Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Prevalence of Bursty Star Formation in Low-Mass Galaxies at z=1-7 from Hα-to-UV Diagnostics

Marissa N. Perry, Anthony J. Taylor, Oscar A. Chavez Ortiz, Steven L. Finkelstein, Gene C. K. Leung, Micaela B. Bagley, Vital Fernandez, Pablo Arrabal Haro, Katherine Chworowsky, Nikko J. Cleri, Mark Dickinson, Richard S. Ellis, Jeyhan S. Kartaltepe, Anton M. Koekemoer, Fabio Pacucci, Casey Papovich, Nor Pirzkal, Sandro Tacchella

TL;DR

This study uses JWST/NIRSpec spectroscopy from the CEERS and RUBIES surveys to quantify burstiness in star formation histories across $0.8 \le z < 7$ and $7.0 \le \log(M_*/M_\odot) \le 10.9$ by comparing dust-corrected $L_{\rm H\alpha}$ and $L_{\rm UV}$ tracers. By modeling the star-formation main sequence with a non-parametric SFH and incorporating mass-completeness and selection biases, the authors find the intrinsic scatter of the H$\alpha$- and UV-based SFMS are nearly identical, suggesting no strong short-timescale variability detectable via the SFMS. The H$\alpha$-to-UV ratio reveals that $73^{+4}_{-4}$% of galaxies deviate from a constant SFH over the past $\sim100$ Myr, with low-mass systems ($7 \le \log(M_*/M_\odot) < 8.5$) being $30 \pm 1$% more likely to be in a recent burst than higher-mass galaxies. Across redshift, the overall bursty fraction shows little evolution, implying that bursty SFHs in low-mass galaxies persist over roughly $0.8$–$6$ Gyr after the Big Bang, and highlighting the importance of time-resolved diagnostics for understanding early galaxy evolution.

Abstract

We present an analysis of bursty star-formation histories (SFHs) of 346 star-forming galaxies at $1\lesssim z<7$, selected from JWST/NIRSpec G395M and PRISM spectroscopy provided by the CEERS and RUBIES surveys. We analyze the correlation of star-formation rate vs. stellar mass (the star-forming main sequence, SFMS) for our sample and find no significant difference between the intrinsic scatter in the H$α$-based SFMS and the UV-continuum-based SFMS. However, the diagnostic power of the SFMS is limited at high redshift and low stellar mass due to observational biases that exclude faint, quenched galaxies. To more directly probe star-formation variability, we examine the dust-corrected H$α$-to-UV ratio, which is assumed to trace deviations a from constant SFH over the past $\sim100$ Myr. In our sample, $73^{+4}_{-4}$% of galaxies exhibit H$α$-to-UV ratios inconsistent with a constant SFH. We do not observe any statistically significant evolution in the H$α$-to-UV ratio with redshift. Additionally, lower-mass galaxies ($7\leq\text{log}(M_*/M_{\odot})<8.5$) are $30 \pm 1$% more likely to lie above this equilibrium range -- indicative of a recent ($\lesssim 100$ Myr) burst of star formation -- compared to higher-mass systems ($8.5\leq\text{log}(M_*/M_{\odot})\leq10.9$). These results suggest that bursty SFHs are more common in low-mass galaxies at $z\sim 1$-$7$ and that this remains relatively stable across $\sim0.8$-$6$ Gyr after the Big Bang.

The Prevalence of Bursty Star Formation in Low-Mass Galaxies at z=1-7 from Hα-to-UV Diagnostics

TL;DR

This study uses JWST/NIRSpec spectroscopy from the CEERS and RUBIES surveys to quantify burstiness in star formation histories across and by comparing dust-corrected and tracers. By modeling the star-formation main sequence with a non-parametric SFH and incorporating mass-completeness and selection biases, the authors find the intrinsic scatter of the H- and UV-based SFMS are nearly identical, suggesting no strong short-timescale variability detectable via the SFMS. The H-to-UV ratio reveals that % of galaxies deviate from a constant SFH over the past Myr, with low-mass systems () being % more likely to be in a recent burst than higher-mass galaxies. Across redshift, the overall bursty fraction shows little evolution, implying that bursty SFHs in low-mass galaxies persist over roughly – Gyr after the Big Bang, and highlighting the importance of time-resolved diagnostics for understanding early galaxy evolution.

Abstract

We present an analysis of bursty star-formation histories (SFHs) of 346 star-forming galaxies at , selected from JWST/NIRSpec G395M and PRISM spectroscopy provided by the CEERS and RUBIES surveys. We analyze the correlation of star-formation rate vs. stellar mass (the star-forming main sequence, SFMS) for our sample and find no significant difference between the intrinsic scatter in the H-based SFMS and the UV-continuum-based SFMS. However, the diagnostic power of the SFMS is limited at high redshift and low stellar mass due to observational biases that exclude faint, quenched galaxies. To more directly probe star-formation variability, we examine the dust-corrected H-to-UV ratio, which is assumed to trace deviations a from constant SFH over the past Myr. In our sample, % of galaxies exhibit H-to-UV ratios inconsistent with a constant SFH. We do not observe any statistically significant evolution in the H-to-UV ratio with redshift. Additionally, lower-mass galaxies () are % more likely to lie above this equilibrium range -- indicative of a recent ( Myr) burst of star formation -- compared to higher-mass systems (). These results suggest that bursty SFHs are more common in low-mass galaxies at - and that this remains relatively stable across - Gyr after the Big Bang.

