The Early Maturity of High-Redshift Galaxies: Insights from sSFR, M/L and SFHs at z~7-14

TL;DR

This work analyzes a large, robust sample of $z\approx 7-14$ galaxies from the ASTRODEEP-JWST dataset using non-parametric SFHs to derive $sSFR$ and $M/L$, uncovering a largely non-evolving $sSFR$ and $M/L$ across $\sim$600 Myr and revealing significant diversity in stellar populations, including rare quiescent systems. The authors compare observations to multiple theoretical frameworks (AFM, GAEA variants, negative-$\Lambda$) and find that no single model reproduces the full variety, suggesting a combination of processes is at play and highlighting the potential existence of unseen dusty starbursts that could reconcile theory with data. They further infer the assembly histories of the highest-$z$ and evolved galaxies, showing rapid recent growth for some sources and early mass assembly for others, with progenitors often remaining UV-faint or highly dust-obscured. The results imply that galaxy formation in the early Universe is more complex and rapid than some models predict, underscoring the importance of multiwavelength follow-up (MIRI/ALMA) and deep spectroscopy to constrain dust, SFHs, and mass assembly routes. Overall, the paper provides crucial constraints on early galaxy assembly, the nature of SFHs, and the drivers behind the JWST-detected UV-bright excess at $z>10$.

Abstract

The James Webb Space Telescope (JWST) has revealed an unexpected excess of UV-bright galaxies at z>10, unaccounted for by extrapolations from pre-JWST observations and theoretical models. Understanding the physical properties and star formation histories (SFH) of high redshift systems is key to distinguish between the different proposed scenarios. We identify and analyse a sample of 2420 robust candidates at z~7-14 drawn from the ASTRODEEP-JWST dataset over ~0.2 deg^2 and model their properties with non-parametric SFHs to derive their specific star formation rate (sSFR) and stellar population properties. We find that the median sSFR and M/L remain roughly constant across the probed redshift range. We show that this result is robust against potential systematics unless a hidden population of dust-enshrouded starbursts, undetectable in current data, exists at these redshifts. In any case, the absence of observed high-sSFR systems at the highest redshifts suggests that any dust-free starburst phase must be short-lived. The observed sSFR evolution is in tension with most (though not all) theoretical models, making it a key quantity for discriminating among competing scenarios. The sample shows a wide range of physical conditions and galaxy classes, including systems with low sSFR and high mass-to-light ratios (M/L) up to z~10, indicative of already evolved galaxies only a few hundred Myr after the Big Bang, and different degrees of dust attenuation. We finally attempt to reconstruct the assembly histories of two sub-samples: a) the highest-M/L galaxies at z~7-8, which appear to have formed the bulk of their stars at least 500 Myr before observation, implying progenitors observable as UV-bright sources at z>20, and b) z>11 galaxies, which formed through stochastic SFH, remaining UV-faint for most of their early evolution, before undergoing recent (~50 Myr old) episodes of major growth.

Figures (11)

• Figure 1: Redshift distribution of the total sample (gray filled histogram) and of the galaxies selected in the different fields (coloured open histograms, see legend). The black dashed open histogram shows the distribution of spectroscopic sources.
• Figure 2: Stellar mass distribution for the full sample (filled gray) and in three redshift bins (thick solid lines). The small coloured arrows indicate the median of the distribution in each redshift bin. Thin dashed lines show the mass distribution of the Santa Cruz simulation, mass-matched to the observed galaxies (see Sect. \ref{['sec:massmatching']}), in the three redshift bins normalized to the peak of the observed mass distributions.
• Figure 3: Evolution of the sSFR. The median error bar in bins of redshift is shown at the bottom. The yellow line and shaded region represent the median and 16$^{th}$-84$^{th}$ percentile range of the $\log (M/M_\odot)$$=8-9$ subsample. Points are colour-coded according to the dust attenuation E(B$-$V). Being the E(B$-$V) calculated on a discrete grid, for visualization purposes we adopt the likelihood-weighted average. Pink open diamonds show a compilation of spectroscopic, JWST-based results taken from the recent literature tacchella23robertson23roberts-borsani23arrabal-haro23wang23zavala24deugenio24hsiao24hainline24bharikane24harikane25looser24carniani24carniani25topping25trussler25alvarez-marquez25helton25wu25naidu25witstok25kokorev25donnan25weibel25, with lines connecting different estimates for the same sources. Where available, we considered the SFR averaged over a timescale which is as close as possible to our choice of 20 Myr, sometimes averaging estimates over different timescales. The magenta horizontal dotted line represents the sSFR threshold for developing radiation driven outflows capable of expelling dust according to the model of ferrara24a (see text). The dark red dashed line and solid thin curves are the median and 10%, 50%, 80% and 99.9% probability densities for the SC-SAM.
• Figure 4: Relation between the stellar mass and the rest-frame UV absolute magnitude in three redshift bins. Thin black lines enclose 10%, 30%, 50%, 70%, and 90% probability densities. Large black symbols represent the median mass and 16$^{th}$-84$^{th}$ percentile range in bins of absolute magnitude. The black line shows the best-fit relation, and the dashed lines replicate the best-fit at $z\sim 7$ in the other redshift bins. Coloured lines and symbols illustrate previous results in the literature: duncan14, song16, bhatawdekar19, kikuchihara20, stefanon21, santini23, rojas-ruiz25. The dark red thin curves enclose 10%, 50%, 80% and 99.9% probability densities for the SC-SAM.
• Figure 5: Evolution of the $\rm M/L$, where L is the rest-frame UV luminosity. Points are colour-coded according to the dust attenuation E(B$-$V). The thick yellow line indicate the median of the $\log (M/M_\odot)$$=8-9$ subsample and the shaded region spans the 16$^{th}$-84$^{th}$ percentile range. The dark red dashed line and solid thin curves are the median and 10%, 50%, 80% and 99.9% probability densities for the SC-SAM.