Stochastic star formation activity of galaxies within the first billion years probed by JWST

¹ Aix Marseille Univ., CNRS, CNES, LAM, Marseille, France
² Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS, Laboratoire Lagrange, 06000 Nice, France
³ Institut d’Astrophysique de Paris, UMR 7095, CNRS, Sorbonne Université, 98 bis Boulevard Arago, F-75014 Paris, France
⁴ Cosmic Dawn Center (DAWN), Denmark
⁵ Niels Bohr Institute, University of Copenhagen, Jagtvej 128, 2200 Copenhagen, Denmark
⁶ Department of Astronomy, University of Geneva, Chemin Pegasi 51, 1290 Versoix, Switzerland

^★ Corresponding author: cristian.carvajal@lam.fr

Received: 17 July 2025
Accepted: 31 October 2025

Abstract

Early observations with the James Webb Space Telescope have highlighted the excess of UV-bright galaxies at z > 10, with a derived UV luminosity function (UVLF) that exhibits a softer evolution in redshift than expected. This unexpected trend may result from several proposed mechanisms, including a high star formation efficiency (SFE) or a bursty star formation history (SFH). In this work, we aim to characterize the burstiness level of high redshift galaxy SFHs and its evolution. We implemented a stochastic SFH module in CIGALE using power spectrum densities, to estimate the burstiness level of star formation in galaxies at 6 < z < 12. We find that SFHs with a high level of stochasticity better reproduce the Spectral Energy Distributions of z > 6 galaxies, while smoother assumptions introduce biases when applied to galaxies with bursty star formation activity. The assumed stochasticity level of the SFH also affects the constraints on galaxies’ physical properties, producing a strong and tight relation between the star formation rate (SFR) and stellar mass in the case of a smooth SFH, down to a weak relation at z ≥ 7 for an SFH with a high level of stochasticity. Successively assuming different levels of burstiness, we determined the best-suited SFH for each 6 < z < 12 galaxy in the JADES sample from a Bayes factor analysis. Galaxies are classified according to their level of burstiness, and the corresponding physical properties are associated with them. For massive galaxies (8.8 < log M_★/M_⊙ < 9.5), the fraction of bursty galaxies increases from 0.38 ± 0.08 to 0.77 ± 0.2 at z ∼ 6 and z ∼ 12, respectively. At all redshifts, only < 20% of low-mass galaxies are classified as bursty; although, this estimate is uncertain because their faintness leads to a low signal-to-noise ratio. For bursty galaxies, the log₁₀(SFR₁₀/SFR₁₀₀) ratio, another indicator of bursty star formation, does not evolve with redshift, but the fraction of galaxies with a high log₁₀(SFR₁₀/SFR₁₀₀) slightly increases from 0.28 ± 0.06 to 0.38 ± 0.11 between z ∼ 6 and z ∼ 9. We include additional constraints from observations on σ_UV, the dispersion of the UV magnitude distribution, and SFE, finding a maximum of 0.72 ± 0.02 mag and 0.06 ± 0.01 for σ_UV and SFE, respectively. This confirms that neither alone is responsible for the weak evolution of the UVLF at z > 10. Our results add further evidence that a combination with other mechanisms is likely responsible for the high-z UVLF. The stochastic SFH module is public as part of CIGALE version 2025.1.