Paper Structure

This paper contains 17 sections, 6 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Data
  3. Spectroscopy
  4. Photometry
  5. Methods
  6. Emission-Line Fitting
  7. SED Fitting
  8. UV-Continuum Measurement
  9. Stellar-Mass Completeness
  10. Results
  11. The Star-Forming Main Sequence
  12. The H$\alpha$-to-UV Luminosity Ratio
  13. Discussion
  14. Fraction Around the H$\alpha$-to-UV Equilibrium
  15. The Duty Cycle of Star Formation
  16. ...and 2 more sections

Figures (5)

  • Figure 1: Spectroscopic redshift distribution of our sample split by survey (top) and by redshift bin (bottom). The full sample consists of 346 spectroscopically confirmed galaxies. We divide the sample into four redshift bins: $z=1$--$2.5$, $z=2.5$--$4$, $z=4$--$5.5$, and $z=5.5$--$7$, which we use throughout our analysis to examine redshift evolution in galaxy properties across cosmic time.
  • Figure 2: H$\alpha$- (left) and UV-continuum-based (right) SFRs vs. stellar masses for our galaxy sample binned by redshift. Vertical gray regions and gray data points indicate where our sample is not mass complete, as defined in Section \ref{['sec:stellar-mass-completeness']}. These data points are excluded from the fitting procedure. In the left panel, horizontal gray regions indicate the 5$\sigma$ NIRSpec G395M grating emission line sensitivity limits derived from Table 6 in finkelstein2025. In the right panel, we compute 3$\sigma$ limiting SFR(UV) values based on the faintest reliably measured rest-frame UV-continuum in each redshift bin, defined as the maximum of the 16th percentile $M_{1500~\mathring{A}}$ values from our SED-fitting posteriors of galaxies with $\text{SNR}>3$. The colored shaded regions denote the 1$\sigma_{\rm{}int}$ intrinsic scatter intervals about the SFMS and its associated error. The black dashed and dashed-dotted curves represent literature values from Popesso2023 and Speagle2014, respectively. On average, the H$\alpha$-based SFMS and the UV-based SFMS exhibit the same degree of intrinsic scatter, indicating that we do not detect short-timescale ($\sim10$ Myr) SFR variability across the sample with the SFMS.
  • Figure 3: The distribution of H$\alpha$-to-UV ratio values binned by redshift. The vertical gray hatched region shows the equilibrium range of the H$\alpha$-to-UV ratio calculated by Mehta2023. For each redshift bin, the distribution peak lies within the equilibrium range expected for a constant SFH over the past $\sim 100$ Myr.
  • Figure 4: Left panel: The H$\alpha$-to-UV ratio vs. stellar mass for our galaxy sample binned by stellar mass. The large points represent the median values in each stellar mass bin. The vertical error bars denote the 16th and 84th percentiles of each bin and the horizontal error bars denote the stellar mass range of each bin. The horizontal gray hatched region shows the equilibrium range of the H$\alpha$-to-UV ratio calculated by Mehta2023. Right panel: The H$\alpha$-to-UV ratio vs. redshift for our galaxy sample binned by stellar mass. The large points represent the median values in each stellar mass and redshift bin. The vertical error bars denote the 16th and 84th percentiles of each bin and the horizontal error bars denote the redshift range of each bin. Lower-mass galaxies maintain higher H$\alpha$-to-UV ratio values across all redshifts probed compared to higher-mass galaxies, suggesting lower-mass galaxies are more prone to bursty star formation.
  • Figure 5: The fraction of galaxies whose H$\alpha$-to-UV ratios are consistent (middle panel) and inconsistent (outer panels) with a constant SFH computed by Mehta2023 shown as a function of redshift. For each stellar mass and redshift bin, we perturb the observed ratios using Gaussian noise based on 1$\sigma$ errors and compute fractions via bootstrap resampling. Measured values correspond to the median of the bootstrap distribution, with error bars from the 16th and 84th percentiles. Colored curves represent this work, binned by stellar mass and redshift, compared to the dotted gray curve from Clarke2024, which includes galaxies with stellar masses in the range $10^7$--$10^{10.5}~M_*/M_{\odot}$. Within each bin, we compute the median redshift value. We estimate uncertainties on the fractions using bootstrap resampling. These trends suggest little evolution in the relative fraction of galaxies above, below, or within the equilibrium range over $z=1$--$7$.