Fueling, Evolution, and Diversity of AGN in Dwarf Galaxies: Insights from Star Formation and Black Hole Scaling Relations

TL;DR

This work analyzes star formation and black hole–host scaling in a large sample of AGN hosted by dwarf galaxies ($ \log(M_\star/M_\odot)\lesssim 10$) at $0.5<z<0.9$ from VIPERS, using optical (Seyfert/LINER) and IR (WISE) selections. It introduces $SFR_{\mathrm{norm}}$ to compare AGN hosts with mass- and redshift-matched star-forming dwarfs, and examines its dependence on $L_{\mathrm{[OIII]}}$, $M_{\mathrm{BH}}$, environment, and stellar age, while exploring the $M_{\mathrm{BH}}{-}M_\star$ relation across AGN types and comparing with X-ray AGN and high-$z$ quasars. The results show suppressed star formation at low accretion power, type-dependent trends of $SFR_{\mathrm{norm}}$ with black hole mass, and elevated $M_{\mathrm{BH}}/M_\star$ ratios relative to local scaling relations that remain roughly constant with redshift; the data suggest diverse fueling modes and evolutionary pathways in dwarf hosts rather than a single evolutionary sequence. These findings imply rapid early black hole growth in low-mass galaxies and highlight the importance of multiwavelength selection to capture the full diversity of AGN activity in dwarfs, with implications for seed black hole formation and galaxy evolution in the low-mass regime.

Abstract

We investigate the star formation activity and black hole scaling relations in a sample of 1451 AGN hosted by dwarf galaxies at redshift 0.5 to 0.9, drawn from the VIPERS survey. The sample comprises Seyferts and LINERs identified through emission-line diagnostics, as well as IR-selected AGN based on WISE colors. Using the parameter SFRnorm, defined as the ratio of the SFR of a galaxy hosting an AGN to the median SFR of star-forming galaxies of similar stellar mass and redshift, we compare AGN hosts to a control sample of non-AGN star-forming galaxies. We examine how SFRnorm varies with AGN power ([O III] luminosity), black hole mass, local environment, and stellar population age. We also analyze the MBH-Mstar relation and the evolution of the MBH/Mstar ratio, incorporating comparisons to X-ray AGN and high-redshift quasars (z > 4). Our key findings are: (i) all AGN populations show suppressed star formation at low AGN luminosities, with SFRnorm rising above unity at different luminosity thresholds depending on AGN type; (ii) LINERs show flat SFRnorm trends with MBH, remaining broadly consistent with unity; Seyferts display a mild increase with MBH, while IR AGN show a more pronounced positive trend; (iii) LINERs exhibit older stellar populations than Seyferts; (iv) at fixed stellar mass, Seyferts host more massive black holes than LINERs, with IR AGN falling in between; (v) the MBH/Mstar ratio is elevated relative to local scaling relations and remains approximately constant with redshift, in agreement with high-z AGN; (vi) the ratio decreases with stellar mass up to log(Mstar/Msun) approximately 11, beyond which it flattens toward values consistent with those of local, inactive galaxies, with this trend clearest for Seyferts and IR AGN. These results suggest that AGN in dwarf galaxies follow diverse evolutionary pathways shaped by gas availability, feedback, and selection effects.

Figures (5)

• Figure 1: Distributions of stellar mass (left) and redshift (right) for the three AGN populations. The mean ($\mu$) and median (med) values for each distribution are indicated in the legends. Vertical lines correspond to mean values.
• Figure 2: Distributions of specific SFR ($\rm sSFR=\frac{SFR}{M\star}$) for the three AGN populations.
• Figure 3: Normalized star formation rate (SFR$_\mathrm{norm}$) for the three AGN populations as a function of key physical parameters. The top left panel shows SFR$_\mathrm{norm}$ versus [OIII] luminosity, the top right panel shows SFR$_\mathrm{norm}$ as a function of black hole mass, and the bottom left panel presents SFR$_\mathrm{norm}$ as a function of environmental density. The bottom right panel displays the distributions of stellar ages, using the D4000 index as a proxy.
• Figure 4: Black hole mass versus stellar mass for the three AGN populations. Small colored symbols represent individual sources, with distinct shapes and colors indicating the different AGN populations as shown in the legend. Large symbols indicate median values in stellar mass bins of width 0.25 dex. Error bars correspond to the standard error of the mean ($1\,\sigma$) in each stellar mass bin, computed as the standard deviation divided by the square root of the number of sources in the bin. Solid lines show least-squares fits to the binned data. Dashed lines correspond to established $M_\mathrm{BH}$–$M_\star$ relations in the local universe from the literature. Black and green markers denote X-ray selected AGN from the XMM-XXL and 4XMM surveys, respectively, restricted to the same redshift range as our sample ($\rm 0.5 < z < 0.9$). Corresponding solid lines show the best-fit relations for these X-ray AGN samples.
• Figure 5: Evolution of the black hole–to–stellar mass ratio. Top: $\log(M_\mathrm{BH} / M_\star)$ vs. redshift for the three AGN populations in this study, with high-redshift ($z > 4$) AGN from the literature shown for comparison. Bottom: $\log(M_\mathrm{BH} / M_\star)$ vs. stellar mass, including high-redshift dwarf AGN Sun2025b and X-ray AGN from the XXL and 4XMM surveys. Dashed lines show linear fits to the binned data; for XXL AGN, the fit is limited to $\log(M_\star / M_\odot) < 11$. Solid and dashed horizontal lines indicate the average ratio for XMM-XXL AGN at $z < 2$Mountrichas2023b and for local inactive galaxies, respectively.