A new sample of Little Red Dots at $z<0.45$ in DESI DR1: Broad Balmer lines, low ionization spectrum and no variability

Kevin Park, Alberto Torralba, Jorryt Matthee, Sara Mascia, Zoltán Haiman, Rohan P. Naidu, Anna de Graaff

First listed 2026-05-14 | Last updated 2026-05-14

Abstract

JWST has unveiled an abundant population of compact broad-line emitters largely at $z\gtrsim4$, the Little Red Dots (LRDs), which might represent a previously unprobed supermassive black hole evolution channel predominant at high redshift. However, the LRDs have remained mostly elusive at lower redshift ($z\lesssim2$) where detailed studies are possible from ground-based observatories. We searched for low-redshift LRDs in the Dark Energy Spectroscopic Instrument (DESI) survey. Our search is primarily based on emission line properties, as opposed to earlier approaches that searched for compact sources with specific photometric spectral energy distributions. We report the discovery of eight LRDs at $z=0.2-0.45$, which show spectral features akin to the high-redshift LRDs in the rest-frame optical. The sources are characterized by broad Balmer lines, steep Balmer decrements, compact morphologies, Balmer absorption features and/or strong He I emission, but weak or absent He II, [Ne V] or other high excitation lines typical of Type I AGN. For 7 out of 8 sources, we retrieve dense-cadence light curves from time-domain surveys and for most sources we find weak to no intrinsic variability ($0.0-0.1$ mag) over $4-17$ years in the rest-frame. We also highlight the identification of a quasar with similar Balmer line profiles as LRDs, but shows differences in Balmer decrement, significant variability, and high-ionisation lines. Given the effective volume $4.9{\rm Gpc^3}$ covered by DESI DR1 at $z<0.45$, our sample corresponds to a number density of $1.6\times10^{-9}$Mpc$^{-3}$, indicating a number density $\sim$10,000 times lower than in the first billion years of cosmic time. We find a dearth of luminous and red LRDs at $z<1$ compared to higher-redshift, which could suggest lower gas feeding rates of LRD activity due to higher metallicities at later cosmic epochs.

Short digest

Park et al. use an emission-line-driven search in DESI DR1 to identify eight little red dots at z=0.2-0.45, building a new low-redshift sample with rest-optical spectra that closely resemble JWST-selected LRDs. The objects show broad Balmer lines, steep Balmer decrements, compact morphologies, Balmer absorption features and/or strong He I emission, but weak or absent He II, [Ne V], and other high-ionization lines typical of ordinary Type I AGN. For 7 of the 8 sources, dense time-domain light curves reveal only 0.0-0.1 mag intrinsic variability over 4-17 rest-frame years, while a comparison quasar with similar Balmer profiles separates itself through stronger variability and high-ionization emission. Their inferred number density of 1.6×10^-9 Mpc^-3, about 10,000 times below the first-billion-year LRD abundance, sharpens the case that luminous red LRD activity becomes both far rarer and less extreme at late cosmic times.

Key figures to inspect

• Figure 1. Use Figure 1 for the paper’s sample-definition logic. This is the figure that should establish how the DESI DR1 broad-line search and template-matching strategy isolate low-redshift LRD candidates without relying on the classic photometric V-shape selection, which is central to what this paper newly contributes.
• Figure 3. Use Figure 3 for the core spectroscopic evidence. This is the figure that most directly shows why these eight sources are being called LRDs in the rest-frame optical: broad Balmer emission, unusually steep Balmer decrements, Balmer absorption and/or strong He I, and the contrast with the weak high-ionization lines expected in standard Type I AGN.
• Figure 8. Use Figure 8 for the mid-paper physical synthesis. This is the figure most likely to pull together the low-ionization diagnostics, Balmer-line behavior, and compactness or SED context across the sample, making it the best single place to show that the classification rests on a package of properties rather than on broad-line width alone.
• Figure 12. Use Figure 12 for the variability result, which is one of the paper’s strongest bottom-line claims. The recommended figure should show the long-baseline light-curve behavior of the DESI LRDs and the comparison quasar, making clear that the LRDs are nearly non-variable over 4-17 rest-frame years while the quasar analog is not.
• Figure 17. Use Figure 17 for the final population-level takeaway. This is the figure that should capture the low-redshift rarity of the sample and the comparison to higher-redshift LRDs in number density and luminosity-color space, which is what makes the paper matter beyond the eight individual objects.