authors show that LRDs at z > 4 preferentially live in under-dense regions relative to the general galaxy population, but by z ~ 3.5 that contrast largely vanishes, while clustering implies their typical halo masses rise rapidly from less than about 10^10.1 solar masses at z ~ 7.5 to about 10^11.3 solar masses at z ~ 3.5. Mapping those halo masses to stellar masses, they argue that LRD black holes are strongly over-massive relative to their hosts at early times but move toward the local M* - MBH relation by z ~ 3.5. Their proposed explanation for why LRDs disappear is therefore evolutionary: as halos grow, hosts become less compact and their gas and star-formation conditions change enough to erase the compact red SED phenotype that defines high-redshift LRDs.

Key figures to inspect

• Figure 1. This is the core environmental result of the paper. It directly shows the projected overdensity profiles of LRDs versus the matched galaxy population across redshift, plus the redshift evolution of the integrated LRD-to-galaxy overdensity ratio, making the case that LRDs start in relatively under-dense regions at z > 4 and approach normal environments by z ~ 3.5.
• Figure 2. This figure carries the paper's main physical synthesis by translating the clustering measurements into halo mass evolution and placing LRDs alongside galaxies, low-luminosity AGN, quasars, and earlier LRD constraints. It is the cleanest visual summary of the claimed rapid halo growth from very low-mass hosts at high redshift to halos comparable to ordinary galaxies at lower redshift, which underpins the disappearance argument.
• Figure 3. This is the most important downstream interpretation figure because it connects the environment and halo results to black hole-galaxy coevolution. By showing where the LRD sample falls relative to local and high-redshift MBH scaling relations, it visualizes the claim that LRD black holes are initially over-massive for their hosts and then trend toward the local relation by z ~ 3.5.
• Figure 7. This is the key measurement figure behind the halo-mass inference, not just a derived comparison product. The projected LRD-galaxy cross-correlation functions in four redshift bins show the actual scale-dependent clustering signal and fitted correlation lengths that are later converted into the halo-mass evolution plotted in Figure 2.