Star formation histories and gas content limits of three ultra-faint dwarfs on the periphery of M31

TL;DR

The paper analyzes Pegasus V, Pisces VII, and Pegasus W—three ultra-faint dwarfs near M31—to measure distances, constrain gas content, and recover star formation histories from HST CMDs complemented by VLA HI limits. A Python-based StarFISH-like SFH fitting framework uses three isochrone libraries (PARSEC, MIST, BaSTI) to build Hess diagrams and perform maximum-likelihood fits, with uncertainties assessed via Monte Carlo realizations. The results show Pegasus W retaining star formation until ~1 Gyr ago, while Pegasus V and Pisces VII appear quenched by reionization, with all three possessing stringent HI upper limits (M$_{HI} < 10^4$ M$_\odot$). These findings support a scenario where low-mass UFDs around M31 quenched early, akin to MW satellites, though environmental effects may play a role for the more massive Pegasus W; deeper follow-up (e.g., JWST/HST) is needed to fully resolve the ancient populations and the nature of Pegasus W's recent features. This work provides crucial constraints on the interplay between reionization and environment in shaping the faint end of the dwarf galaxy population around different hosts.

Abstract

We present Hubble Space Telescope (HST) imaging of Pegasus V and Pisces VII, along with a re-analysis of the archival imaging of Pegasus W, and Jansky Very Large Array (VLA) neutral gas (HI) observations of all three. These three ultra-faint dwarfs (UFDs) are all within the Local Group in the approximate direction of M31. The VLA observations place stringent upper limits on their HI content, with all having $M_\mathrm{HI} < 10^4\;\mathrm{M_\odot}$. As the red giant branches of these UFDs are sparsely populated, we determined distances from the HST photometry of horizontal branch (HB) stars in comparison to a fiducial HB population (from M92), with all three falling in the range 0.7-1 Mpc. Using a new Python-based star formation history (SFH) fitting code (based on StarFISH), we derive SFHs of all three UFDs. As found previously, the best fit SFH for Pegasus W includes significant star formation well beyond the end of reionization, while the SFHs calculated for Pegasus V and Pisces VII are consistent with them having quenched shortly after reionization. These findings for the latter two objects indicate that, like those in the vicinity of the Milky Way, lower mass UFDs in the vicinity of M31 likely quenched at early times.

Figures (10)

• Figure 1: HST ACS F606W+F814W images of Pegasus W (top), Pegasus V (bottom-left) and Pisces VII (bottom-right). Point sources that meet our RGB+HB selection criteria (§\ref{['sec:HSTdata']}) are marked with yellow circles. The large dashed green ellipses indicate double the size of the half-light ellipse of each object. The large white X in the bottom-right panel indicates the center of Pisces VII given in Collins+2024. The background galaxy near the revised center likely adversely impacted the original ground-based photometry and led to an offset center being measured.
• Figure 2: VLA H i spectra of Pegasus W (top), Pegasus V (middle), and Pisces VII (bottom) extracted over a single synthesized beam centered on the optical location of each object. The dotted horizontal line in each panel indicates three times the rms noise in the spectrum. None of the spectra contain significant signs of genuine line emission, aside from that of the MW around $v\approx0$ km$\,$s$^{-1}$. The low-significance feature in the Pegasus V spectrum around 275 km$\,$s$^{-1}$ is the result of large scale correlated noise features in the data.
• Figure 3: HST CMDs (extinction-corrected) of Pegasus W (left), Pegasus V (center), and Pisces VII (right) within $2r_h$. The blue and red lines show the fiducial HB and RGB Mutlu-Pakdil+2019, respectively, while the green line shows a young MS population from PARSEC ($\log t/\mathrm{yr} = 8.3$, $[\mathrm{M/H}] = -1.7$), all shifted to the appropriate distance moduli (Table \ref{['tab:props']}). Almost no stars are present at the TRGB and thus we instead determined distances from the location of the HB (§\ref{['sec:struct_params']}). The stars used to fit the distance moduli are those within the HB selection boxes (grey rectangles). The blue shaded bands indicate the 1-$\sigma$ uncertainties in the distance moduli. The dotted and dashed black lines indicate the 90% and 50% completeness limits, respectively, for each observation. The errorbars along the right edge of the right panel indicate the typical photometric uncertainties as a function of F606W magnitude, which are comparable for all three observations.
• Figure 4: Cumulative fractional star formation histories of Pegasus W (left), Pegasus V (center), and Pisces VII (right). The fits from the three different isochrone libraries are shown with different colors and line styles. The diagonal dashed black lines represent a constant SFR for the age of the universe and the grey vertical band indicates the approximate period of cosmic reionization ($10 \lesssim z \lesssim 6$). The shaded color bands indicate the 1-$\sigma$ uncertainties of each model based on re-fitting MC realizations of the best fit artificial CMD. Note that the first time bin extends from the beginning of the universe until 10 Gyr ago. As this is a single bin, the SFR within it is plotted at the average (constant) value. In reality, the majority of the stars formed in this bin likely formed by the end of reionization (grey band). However, the current photometry is not deep enough to distinguish these ancient populations and they are thus grouped into a single bin.
• Figure 5: Differential star formation histories of the three UFDs, plotted in a style analogous to Figure \ref{['fig:cumSFHs']}. Only Pegasus W exhibits significant SF past 10 Gyr ago. We have truncated the SFHs at 100 Myr as SFRs more recent than this are unreliable given the sparsely populated CMDs of these UFDs. Note that the error bars on differential SFHs can be inflated by anti-correlations between adjacent time bins, causing them to overestimate the true statistical uncertainties. Such correlations do not significantly impact error estimates for the cumulative SFHs, thus the error bands in Figure \ref{['fig:cumSFHs']} provide a more robust representation of the uncertainties.