A Rapid Evolution in the Observed Mbh/M* Relation at z > 3 Revealed via Spectro-photometric SED-Modeling

Ansh R. Gupta, Anthony Taylor, Emma Curtis-Lake, Maddie Silcock, Óscar A. Chávez Ortiz, Steven L. Finkelstein, Hollis B. Akins, Bren E. Backhaus, Guillermo Barro, Laura Bisigello, Madisyn Brooks, Caitlin M. Casey, Stephane Charlot, Jacopo Chevallard, Anna Feltre, Giovanni Gandolfi, Mauro Giavalisco, Norman A. Grogin, Michaela Hirschmann, Tiger Yu-Yang Hsiao, Junehyoung Jeon, Shardha Jogee, Jeyhan S. Kartaltepe, Dale D. Kocevski, Anton M. Koekemoer, Vasily Kokorev, Gene C. K. Leung, Ray A. Lucas, Fabio Pacucci, Nor Pirzkal, Adele Plat, Rachel S. Somerville, Jonathan R. Trump, Alba Vidal-García, Xin Wang, L. Y. Aaron Yung

First listed 2026-06-01 | Last updated 2026-05-28

Abstract

Spectroscopic observations from JWST have uncovered a plethora of active galactic nuclei (AGN) at z > 4 with black hole (BH) mass (Mbh) to stellar mass (M*) ratios significantly above the local relation when using standard virial mass scaling relations. However, M* estimates of AGN may be inaccurate due to limitations in spectral energy distribution (SED) fitting codes, exemplified by a lack of physically-motivated AGN line emission models. Here, we fit NIRSpec/PRISM spectra of 39 galaxies at z ~ 3.5-7 selected as broad-line AGN from the CEERS and RUBIES surveys. Applying kinematic decompositions from NIRSpec/G395M spectra, we fit their continuum and narrow-component line fluxes using the BEAGLE-AGN SED fitting tool. While limitations of BEAGLE-AGN make it difficult to model little red dots (LRDs), we find that M* estimates of non-LRDs are, surprisingly, only modestly impacted by the inclusion or not of AGN narrow-line region (NLR) and continuum emission model components. We further find that non-LRD AGN at z < 3.5 are consistent with the local Mbh/M* relation while those at z > 4.5 display elevated ratios. While we cannot rule out observational biases or systematic uncertainties as partial causes, this transition over just ~500 Myr is driven entirely by changes in M* rather than an evolving Mbh distribution. These findings are consistent with models in which rapid BH growth results in elevated Mbh/M* ratios at early times, with a swift late-time assembly of host galaxies returning sources to the local relation at z < 4.

Short digest

This paper re-derives host-galaxy stellar masses for 39 CEERS and RUBIES broad-line AGN at z~3.5-7 by fitting NIRSpec/PRISM spectra and narrow-line fluxes with BEAGLE-AGN, using G395M-based kinematic decompositions to separate broad and narrow emission. For non-LRD AGN, adding AGN narrow-line region and continuum components changes M* only modestly, implying that their elevated Mbh/M* values are not mainly a simple SED-fitting artifact from omitted AGN emission. The central result is a rapid transition in the observed relation: non-LRD systems at z<~3.5 are consistent with the local Mbh/M* relation, while those at z>~4.5 remain elevated. Because the shift is driven by changing M* rather than an evolving Mbh distribution, the paper favors a picture where black holes grow early and hosts assemble stellar mass quickly later, while noting that residual biases and systematics cannot be fully ruled out.

Key figures to inspect

• Figure 1. Use this figure to show how the sample is defined and how the spectroscopy is handled. It is the most likely place to anchor the 39-source CEERS plus RUBIES broad-line AGN sample, the PRISM and G395M data combination, and the broad-versus-narrow kinematic decomposition that underpins every later stellar-mass inference.
• Figure 3. Recommend the figure that directly compares BEAGLE-AGN stellar-mass estimates with and without AGN narrow-line region and continuum components, ideally split by LRD versus non-LRD behavior. This is the paper's key methodological result, because it shows that non-LRD host masses are only modestly perturbed by the added AGN terms even though LRDs remain hard to model.
• Figure 4. This should be the main science figure if it is the Mbh versus M* or Mbh/M* comparison against the local relation across redshift bins. It is the cleanest single visual for the paper's headline claim that non-LRD AGN move from near-local ratios at lower redshift to elevated ratios at z>4.5.
• Figure 5. Include the later diagnostic figure that separates the evolution of Mbh from the evolution of M* or otherwise demonstrates that the transition happens through changing stellar masses rather than changing black-hole masses. That distinction is the paper's most important physical takeaway, because it reframes the result as rapid host assembly after an earlier phase of comparatively faster black-hole growth.